INFORMATION TO USERS This reproductionwas made from a copy of a manuscript sent to us forpublication and microfilming. While the most advanced technology has been used to pho- tograph and reproduce this manuscript, the qualityof the reproduction is heavily dependent upon the quality of the material submitted. Pages in any manuscript mayhave indistinctprint. In all cases the best available copy has been filmed. The followingexplanationoftechniquesis provided to help clarifynotationswhich may appear on this reproduction. 1. Manuscripts may not always be complete. When it is not possible to obtain missing pages, a note appears to indicate this. 2. When copyrighted materials are removed from the manuscript, a note ap- pears to indicate this. 3. Oversize materials (maps, drawings, and charts) are photographedbysec- tioning the original, beginningat the upper left hand comer and continu- ing from left to right in equal sections with small overlaps. Each oversize page is also filmed as one exposure and is available, for an additional charge, as a standard 35mm slide or in black and white paper format. • 4. Most photographs reproduce acceptably on positive microfilm or micro- fiche but lack clarity on xerographic copies made from the microfilm. For an additional charge, all photographs are available in black and white standard 35mm slide format. * *For more information about black and white slides or enlarged paper reproductions, please contact the Dissertations Customer SelVices Department.
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
INFORMATION TOUSERS
This reproduction was made from a copyofa manuscriptsent to us for publicationand microfilming. While the most advanced technology has been used to photograph and reproduce this manuscript, the quality of the reproduction is heavilydependent upon the quality of the material submitted. Pages in any manuscriptmay have indistinct print. In all cases the best available copy has been filmed.
The followingexplanation of techniques is provided to help clarify notations whichmay appear on this reproduction.
1. Manuscripts may not always be complete. When it is not possible to obtainmissing pages, a note appears to indicate this.
2. When copyrighted materials are removed from the manuscript, a note appears to indicate this.
3. Oversize materials (maps, drawings, and charts) are photographed by sectioning the original, beginning at the upper left hand comer and continuing from left to right in equal sections with small overlaps. Each oversizepage is also filmed as one exposure and is available, for an additionalcharge, as a standard 35mm slide or in black and white paper format. •
4. Most photographs reproduce acceptably on positive microfilm or microfiche but lack clarity on xerographic copies made from the microfilm. Foran additional charge, all photographs are available in black and whitestandard 35mm slide format. *
*For more information about black and white slides or enlarged paper reproductions,please contact the Dissertations CustomerSelVices Department.
8603661
Wylie, Alexander Michael
RATIO STRAIN: CONDITIONS OF PERFORMANCE DECREMENT OF HUMANOPERANT BEHAVIOR
University of Hawaii
UniversityMicrofilms
International 300N. ZeebRoad, AnnArbor,MI48106
PH.D. 1985
Copyright 1985
by
Wylie, Alexander Michael
All Rights Reserved
PLEASE NOTE:
In all cases this material has been filmed in the best possible way fromthe available copy.Problemsencountered with this document have beenidentified herewith a check mark_",_.
1. GI"ssy photographs or pages__
2. Colored illustrations, paper or print _
3.
4.
5.
6.
7.
8.
9.
10.
Photographs with dark background~
Illustrations are poor copy _
Pageswith black marks, not original copy __
Print shows through as there is text on both sidesof page _
Indistinct, broken or small print on several pages /"
Print exceedsmargin requirements__
Tightly bound copy with print lost in spine _
Gomputer printout pageswith indistinct print _
11. Page(s) lacking when ml1terial received, and not available from school orauthor.
12. Page(s) seemto be missing in numbering only astext follows.
13. Two pages numbered . Text follows.
14. Curling and wrinkled pages__
15. Dissertation contains pages with print at a slant,filmed as received ",/
16. Other _
UniversityMicrofilms
International
RATIO STRAIN:
CONDITIONS OFPERFORMANCE DECREMENT
OF HUMAN OPERANT BEHAVIOR
A DISSERTATION SUBMITTED TOTHE GRADUATE DIVISION OFTHEUNIVERSITY OFHAWAII INPARTIAL FULFILLMENTOFTHEREQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
INPSYCHOLOGY
DECEMBER 1985
By
Alexander Michael Wylie
Dissertation Committee:
Richard A. Dubanoski, ChairmanJacCarison
Elaine M. HeibyKarl A. Minke, Jr.Arthur W.StaatsCharles Glisson
© Copyright byAlexander Michael Wylie 1985
All Rights Reserved
iii
ACKNOWLEDGEMENTS
I wouldliketo briefly, yet sincerely, acknowledge a number of people who have, either
directly or indirectly, influenced this product. Firstandforemost I would liketo acknowledge the
membara of nitJ dissertation committee, andparticularly Richard Dubanoski, for whose
permission andguidance I amgrateful in thecurrent research. I would distinctly liketo
acknowledge the friendship and intellectual stimulation of Michael Layng who contributed
explic~ly to the ideas contained herein andwhose knowledge of behavior analysis andthings
mechanical continues to be invaluable. I amalsoindebted to a number of colleagues andpeers
including, but not limhed to, JoeGrossmann, KhJones, andFrankWojnicki. Iwouldliketo
acknowledge andthankEnabling Technologies of Chicago, IL for theextensive useof their
facilhies in the production of thisdocument. I wouldliketo particularly thankT.V.Joe Layng and
Israel Goldiamond who haveeachsubstantially influenced and shaped my professional
development.
Personally, I amgrateful to JoanTagashira for herhelpand support overa pretty demanding
(and lengthy) period of time. I would alsoliketo thankmyparents for their continued support in
mypursuit of knowledge. Last, but not least, I would liketo acknowledge andthankall of my
friends fromthe Behavior Analysis Research Laboratory of the University of Chicago, including
thosepreviously mentioned, who although are not getting any younger, arecertainly "fast and
scientific."
iv
ABSTRACT
Ratio strain, or the disruption of established patterns of operant performance. was examined
in series of three single-subject design experiments specifically observing theconditions under
which performance decrement occurred. Leverpressing performance of six human responders
under the control of large fixed ratio schedules of monetary reinforcement wassubstantially
affected: by thedegree of increase in a regimen of increasing fixedratios. with performance
deterioration onlyoccurring with relatively large schedule increases (Experiment I); by the
presence or absence of alternative behavioral opportunities (Experiment III);and to a lesser
extent by the response effort requirements of the operant involved (Experiment II). Otherthan
underconditions of response decrement. pertormance wasmaintained atsubstantially higher
ratios than heretofore reported. In each of the threeexperiments. pattems of response
decrement werequalitatively andquantitatively different. suggesting thatstrained performance
canbe a function of multiple variables. Thetheoretical implications of these findings to the
concept of ratio strain arediscussed.
In addition to patterns of responding maintained by the referent fixed ratio schcdele
programmed to the main response lever, patterns of responding were observed in the presence
of a second time out lever and a thirdextra lever. A self-iliaiated andcontrolled timeout
contingency wasin effect on thetimeout leverwhile nocontingency arrangements wereplaced
onthe extra lever. It is suggested thatobserved patterns of extra leverand time out responding
wereactually induced patterns being brought about bydegrees of contingency change. Each
occurred predictably underconditions of schedule variability in the referent contingency
produced by a number of change conditions. Patterns of timeout reponding didnotsupport the
notion of such behaviors being maintained bythe removal of stimuli associated withaversive
schedule conditions (Azrin. 1961). TIme out responding was lessthanthat to the no contingency
extra leverand response rates to the time out keydecreased uponthe introduction of the time
outcontingency. Thissuggests that thetimeoutcontingency may beprocedurally classified as
punishing in thissituation in that it restricted overall responding to the timeoutkeyunderthe
reported conditions of contingency induced variablility. Thetheoretical andclinical implications of
these findings are discussed.
v
TABLE OF CONTENTS
ACKNOWLEDGEMENTS. . . . . .
ABSTRACT .
LIST OFTABLES . . . . . . . . . . . . . . . . . . . . . .
LIST OFFIGURES ... . . . . . . . . . . . . . . . . . .
LIST OFABBREVIATIONS .
INTRODUCTION . . . . . . . . .
iv·
. ...v
........ viii
. . . . . . . . ix
. xi
. 1
Physical Fatigue andRatio Strain . . . . . . . . . . . . . . . . . . . . . . . . . . . 1Definitions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2The Post Reinforcement Pause 4The Ratio Strain Literature . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . 8Trne OutPerfonnance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11ConciJsions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
EXPERIMENT I . . . . . . . . . . . . . . . . . . .19
Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19Subjecls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 19Apparatus . . . . . . . . .. 21ProceclJre . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Results 29Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
EXPERIMENT II . . .64
Method . . . . . . . . .65Subjects . . . . . . . . . . . . . . . . . . . . . . . .. . 65Apparatus . . . . . . . . .. . .66ProcedJre . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. 66
Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. ..67Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
EXPERIMENT III. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. 85
Method .. . . . . . . . . . . . . . . . . . . . . . . . . . .. . .86Subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. " 86Apparatus . . . . . . . . .. " 86Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. ..87
Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .89Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. " 95
vi
GENERAL DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . " 99
APPENDIX I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
APPENDIX II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Method 130Subjects 130Apparatus . . . . . . . .. 130ProcedJre. . . . . . 131
Resulls. '.' 134Discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
REFERENCES 149
vii
LIST OF TABLES
TABLE PAGE
1 Orderandincrementation sizeof theregimen of fixed ratio schedulesandthenumber of sessions spent responding to each FRby eachsubject in Experiment I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
viii
LISTOF FIGURES
FIGURE
1 The experimental response apparatus
PAGE
.. 22
2 Overview of the main keyexperimental result for each sUbject in Experiment I .... 30
3 sarJl)le curTlJlative records ofAR; Expo I
4 8arJlJIe CUrTlJlative records ofIN; Exp. I
5 8aJTl:>Ie curTlJlative records ofPW; Exp. I
6 5ar1lJIe curTlJlative records ofHN; Exp. I
.33
.35
.39
.41
7 Frequency of responding to the main, timeout,andextrakeysbyARin Experiment I and the clnation oftime out 44
8 Frequency of responding to the main, timeout,andextrakeysbyIN inExperiment I 45
9 Frequency of responding to the main, timeout, andextrakeysby PWinExperiment I . 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • 47
10 Frequency of responding to the main, timeout, andextrakeysbyHNin Experiment I . . . . . . . . . . . . . . . . . . . . .. 48
11 The effects 01 increasing the response forceof the mainkeyfrom 115gm.to 500 gm. on responding byall subjects in Experiment II
12 sarJllleCUrTlJlative records ofAR; Exp. II
13 saf11)1e curTlJlative records of PW; Exp. II
14 sarJlllecurTlJlative records of LD; Exp.1I
.68
.70
.72
075
15 Frequency of responding to the main, timeout, andextrakeysbyAR inExperiment II . . . . . . . . . . . . . . . . . . . . .78
16 Frequency of responding to the main, timeout, and extrakeysbyPW in Experiment II . . . . . . . . . . . . . . . . . . . . .79
17 Frequency of responding to the main, timeout, andextrakeysby LDandduration of timeout,andmain key response frequencydJring a time outinExperiment II ... .. . . . . . . . . . . . 0 • • • • • 0 80
ix
18 Frequency of responding to themain, timeout,andextra/alternative keysbysubjects LO andAR, and total eamings persession in Experiment III . . . . . 88
19
20
salll>lecumulative records of LO; Exp. III
sarJl)le culTlJlative records ofAR; Exp. III
..........90
.93
21 Overview of main keyresponse frequency in each conditionofExperiment IIIbby LM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
22 Frequency of responding to the main andextra keys by subject LMandduration of timespent engaging inthealternative in Experiment IJIbduring sessions when the altemative was present . . . . . . . . . . . . 135
23 Frequency of responding to themain, timeout,andextrakeysduring sessions when noalternative waspresent byLM in Experiment IIIb . . 136
24 sample cumulative records of LM w~h alternative; Exp. IJIb . . 139
25 Sarrple cumulative records of LM Mh noalternative; Exp. IIIb 141
x
CER
DRL
Ext.
FI
FR
GR
PR
PRP
VR
ABBREVIATIONS
. . . . . . . . . . . . . . . . . . . . Conditioned Emotional Response
Differential Reinforcement of Low Rates
Extinction
Fixed imarval
Fixed Ratio
Geometric Ratio
Progressive Ratio
PostReinforcement Pause
..... Variable Ratio
xi
INTRODUCTION
Response patterns engendered by ratio schedules of reinforcement aredistinct and
replicable. Both fixed and variable ratio schedules of reinforcement produce rapid and
consistent response rates when theoperant behavior selected for analysis is fairly simple.
However, under certain conditions these consistent response patterns become variable or
nonexistent. A general pause or abulia in response patterning under thecontrol of ratio
reinforcement schedules has been referred to asstrain. Catania (1968) defines ratio strain as
the appearance of pauses in fixed ratio or variable ratio responding at times other than after
reinforcement (ct. the post reinforcement pause, PRP) which is a consequence of large ratio
sizes and/or insufficient reinforcement. Variabilny in theconditions producing ratio strain is
suggested by the factthatsome authors report sustained responding under schedules
requiring 2000 (Reynolds, 1975) and4000 responses (Findley & Brady, 1965), while others
report strain under ratio schedules requiring asfewas4, 7, (Staats, Finley, Osborne, Quinn, &
Minke, 1963), or 20 responses (Thompson & Moershbaecher, 1980). Thisvariability andthe
presentation of a number of inconsistent and/or limiting definitions of ratio strain suggest that
the phenomenon is notwellunderstood, or at least hasnotbeen systematically analyzed. The
following review andexperiments attempt to explore issues important in theanalysis of strain
while clarifying the nature of theconcept.
Physical Fatigueand RatioStrain
Although fatigue is implied bytheterminology "ratio strain", authors arequick point outthat
strain does notreferto physical fatigue. This paradox canbeunderstood when examining the
introduction and laterevolution of theconcept. Inhis initial presentation ofoperant procedures,
Skinner (1938) discusses strain in thecontext of fatigue of a hypothesized response reflex
reserve. It should be noted that, by reflex, Skinner refers to anobserved relation between
stimulus and response asan analytical unit, descriptive rather than neurological. Purportedly,
theprocess of conditioning strengthens a reserve of responses. During extinction then, thIs
reserve becomes fatigued anddepleted. Conversly, responding will be maintained if
reinforcement is provided on a ratio schedule significantly smaller than the extinction reserve.
Hence, strain as initially formulated refers to conditions where the ratio schedule does not
provide reinforcement frequently enough to maintain or strengthen the reflex reserve.
Significantly, Skinner laterrefines many of his innial speculations on conditioning andthis
2
hypothetical reflexreserve is abandoned. However, the termstrain haspersisted in the
nomenclature, generally referring to observed decreases in frequency of responding under
certain condhlons of ratio schedules. An important distinction is seenin manylaterreferences to
ratio strainwhenthe concept is actually applied to the environment, as in expanding, increasing,
stretching, or "straining a ratio", rather than in thecontext of exhausting a hypothesized
construct. Evidence as to whyphysical fatique is not implicated in strain is provided by Ferster
andSkinner(1957). Inone example of a badly strained ratio, pauses of 80,40,45,83, and35
minutes were recorded during one experimental session with a totalof one response occurring
to the variable ratioschedule in effect. All pauses wereterminated by the introduction of a
stimulus lightcorrelated with anintermediate interval schedule of reinforcement, underwhich
responding occurred as normal. In thiscase, responding occurred underthe discriminative
control of one reinforcement schedule andnotthe other. Physical fatigue wouldprohibit
responding equallyundereachschedule. If anything, one might referto stimulus, rather than
physical fatigue. A properacknowledgement might be thatprecise discriminative control by
each schedule andassociated stimuli wasoccurring in this s"uation, andthat responding reliably
occurred onlyunderone component of the multiple schedule.
Definitions
A number of definitions of strainhavebeenproposed. In general, these definitions
suggest that strain is a function of large ratio sizes (Catania, 1968; Ferster, Culbertson, & Boren,
1975) ancllor abrupt increases in ratio size (Sulzer-Azaroff & Mayer, 19n; Whaley & Malott, 1971;
Reese, 1978). Thatthe step-size, or ratio increment regimen, maybe as important in the
development of strain as the overall raiiu~ize is suggested by a number of authors.
Sulzer-Azaroff & Mayer (1977) define strain as:
a disruption inperformance when ratio requirements areveryhighor are raisedabruptly. An individual is saidto besuffering from"ratio strain" when previouslyhighratesof responding disintegrate.
Otherauthors simply referto "training theratio" asextinction resulting frommaking a ratio
too high, too quickly(Whaley & Malott, 1971). Reese (1978) suggests that:
a behavioral requirement for reinforcement canbe quitehighif progress alongthe wayis indicated or if increases aregradual. However, abrupt increases in ratiosizecan leadto ratiostrain, that is, extended pausing afterreinforcement andoftena breakdown of the performance.
3
Reynolds (1975) similarly suggests that, although pigeons can bebrought under the
control of highratio schedules demanding 2000 responses when an initially small ratio is
gradually incremented, strain occurs when the value of theratio is increased too rapidly resulting
inabrupt pauses innormally smooth andrapid rates of responding.
If an optimal response level is desired, Sulzer-Azaroff and Mayer (1977) recommend using
small progressive andadjusting schedules to reach a response asymptote. Theyalsosuggest
thatratio strain can be avoided byincreasing ratio requirements verygradually andit canbe
treated by a temporary reduction in the ratio requirement. Ferster andSkinner (1957) provide
empirical evidence for these suggestions from the animal laboratory. They report sustained
performance without prolonged pausing by three pigeons on large fixed ratio schedules
through the useof an adjusting schedule. Anadjusting ratio is a schedule wherein the number
of responses required for reinforcement is changed progressively in terms of some
characteristic of thebehavior of theorganism (the PRP inthis case) during the preceding ratio. If
anystraining appears, as in thiscase evidenced by long PRPs, thesizeof the ratio is
automatically andtemporarily decreased. Thus, the Importance of theprocedures used when
increasing ratio schedule requirements hasbeen acknowledged boththeoretically and
empirically, yet thispoint Isoften notacknowledged during discussions of ratiostrain. For
example, the defin~ion of strain presented earlier (pg. 1)by Catania omits thispoint asdoes the
forthcoming definition presented by Ferster, CUlbertson, andBoren.
Definitions of strain may, then, differ in emphasis or emphasize contradictory points. In
otherexamples, Ferster, Culbertson andBoren (1975) limit the phenomenon to fixed ratio (FR)
schedules, while Catania (1968) suggests that strain canoccurundervariable ratio (VR)
schedules aswell. Catania indicates thatstrain is independent of the PRP whileReese (1977)
andThompson (1964) suggest thatstrain refers to extended pausing afterreinforcement. Strain
is defined by Ferster, Culbertson, andBoren (1975) as;
A termusedto describe thedecreased frequency of a performance that occurswhen the performance is reinforced ona large fixed ratio schedule. Under mostconditions strain occurs between long periods during which the performancedoes notoccur at all and periods when there arebursts of perfomance at highrates.
Although thisdefinition suggests thatstrain refers to perfomance ofbehavior at a
decreased rate rather than the absence of behavior, the authors suggest elsewhere thatstrain
refers to "thelongperiod of nobehavior thatoccurs when the number of (responses) required
4
for reinforcement is increased". In thisformulation strain is referring to an absence of behavior,
thusevenw~hin definitions, contradiction occurs.
What isclearabout ratio strain is thatthe nature of theconcept is notveryclear. This is
perhaps notvery surprising given the factthat thehistorical referent of the concept haschanged
overthe last40years. Yet. ontheotherhand, ratio strain is acknowledged asa basic concept in
thefield of behavioral psychology. It Isperhaps thisprimacy thathasledto confusion. Strain is
readily acknowledged asa general termfor rate decreases underratio schedules, thespecifics of
which areassumed to be apparent or arenotquestioned. However, upon scrutiny thegenerality
of theconcept camouflages a number of intricate andpotentially confusing relations. These may
befairly trivial, for example as in the above discrepancy between the occurrence of strain and
Iim~ation to VRancllor FRschedules. Other questions arise aswell. Therelationship of the PRP
and strain is notwellunderstood, nor is thatof strain andextinction. Does strain result from too
large of ratio sizes or from some relationship of how the ratio sizeis increased? Doother
unspecified conditions influence strain? Some confusion exists as to whether strainrefers to
behavioral disruptions or to thecomplete cessation of behavior. Some of theseissues may be
answered through theoretical analysis andyetothers need to beprimarily addressed empirically.
ThePost ReinforcementPause
The relationship of strain andthe postreinforcement pause hasbeen seento be particularly
confusing. Catania (1968) hassuggested that strain is independent of the PRP whileother
authors equate postreinforcement pausing with strain (Reese, 1977) or usethe PRP as an
indicator of strain (Thompson, 1964). Historically, post reinforcement pausing wasposited to
occur by Skinner (1938) when the reflex reserve wasin a temporary stateof strain or exhaustion
from a preceding ratio run. This would suggest that indeed, at leastInearlytheoretical
formulations, strain andthePRP areessentially analogous. This issue of similarity is not
particularly addressed in laterformulations. Skinner (1953) does suggest however thatboth in
the laboratory study of ratio reinforcement and inthepractical applications of everyday life, the
firstsignsof strain areseen in,or indicated by,thepostreinforcement pause. This relationship
canbe further clarified by exploring the nature of the PRP andresearch examining thispausing.
Ferster andSkinner (1957) describe in detail the response patterns produced by a number
of schedules of reinforcement. Final performance on a small FR willshowa high andstable rate
of responding whereas intermediate andlarge FR's will likely produce post reinforcement
pauses, growing longer asthe number of responses in the FRis increased. The length of the
5
PRP may varyalthough short pauses tendto group. After pausing, responding usually abruptly
shifts to a stable terminal rate until thenextreinforcer resulting in square break andrunpatterns
on a cumulative record. Similar response patterns aregenerated by variable ratio schedules,
however the overall rate is much more consistent withpostreinforcement pausing notreadily
apparent, occurring onlyunder very high variable ratios and then perhaps in smaller magnitude.
Theincreasing relationship between PRP length and FRsize has beendocumented by a
number or researchers (Thompson, 1964; Felton & Lyon, 1966; Powell, 1968). This relationship
haslead to confusion intheanalysis of ratio strain. It seems thatif strain is a function of large ratio
sizes, andif increases in ratio sizes produce increases in post reinforcement pausing, thenthey
might beone andthesame. Indeed, the two are treated asthesame bysome researchers
(Thompson, 1964). Ontheother hand, it hasbeen argued thatthe post reinforcement pause is a
basic and stable property of certain schedules of reinforcement (Nevin & Reynolds, 1973) and as
such is r.ot related to strain which is likely to occur under conditions of instability.
The perspective of thepostreinforcement pause being abasicproperty of certain
schedules arises from a fine grain analysis of response patterns produced byschedules of
reinforcement. Anybasic schedule of reinforcement generates a distinct andreplicable
response pattern (Ferster & Skinner, 1957). The particular pattern generated by anygiven
schedule is a function of theparticular schedule requirements andof subtle within-schedule
stimulus control relations dictated bythese requirements, aswellasbythe response repertoires
andschedule histories brought to thesituation bythe responder.
An example of how these subtle relations of stimulus control influence schedule
performance canbegleaned from a comparison of Interval and ratio performance. Typical ratio
performance has been described above andingeneral ratio schedules show consistent and
highrate patterns of responding, although differences exist between patterns produced under
fixed or variable ratio schedules. Interval schedules, ontheotherhand, produce lower rates of
responding again withdistinct patterns generated by fixed or variable intervals. A scalloped
pattern, asobserved on a cumulative record, of fewresponses following reinforcement followed
by rate increases throughout theinterval is likely to be seen undera fixed Interval, while steady
patterns aregenerated by variable interval schedules without scalloping.
The differences in rate observed underinterval and ratio schedules canbe analyzed as
being related to differences in discriminative schedule control, which actto optimize certain
patterns of performance under each schedule as a result of the different requirements imposed
byeach particular schedule. That is,given a particular reinforcement schedule, a number of
6
behavioral regimens mayultimately fulfill reinforcement requirements. Some of these patterns
willbe efficient in terms of maximizing reinforcement overtimeor Interms of the amount of effort
involved, etc., while others will be inefficient. Discriminative schedule control refers to the
particular response patterns thatevolve or areselected bythe specific schedule requirements.
In terms of thethree termoperant contingency. thediscriminative stimulus in this case is notany
exteroceptive stimulus complex, but rather a complex of subtle relations and requirements
imposed bythe governing schedule.
Ratio schedules produce higher response rates thaninterval schedules. When looking at
thespecific schedule requirements it becomes apparent that ratio schedules differentially
reinforce higher response rates, as interval schedules differentially reinforce more moderate
rates. That is, the frequency of reinforcement is increased undera ratio schedule when higher
rates of responding occursince higher rates fulfill the schedule requirements sooner than lower
rates. Under an interval schedule onthe otherhand the rate of responding is for the roost part
inconsequential andthe actual pattern of responding that is reinforced iswaiting for the interval
to elapse, hence lower rates of responding areselected. Schedule requirements can thusbe
seen to selectively control response patterns generated by each schedule.
Even within a particular class of schedules, such asthe interval schedules. thespecific
schedule requirements will influence response patterns. The scalloped pattem of thefixed
interval arises as a result of local within-schedule contingency requirements. That is, post
reinforcement pausing is pronounced underfixed Interval schedules since responses arenever
reinforced inthe initial temporal area of the interrelnforcement Interval. Ontheotherhand,
variable interval schedules willoccasionally provide reinforcement within a short time of the
preceding reinforcer. hence consistent. steady lowrate pattems of responding areproduced
with lessapparent postreinforcement pausing.
These within-schedule contingency requirements areessential in finegrainanalysis oi
schedule performance andin thisparticular case in the analysis of thepostreinforcement pause
as it pertains to ratio strain. Nevin and Reynolds (1973) suggest that although fixed ratio
schedules do notspecify a timebetween reinforcement and in general reinforce highrates, in
actuality, a minimum temporal interval between reinforcement is imposed bythe time required to
complete theratio, so that. as in fixed interval schedules, stimuli associated withshort times after
reinforcement arealso associated withnonreinforcement. ThatIs. responses arenever
reinforced in the initial temporal area of the interreinforcement interval. Therefore. these
temporal stimuli control lowrates. as infixed interval schedules. hence postreinforcement
7
pausing. Asthe sizeof thefixed ratio increases, theduration of the period of nonreinforcement
must necessarily increase, leading to increases in pausing afterreinforcement. This analysis of
post reinforcement pausing under fixed ratio schedules suggests that it is essentially
independent of strain and reconciles data suggesting an increasing relationship between FRsize
and postreinforcement pausing as being essentially similar to those patterns of pausing selected
byfixed interval schedules. Thedistribution of ratios is important aswell (Reynolds, 1975).
When, as in thevariable ratio schedule, short and medium ratios areincluded (hence variability in
thetemporal stimuli controlling reinforcement), postreinforcement pausing is notseen.
Responses, aswell astemporal relations, maybeadditional controlling dimensions in fixed
ratio schedule control which influence post reinforcement pausing. Ferster and Peele (1980)
suggest that postreinforcement pausing by pigeons under fixedratio schedules is influenced
bythe discriminative control of the number of pecks since reinforcement. Theycompared
alternating FRperformance withand without an added counter (counting pecks) to test this
notion. Reliable increases in postreinforcement pauses were witnessed with the addition of
thecounter, leading the authors to suggest thatthe added counter increases past
reinforcement pausing because of an increased discriminative effect. That is, PRPs may result
from thefact that reinforcement is never contingent on initial FR responding andthis relation
becomes more precise when anexteroceptive monitor ispresent.
A number of factors in addition to FR size have been shown to influence the PRP including
delayed reinforcement (Morgan, 1972); level of deprivation (Sidman & Stebbins, 1954); the
force requirement of the response lever (Azrin, 1959; Notterman & Mintz, 1965); theprobability
of reinforcement (McMillan, 1971); andreinforcement magnitude (Powell, 1969; Priddle-Higson,
Lowe & Harzem, 1976). Thepostreinforcement pause has, as is evident, been extensively
examined by a number of investigators largely instudies notexamining strain. Lawful relations
have beenshown by these investigators at ratio values notproducing strain. Capehart,
Eckerman, Guilkey andShull (1980), instudying thesimilarity anddissimilarity of PRPs infixed
interval andfixed ratio schedules, sU~-Jest thatthe PRP is unlikely to be dueto ratio strain or
some otherinstability in ratio performance at high values since a decreasing linear function exists
underFRschedules relating theproportion of pausing to timesince reinforcement andthis
function isat least asapparent atshorter FR values asat large FRvalues.
Thisresearch suggests then thatthepostreinforcement pause is a distinct and separate
phenomena from strain. ThePRP is infuenced by a number of varying conditions. It is possible
thatsome of the conditions influencing thepostreinforcement pause mayaswellbe influential
8
in thedevelopment of ratio strain. If thisis thecase it may be appropriate to suggest, asSkinner
does, that the first signs of strain appear in the post reinforcement pause. That is, although the
PRP is a separate phenomena, the PRP may serve asan indicator of a class of controlling
conditions that may potentially lead to strain. However, this relationship cannotbe adequately
explained until theconditions defining ratio strain aredelineated.
The Ratio Strain Literature
Little research is available specifically exploring the nature of strain or theconditions
responsible for itsdevelopment. Ratio strain hasin various usages referred to the actual
conditions producing response decrement under thecontrol of ratio reinforcement schedules,
as in "straining a ratio" (Ferster &Skinner, 1957); to those behavioral changes resulting from
certain ratio schedule conditions (Catania, 1968); andhistorically at least to thetemporary
depletion of a response reserve (Skinner, 1938). Ratio strain, while notsystematically
examined, hasbeen reported in a number of research investigations examining a variety of
diverse phenomena andbehavioral patterns (Morse & Hermsteln, 1956; Ferster & Skinner,
1957; Dews, 1958; Green, Sanders & Squire, 1959; Ferster, 1960; Hutchinson & Azrin, 1961;
Ferster, Appel & Hiss, 1962; Thompson, 1964; Findley & Brady, 1965; Weissman &
Crossmann, 1966; McMillian, 1969; Laursen, 1972; Bryd, 1973; Dardano, 1973, 1974; Barnett,
1976; Boren, Moershbaecher & Whyte, 1978; Thompson & Moershbaecher, 1980; Wood,
Happ & Adams, 1983). The majority of these investigators report or inferstrain asoccurring
uponchanges in previously stable rates of responding underthe control of ratio schedules or
upona temporary or prolonged cessation of responding andmost experimental investigators
seem content using the term primarily asa descriptor of lowered response rate. Theincidental
nature of many of these reports makes it difficult to precisely assess specific variables
occasioning the occurrence of strain.
Strain, although again incidently, hasbeen perhaps more rigorously examined in a number
of studies exploring progr6ssive ratio performance (Dardano, 1968; Hodos & Trumble, 1967;
Hodos & Kallman, 1963; Thomas, 1974; Dardano & Sauerbrunn, 1966). Under a progressive
ratio (PR) schedule, the number of responses required in successive ratios increases bya fixed
increment. In general, andalthough notexplicitly addressed, these studies usually inferstrain
asoccurring at the "breaking point", or thatpoint in the increasing series of schedules where
responding "breaks" or stops. Thebreaking point is empirically defined by Individual
researchers as thefirstperiod of X-minutes (usually somewhere in the range of 3-15) where
9
responding is absent. Thefocus of research utilizing PR ratio schedules has been directed at a
number of factors influencing progressive ratio performance or preference forprogressive ratio
versus concurrent alternatives undervarying conditions. Forexample, Thomas (1974) reports
that increased pressures of nitrogen, butnothelium, ina hyperbaric chamber lead to both
increases anddecreases interminal ratio size withthedifferential effects depending on both air
pressure andon the sizeof theprogressive ratio increment.
The sizeof the increasing series of ratios is important inprogressive ratio performance.
Hodos andKallman (1963) and Thomas (1974) each report the sizeof the PR increment to
affect the occurrence of the breaking point, with smaller increment sizes leading to greater
numbers of completed ratios priorto strain. Even though gleaned from the study of progressive
ratio performance, these results areconsistent withthose views of ratio strain positing the
importance of the regimen of schedule increases inthedevelopment or prevention of strain.
Hodos andTrumble (1967) report thatthe likelihood of engaging in a concurrent alternative
which resets the PR value increases astheprogressive ratio increases, yet isalso a function of
the number of responses required under thealternative priorto each reset. The likelihood of
responding to.thePRor engaging in thereset alternative Is alsoa function of shock intensity,
when responses to the alternative produce electric shock (Dardono & Sauerbrunn, 1966;
Dardono, 1968). Thus, Insummary, performance and the occurrence of strain under
progressive ratio schedules is influenced bythe increment sizeof the PR, whether an
opportunity exists to reset the PR, aswellas by theconsequences involved in responding to
thisalternative, andby theresponse costinvolved in responding to the ahemative in terms of
frequency requirements.
A few instances of ratio strain have been reported withhuman responders. Hutchinson
and Azrin (1961) conditioned mental patients on a FR 300knob-pulling schedule of
reinforcement without performance disruption. However, performance wasdisrupted in
instances where ratios were increased to values much greater than those previously
programmed. Contrary to theanimal literature, nosystematic increases in postreinforcement
pausing occurred asa function of increasing ratio sizepriorto disruption. This lack of post
reinforcement pausing in humans is a finding consistent with results reported by Holland (1958),
ina study of normal responders, andEllis, Barnett, and Pryer (1960), whoexamined patients
response patterns. Sanders (1969) investigated human schedule performance andfound no
evidence of PRP or strain, even at FR1000, leading the authorand others (Weiner, 1966) to
suggest thatcharacteristic animal response patterns may notoccurinhumans. Thisdiscrepancy
10
mayhowever be a function of the procedures used, as Azrin (1958) produced PRPs under fixed
interval schedules by increasing theweight required to close a response keyfrom 15gm. to 300
gm.afterPRPs were notseen with the lighter response effort.
A study by Staats, Finley, Osborne, Quinn and Minke (1963) examining complex reading
behavior suggests thattheoperant selected for analysis caninfluence theschedule parameters
sustaining performance. Theauthors examined reading acquisiton under a fixed ratio schedule
of reinforcment andfound ratio strain of response acquisition by a 4-year-old child when the FR
schedule was raised from FR2to FR 4, leading theauthors to suggest thatthe response effort
involved in the complex verbal, motor and discriminative response prevented schedule
maintenance. Following thistheauthors implemented a procedural change and examined
acquisition undera chained FRCRF schedule, consisting of alternating FR and CRF
components. "In the CRF component fourone-response schedule completions were allowed
and reinforced before a return to the FR component. Reinforcement during the FRcomponent
was limited to access to theensuing CRF series. Under thisarrangement responding was
maintained undertheFR4, and schedule Increases thereafter occurred untilstrain wasseen at
FR 7, upon which time theexperiment was terminated.
Theauthors conclude thatthechained schedule produced general changes in rate
analogous to those observed with more simple behaviors of humans andInfrahumans, and that
theprocedure wasvaluable in thesystematic study of Intermittent reinforcement andthe
complex repertoire of reading. Additionally, theauthors suggest thatthe FR component could
likely be increased to even higher ratios through appropriate manipulation of the CRF
component. Thus procedural arrangements and details of response requirements have been
seen to influence theoccurrence of strain withhuman responders.
Ona theoretical basis littlework has been undertaken specifically developing or examining
the nature of ratio strain. Hursh (1980) ina theoretical article arguing therelevance of economic
principles to the experimental analysis of behavior, suggests thattheavailability of reinforcement
for otherthanthe targeted response will influence thepersistence of a targeted response under
increasing schedule requirements. Specifically, Hursh proposes thattheeconomic concept of
open vs.closed economy Is important in understanding ratioperformance and strain. A closed
economy, in research examining appetitive performance, results when daily consumption
occurs onlyas a resu~ of thesubjects interaction withthesupply (or reinforcement) schedules.
Bycontrast, anopen economy isanyof a variety of experimental arrangements providing a
measure of independence between daily consumption and the experimental contingencies, for
11
example, when free feeding episodes follow experimental sessions. Hursh reviews and
compares dataof FRresponding under open (Felton & Lyon, 1966) andclosed (Collier, Hirsh, &
Hamlin, 1972) experimental economies, concluding that economic contingencies of
consumption strongly influence performance disruption and persistence with performance
being sustained underhigher ratios when the behavioral economy is closed. Goldiamond
(1975, 1984) similarly hasargued thatanythorough understanding of behavior must necessarily
examine notonly target behaviors and thecontingency relations influencing these but
alternative contingency relations and behaviors as theyrelate to the target or referent behavior.
In othertheoretical work, Rachlin (1978) andPear (1975) have independently proposed
theoretical quantitative formulae predicting points upon which stable schedule performance may
be disrupted, however, thiswork has notseen empirical extension. It is interesting to note that
Rachlin, in presenting his theory, makes the assumption thathighrates of responding under
ratio schedules are slightly punishing. Thisassumption is necessary inorderfor hypothesized
mathematical functions to match plotted datafrom existing experimental results. This view, that
high rates of responding are aversive andthatconditions producing ratio strain must be
aversive, is pervasive inthe experimental literature andin theoperant clinical nomenclature.
That is, ona theoretical level, strain asa behavioral phenomena is attributed to have anaversive
emotional component. In large, thisview results from a series of research investigations
examining ratio performance and time out responding. A detailed review of thisseries of articles
suggests thatperhaps thisassumption is unwarranted.
Time Out Performance
Azrin (1961) in naming the research area inanarticle entitled, Time OutfromPositive
Reinforcement suggests that "performance under a schedule of positive reinforcement may at
certain stages be aversive in spite of theapparent absence of aversive stimuli". This claim is
based on evidence of pigeons escaping fromconditions associated witha schedule of positive
reinforcement by initiating and spending increasing amounts of timein timeout as a function of
ratio sizewhen a fixed ratio schedule of reinforcement wassystematically increased andwhen
the opportunity to engage intime outwas available. Under thisexperimental arrangement, a
timeout produced a change in theambient chamber lightwhile interrupting the contingency
relanon to the main response lever such that noreinforcement wasavailable during thisperiod.
Indeed, oversessions, responding to the main keyduring a time outwas seento dropto zero.
Theduration of time spent engaging in timeoutwas reported to increase with increasing ratio
12
sizes to the point that at FR200approximately 1/2of a 60 minute session might bespent in time
out,whereas onlya fewseconds were spent in timeout atFR 65.
Pigeons controlled the length of the timeout by initiating both the time out anda second
response to thesamekey returning the experimental conditions to "time in"andthere':>y
occasioning reinforcement under the referent fixedratio contingency. Azrln examined
performance of fourpigeons responding underan increasing series of fixed ratiovalues ranging
fromFR65to FR200andreported no procedural reversals. Thetimeouts wereseen to occur
eitherlate in the post reinforcement pause or early during the ratio run leading the author to
suggest that the subjects appeared to initiate timeouts justbefore engaging in the number of
responses required by the schedule.
A systematic replication of these results wasprovided by Thompson (1965) who, in a slight
procedural variation, examined theeffects of the imposition of a 30 sec. fixedduration timeout
onperformance by two pige~ns under ratio schedule requirements. A similar increasing
monotonic function between incidence of timeout andFR sizewas obtained. Thompson
reported frequency rather thanduration measures of timeout responding when examining
response patterns to the timeoutkeyunderan increasing series of main keyFRvalues. An
abrupt reversal of FRschedule value fromthe highest ratio examined to the smallest was
undertaken following the increasing series anda decrease in timeout responding wasseen,
returning to a levelcommensurate withthatof the initial lower presentation. Thompson also
observed timeoutsto occurduring the PRPandsuggested that the PRPwasthe mostaversive
partof the FRschedule as thatwaswhere time outstypically occurred.
Theoccurrence of timeoutswithin the PRP is well documented in a second studyby
Thompson (1964) examining performance of ratsunderincreasing FRrequirements giventhe
opportunity to engage in a 30 sec. fixed duration timeout. Thompson presents datafor four
animals suggesting thatthe frequency of timeout responding increases concomitantly as PRPs
increase underincreasing FRschedules. Time outsoccurred during the PRPs. Also whena
fifth animal wasplaced undera Mixed FR25 FR225schedule. timeouts were displaced at least
25 responses after reinforcement untilthat point where ratio discriminabi!ity produced pausing
(under such a schedule, noexteroceptive stimulus is present to signal theparticular schedule
component in effect andverybriefPRPs areseen, butextended pauses arelikelyto followan
initial burst of responses which meet the lower ratio value). The results of thefirst experiment
incorporating the foursubjects suggest thatwhen looking at timeout as a function of FRsize,
timeouts increased with increasing FRs. Twoascending and descending series of FR
13
schedules were examined and reversal of effect wasobserved at each descending FRvalue
providing a comprehensive replication of the time outfromreinforcement phenomena.
Several othermanipulations, each wnh asingle animal, were additionally undertaken by
Thompson (1964). Extinction of a contingency of continuous reinforcement wasseento
produce an initial substantial increase Intimeout responding, even though timeout responding
wasalso represented underCRF. Another experiment examined the effects of increasing the
number of responses required to produce a timeouton the frequency of timeout responding
undera large FA. Time outswere observed to progressively decrease asthe number of
responses required to produce them were increased. At thehighest timeout response
requirement of FR27,notimeoutswere engaged in. Thompson reports thatoccasionally the
animals responded to the main response lever during a timeout. Thisandthe occurrence of
timeout during CRFseems to call intoquestion the j:iurported aversive nature of the ratio
response requirements. However, the three studies citedstrongly support the notion of
increasing response patterns to a timeoutkeyunderconditions of increasing response
requirements to a referent key.
Thetimeout from reinforcement relationship was further explored by Appel (1963) in a
similarexperiment withpigeons specifically examining the effects of stimulus change, alone and
in conjunction withthe timeoutcontingency, on responses to the second (time out)key.Azrin
(1961) reported that it wasthecontingency precluding reinforcement during timeout rather than
thesimultaneous stimulus change of the ambient chamber lightinvolved in the occurrence of
timeout thatwasthefactorresponsible for the observed patterns. Thisconclusion is based, in
part, on an explicit examination of thechanges inthe ambient chamber light. Azrin reported that
thechange in lightwas not maintaining the responses sincethetimeoutswereengaged in,
"regardless of whether an increase or decrease in illumination wasassociated withthe timeout".
However, hedid not examine the incidence of responses to thiskeywhen onlya change in the
ambient lightoccurred with nochange in thecontingency condition on themain response lever,
as examined by Appel.
Theworkof Appel replicates thefindings reported by Azrin wnh regard to the increasing
incidence of self-controlled timeoutsat higher ratio values, particularly when response
frequency rather thanduration measures were examined. However, a similar Increasing
response pattern wasalsoseen when onlythestimulus change component of the timeout
"package" was included. In fact, theduration of timespent engaging in thecondnion of only
stimulus change wasreported to increase as anexponential function of thefixed ratio size when
14
reinforcement wasavailable und6:" each .>1imulus condition. that is.when noextinction
c.ontingency waspresent during ,flestimulus change. This resulted in a substantially greater
duration of timespent underthestimulus change condition when contrasted tc thetimeout
condition at the highest FRexamined. In addition to measures of duration. Appel reported
incidence of response frequency to the second keyandthis index showed similar increasing
patterns of response frequency under each of these twoconditions. asthe referent ratio
contingency was increased. The substantial difference in duration noted at the highest FRwas
not evident when patterns were examined in terms of frequency andin general. from a
frequency orientation, a slightly higher incidence of responding to thesecond keyoccurred
when thetimeoutcontingency was included with the stimulus change.
Incomparing the results of time outduration reported by Appel withthe earlier results
presented by Azrin, a decreased amount of timecanbeobserved to bespent in time outat the
higher schedule values. Forexample. whereas Azrin had reported a 50%incidence of time out
duration at FR200. Appel reports a 10%incidence at FR210. A number of interesting
similarities are reported by Appel. The time outs thatwere engaged inwere likely to occurduring
the PRP or early in theratio run. Also. the birds were notlikely to engage insustained main key
responding during a timeout. OCCasionally. particularly priorto stability of performance. thebirds
would. however. runoff ratios during thetime out. Inaddition to thetimeout and onlystimulus
change conditions examined by Appel. responses to thesecond keywere alsoobserved
during a thirdcondition underwhich responses onthe second key had noprogrammed
consequences andweresimply recorded. Substantially fewer responses were reported to the
keyunderthis arrangement when contrasted to the twopreviously reported conditions.
supporting thecontention of stimulus change as a condition potentiating response patterns.
however, a slightincreasing pattern underthis condition wasalsoobserved across the
increasirlQ FRs.
Theresults reported by Appel. then. both support and replicate the earlier research of Azrin
andyetsuggest thatstimulus change alone may besufficient to maintain the increasing
incidence of responses to a second keyunder increasing schedule parameters. Appel does
concur with the suggestion of Azrin andothers that it is theaversive nature of theoriginal
schedule that is responsible for the increased frequency of behaviors which function to change
the existing stimulus conditions to ones notassociated with that (aversive) schedule. This
hypothesis is called intoquestion by the large number of variables which seem to influence the
occurrence of timeout responding (Zimmerman & Ferster. 1964).
15
Zimmerman and Ferster examined time out responding undera number of experimental
conditions and reported systematic relations between responding to a time outkey and the
controlling conditions of a referent schedule. However, the reported resuhs and relations were
substantially different from those previously reported. Theauthors suggest that some of these
differences maybe dueto a number of procedural variations in the respective studies. Notably,
Zimmerman andFerster report no monotonic relation between timeoutresponding and FRsize.
Instead theyreport timeout responding to bevariable andyet influenced bya number of
conditions. Theprocedures specific to thetime outwereseento be important. Results most
similar to those reported by Azrin (1961), Thompson (1964, 1965) andAppel (1963) above
were obtained when pigeons could initiate both thetimeoutandterminate the time outat any
timewitha second time in response. Very little timeout responding occurred whenthetimeout
wasof a fixed duration and which further required an absence of responding to anykeyduring
the time-down period inwhich thefixedtimeoutduration was elapsing. Thefollowing results
reported by Zimmerman andFerster weregenerally specific to only theself-controlled rather
thanfixed timeoutcondition.
As reported, Zimmerman andFerster didnotobserve an increasing overall incidence of time
outasthe main key FRwasincreased, rather theywitnessed an increased Incidence of timeout
during thefirstor second session following a schedule increase. This change wastransitory and
was reported to decrease by about thefourth session following a change. Furthermore, the
frequency of timeout responding wasseen to decrease markedly following a schedule change
to a smaller rather thanlarger FA. Time outs were observed to occurduring the PRP, however,
these were generally verybriefandresponses to thetimeout keytended to occurin even
numbered groups of 2 or 4 responses, etc. These response bursts onthetimeout keyresuhed
in time in conditions being reintroduced immediately or shortly afteranoddnumbered groupof
responses. Thus, the duration of timespent engaging in timeoutwasgenerally minimal. An
early and markedly increased incidence of time out responding wasobserved underconditions
of extinction, however a complete decline inthese responses occurred asthe main key
response patterns became weak. Theauthors pointout thata high preponderance of timeout
responding might be expected to occur during these later~Iriods of extinction if timeout
responding was indeed a function of aversive schedule properties, yet thiswasnotthe case.
Theincidence of timeout responding wasalso affected by the degree of stimulus change
occurring as result of the time out response. Zimmerman andFerster assessed the effects of
three increasing amounts of stimulus change following a time out response at a number of
16
different referent FRvalues. Theyfound thatgreater stimulus change occasioned greater
incidence of timeout responding as measured by frequency, duration, andthe percentage of
sessions during which timeoutsoccurred. Stimulus change did not necessarily haveto be
linked to the timeout response to occasion an increase. Timeout responding wasseento
increase when a novel stimulus light wassimply introduced on themainresponse key.
The results reported by Zimmerman andFerster suggest thatthe phenomena of timeout
responding maybe considerably more complex than the escape froman aversive situation
implied by thetimeout from positive reinforcement literature. Theauthors propose that time cut
responding occurs as a relatively weakalternative behavior thatbecomes predominant under
certain conditions of instability in the referent main keycontingency. Theconditions
occasioning temporary lossof control by theschedule of foodreinforcement include thensuch
events asschedule changes resulting in reinforcement decrement, novel stimulus change, and
presumably others.
The aversive emotional component of large ratio schedules as incorporated in some
theoretical notions of ratio strain is thencalled intoquestion by results reported in several of the
articles reviewed in thisseries. The relationship of timeout responding, ratio schedules, and
strain is Iikowise notwellunderstood or documented.
Conclusions
Systematic attempts have notbeen undertaken to quantify thebehavioral patterns referred
to asstrained andas a result of this, a variety of measurement indices havebeen used. Strain in
the sense of performance cessation hasbeen examined by arbitrarily selected breaking points
in the study of progressive ratio performanca (Hodos & Kalman, 1963). Weissman and
Crossman (1966) have assumed straining to be indicated by decreases in response rate and
propose that a quantitative indexmightbederived by measuring changes in interreinforcement
response rate excluding reinforcement time, however nocriterion otherthan a decrease from a
previous level for selecting strained performance is suggested. Pedan andTimberlake (1984) in
suggesting that reinforcement magnitude will influence theoccurrence of strainreview
traditional measures of ratio strain as (a)themaximum number of keypecks or maximum rate of
keypecking on a givenschedule, (b) thecessation of keypecking, (c) the return to the operant
levelof keypecking, and(d)a disruption in the pattern of keypecking in which theduration of
pauses increase. Theauthors alsopropose several novel measures of strainspecific to their
economic analysis (ala, Hursh, 1980) ofclosed economy keypecking bypigeons including
17
(a) changes inbody weight (b) changes in thesize andfrequency of individual meals, and (c) the
maximum price paid in the sense of a ratio relating pecks or cost to reinforcement or gain. It is not
surprising giventhisdiverse array of measurement indices andlack of consensus inthe
quantification of strain that behavioral andterminological inconsistencies abound asobserved
earlier when reviewing definitions and conceptualizations of strain.
In summary, it is suggested thatratio strain ascommonly used refers to thebehavioral
effects of a specific group of operant procedures, whereby ongoing performance decreases in
frequency underfixed or variable ratio schedules of reinforcement. Theparticulars of thisclass
of decremental operations have notbeen well idflntified, except perhaps through exclusion.
Forexample, rate decreases occurring under ratio schedules arenotlikely to becalled ratio
strain if those decreases areoccurring during extinction or following a pre-aversive stimulus in
research utilizing a condnioned emotional response (CER) paradigm (Brady & Hunt, 1955). It is
alsosuggested thatratio strain is exclusive of the PRP and does notrefer to decreases in
response probability which occur asthe PRP increases, nordoes strain referto physical fatigue.
Onthe otherhand, evidence from a number of areas certainly suggests thattheseries of
schedule increases used when increasing a ratio requirement iscritical asanoccasioning
condltlon of ratiostrain. That is, experimental investigators arelikely toclassify rate decreases as
strain following anabrupt increase ~n theschedule parameter.
Specific operations, then, which have been suggested to be influential in thedevelopment
of straininclude theabrupt increase of schedule requirements or, in general, too large of
schedule requirements. Ratio strain is often directly attributed or inferred to occurasa result of
too large of ratios, however these reports, asreviewed, are largely anecdotal by investigators not
specifically addressing performance disruption. Research reviewed also suggests thatthe
specifics involved in the constitution of "toolarge of schedule requirements" canvary
considerably, aswell. Theultimate value sustaining schedule performance is inpartdependent
onthe program of schedule increases, suggesting thatultimate ratio size is in some cases
redundant as a controlling condnion of ratio strain or is at least a function of theprocedures used
during the schedule increase regimen. Staats, et al. (1963) suggest thatthebehavior selected
for analysis is critical indefining schedule requirements thatare"too large". Other factors canbe
influential aswell. Hursh (1980) and Goldiamond (1975) suggest thatthe availability of
reinforcement for otherthanthetargeted behavior will influence thepersistance of that
behavior. It is suggested that thecontrolling conditions of ratlo strain or schedule disruption are
notadequately portrayed simply bytheacknowledgement of large ratio sizes, and infact, the
18
assessment or attribution of ratio strain in these cases oftenprecludes a thorough functional
analysis examining and delineating specific causes of response breakdown.
Thissituation is, to a large extent, a result of thefact that for the most partsystematic
research examining the behavioral occurrences referred to as strain andtheconditions
producing these has notoccurred. Rather, ratio strainhas entered into thevocabulary of the
behavior analysis community without thebenefit of substantive experimental verification or
examination. The term hascome to beused bythiscommunity asa metaphorical descriptor of
response rate decreases (used onlyundercertain conditons), onewhose historical referent has
undergone considerable evolution andonewhose basic primacy has been reified through
convention. As a result, bothterminological andbehavioral confusion exists asto the
constitution of ratio strain.
Thefollowing research attempts to examine the phenomena of ratio strain and delineate a
number of procedural conditions underwhich performance disruption or breakdown occurs with
human responders. Although nota comprehensive buta preliminary analysis of performance
decrement underratio schedules, the three ensuing experiments attempt to explore issues
arising from the preceding review. Inparticular this research attempts toclarify and define the
nature of ratio strain by undertaking a series of systematic research investigations, as noted,
heretofore lacking. The specifics of each ofthese threeexperiments willbe described in detail,
butgenerally the purposes of thisprogram of inquiry are to; 1) notonlyelucidate the concept of
ratio strain by defining conditions underwhich it occurs, butto also; 2)further examine ratio
responding andtimeout initiation, andanyrelationship of strain andtime out responding, and;
3) usehuman responders. Theutilization ofhuman subjects in the examination of ratio
performance is enlightening notonly in terms of discussions of schedule performance and
across species generality of findings, but also in terms of thefactthat ratio strain ispostulated to
beof inf!uenc.e fin areas of human ciinical concern, notably depression (Ferster, 1973). This
research specifically examines a variety of environmental situations conducive to response
decrement. Although ratio strain, perse, is a thematic issue guiding the ensuing program of
investigation, it is notsuggested or intended for these threeexperiments to provide definitive
conceptual boundaries to ratio strain. Rather, conditions of response decrement might be
considered the topic of current attention. Ideally this broad based approach will shed lighton
those iII-defined behavior-environment relations traditionally referred to as ratio strain.
19
EXPERIMENT I
Although it is often reported thatratio strain is a function ofsome terminal ratio size incapable
of sustaining performance (Catania, 1968; Ferster, Culbertson, & Boren, 1975), others suggest
that strain results fromthe methods inwhich the ratio schedule is increased (Reynolds, 1975;
Reese,1978). Thispoint is supported as well, by research exploring progressive ratio
performance where the sizeof the increasing series of ratios affects the occurrence of response
cessation (Hodos & Kallman, 1963; Thomas, 1974).
Thisexperiment explicitly addresses this issue by examining performance under large ratio
requirements following differential schedule increase regimens across foursubjects. A
hypothesis based on findings fromtheareaof progressive ratio performance predicts that the
critical feature of strained performance in thisexperiment will bethe step-size of the increasing
ratios rather thanthe ultimate ratio value attained. Thus, thefocus of thisexperiment is to
examine theeffects of schedule increases on deterioration in performance, particularly response
breakdown as a function of the different sizedvalues of s~hedule increment, or otherwise to
establish whether ratio strain is indifferent to theseries of schedule increases andmore likely to
be a function of some ultimate value. Ingeneral, this is effected by establishing a stable level of
performance on identical schedules of reinforcement for eachsublect andthen increasing the
requirements for reinforcement to high levels bydifferentially increasing regimens across
subjects andobserving anydifferential effects on schedule performance.
METHOD
SUbjects
Foursubjects were recruited from an advertisement appearing in the bi-weekly student
newspaper of a major midwest university. Theadvertisement read:
Paid subjects (min. wage) wanted for behavioral research. Must beavailable5-7daysperweek(5-6 weeks) for oneandone-half hours each day. Schedulingflexible.
Subjects were excluded from participation if initial questioning revealed thattheyhadmore
thanan introductory levelknowledge of behavioral or experimental psychology. Thefollowing
instructions were givento potential subjects upon their inquiry:
20
We are interested in looking at howpeople perform in research taskssimilar to those studied in animal laboratories with rats andpigeons. Wearelooking to seewhatthesimilarities areand whatthedifferences are.Consequ6iiUy, the experimental taskthatyouwillbe doing is analagous, youwill sit inasmall room with a panel in front ofyouthathaslights and buttonsandpoint counters. You willtry andeampoints which canbe exchanged formoney. Unfortunately. I can't go into detail asto exactly what you willbedoing because it might influence the results. Thebestway to getan idea isto actually come in and dothetaskfor a 30-minute period, andafterthattimeyou can decide if you want to continue or not. There will be noelectricshocks or anything harmful andyou can leave theexperimental room atanytime so don't worry.
Although youwill be required to come in each scheduled dayfor oneandone-half hours theexperimental sessions areonly30 minutes in length.Each daywewill runtwosessions. thefirst andlast1/2hours thatyou arehere. Between thesessions youcan relax in a waiting room soyoumightbring a book or some work to doduring thistime. You will bepaid onlyforthetimeactually spent in the30 minute experimental sessions. You will earn a$1.00 base payfor each 1/2hour(or$2.00 an hour) simply for sitting in theexperimental room. You do not have to do anything to eamthis;however, ifyou leave the room priorto the endof thesession, you will forfeit this $1.00for that session. Additionally, undersome conditions of the experiment youmaybe able to eammore money by interacting with the equipment andearning points. Each point you eamis worth 1 centso If youeam100pointsyou have eamed $1.00. This is an experiment andthings mayvary, but ingeneral, yourbase payand bonus paymight average out to about minimumwage.
If aftersitting through the firstsession you elect to continue theexperiment. you will be expected to enter intoa contractual arrangement withusto come in and participate Ina minimum number of sessions overa 4-7weekperiod. You are free to dropoutof the experiment at anytime andyouwill be paid thebase rate of $2.00 an hour. However, because thedatawecollect willbe most valuable to us onlyat the endof the experiment, youmustattend theminimum agreed upon number of sessions in orderto collect theadditional bonus point money. You willbe paidat theendof the experiment.
Relevant characteristics of each of the foursubjects. herein identified by initials only. areas
follows:
AR Subject AR wasa 45-year-old black male currently working asan intake
coordinator for a sheltered workshop. AR hasa high school education and was
currently taking part-time undergraduate coursework. Subject AR participated in
47 sessions overa seven weekperiod collecting 4315 bonus points for a total
earning of $90.15, and was thuspaid$3.84 per hour.
21
IN Subject IN wasa 21-year-old Caucasion female andcurrently a third year
undergraduate behavioral sciences student. Although IN hadtakenseveral
behavioral sciences courses. she hadlittleknowledge of behavioral psychology
andhadnocoursework inthis area or in experimental psychology. During the
course of the experiment IN worked consecutively part-time asa waitress andan
office assistant. Subject IN participated in 48 sessions overa sixweekperiod
collecting 1620 bonus points fora totalearning of $64.20 andwasactually paid
$80.40 orthe current minimum wage of $3.35 per hour.
PW Subject PWwasa 20-year-old Caucasion female andcurrently a second year
undergraduate chemistry student. During the course of the experiment PWwas
also working asa library assistant Subject PWparticipated in37 sessions overa
sixweekperiod collecting 1965 bonus points fora total earning of $56.65 and
wasactually paid$62.00 or thecurrent minimum wage of $3.35perhour.
HN Subject HNwasa 29-year-old black maleforeign graduate student fonn
Swaziland. Africa andhad livedin the United States for eightmonths priorto
being a subject. HNwasnearing completion of coursework for the M.A. degree at
a Theological Seminary atthe timeof participation. Subject HNparticipated in20
sessions overa twoweek period collecting 2395bonus points for a total earning
of $43.95. andwasthuspaid$4.40 per hour.
Apparatus
Experimental contingencies were controlled by a Rockwell International AIM-65
Microcomputer. A Gerbrands cumulative recorder was used to collectcumulative records of
schedule performance. Thisequipment was located adjacent to a small room 6 feet x 4 feetwide
x 8 feettall housing theexperimental apparatus. A one-way mirror allowed experimenter
observation andvideotaping of subjects. while white noise masked extraneous sounds. during
sessions. However, the cumulative recorder could be heard recording responses, albeit very
faintly. from inside the experimental room.
22
I 0225 I
o I stop I
0)o I Go I
Figure 1. The experimental response apparatus. The consummatory/extra responselever is centered between the main response lever on the right and the time outlever on the left. Specific relationships between each of the response levers.stimulus lights. and the positive point counter are detailed in the text.
23
During each session subjects saton awooden chair facing the experimental apparatus or
response panel, and adiagram of this apparatus canbeseen in Figure 1. Thisapparatus
consisted of awooden box 31 em. wide by36 em. deep wUh three response levers mounted 9
em. apart and centered across thefront. Thecenter lever was extended 2 em. from theoutside
two and was2 em. from thefront of theapparatus. Each lever was actually a standard adjustable
telegraph keyof thetype commercially available at Radio Shack.
Ontheapparatus a wooden housing covered all of theadjustable mechanisms andwiring
connections of these three keys so thatonlytheresponse buttons themselves were accessible
to thesubjects. Only the right-side keywasconnected to thecumulative recorder. The
response panel satona table 70 em. high such thatthelevers were easily accessible to the
subjects.
The experimental apparatus was additionally 31 em. inheight anda number ofcolored lights
andaccessories were mounted on a vertical plane located behind the response levers and set22
em. back fromthe apparatus front. Onthefar left-hand side a vertical string ofthree lights were
centered 4 em. from the panel's leftedge. These lights were colored red, yellow and green from
top to bottom, respectively. The upper redlightwas labeled "stop", while the lowergreen was
labeled "go". A horizontal string of three yellow lights were located 6 em. from thetopand16em.
fromthelefthand edge ofthe apparatus. All yellow lights were unlabeled. Each light in a string
was located 2.5em. from theonenext to U. These commercially available strings are made by
Coulboum Instruments and arecomponent modules designed foruse in animal experiment
chambers.
Two pointcounters, of which only onewas used during the experiment, were located 1 em.
below the horizontal string of yellow lights. The left-most counter was labeled "Pas. Points",
while the right-hand counter wasnot labeled, remained inoperable, andsetto zero during the
experiment. These counters were made by Sodeco Instruments. Centered below these two
counters wasa3 em. indiameter circular P.R. Mallory and Co.Sonalert. Mhoughthis was
capable of emitting a tone, Uwas notused during thisexperiment.
Theexperimental apparatus, supporting table, andchairwere freestanding and could be
moved andarranged within the experimental room bythe subject. Some IimUation was imposed
onthis movement bythree .6emelectrical cables running from the back of the apparatus to the
outside equipment. However, in general, veryIntle rearrangement of equipment occurred by any
of thesubjects.
24
Procedure
All foursubjects went through an initial training procedure andwere then placed on a Fixed
Ratio 250baseline schedule of reinforcement. Aswill be described thisbaseline included a
response initiated option to engage in a timeoutat anytime and for anyduration of thesublect's
choosing. During a time out allexperimental apparatus lights wentout until a second timeout
response (designated the time in response) occurred. Time outandtimein responses were
arranged to occur on the left hand of the three response levers, although thiskeywas
inconsequential during the ·raining portion of the experiment. During andfollowing training all
responses toward schedule completion were required to occuronthe right hand response lever,
hereafter designated as the main response key. Thekeyforce required to close the contacts of
thiskey. located .15em. apart. wasadjusted to 115gms. during this experiment.
At thecompletion of a schedule requirement on themain response keythe fouryellow
apparatus lights were illuminated signalling theavailability of reinforcement, which (throughout
the entire series of experiments) consisted of a 5 point increment to thetotaldisplayed on the
"Pas. Point" counter. When theyellow lights were on. a separate "consummatory" response to
the centerof thethree keysactually collected thepoints by incrementing thecounter. afterwhich
the yellow lights went off leaving onlythe green "Go" light illuminated. Thegreen "Go" lightwas
illuminated constantly during each session inprogress, except during a time out. Aftercollection
of the points another schedule completion could be undertaken. Thus. thethreeresponse
levers included a main response keyto the right. a timeoutleverto the left. anda center
consumatory response key. Responses to thecenterkeyat times otherthanwhen theyellow
lights wereon were inconsequential. however. aswill beseen later. these responses were
recorded bythe control equipment.
Training. Prior to the first session of training each subject was escorted intothe
experimental room and wasseated in frontof the response panel while theexperimenter directly
repeated or paraphrased the information initially provided upon recruitment (reported above on
page 20). During thistimethe red light on theapparatus marked "Stop" was illuminated. The
subject was instructed thatthis light would beon before and aftereach session andthatwhen the
green "Go"lightcame on the session wasbeginning. Noother information wasprovided asto
whatthe experimental taskconsisted of. The subjects were toldthatafterthe session andwhen
the red lightcame backon theyshould wait in the experimental booth untilthe experimenter
25
came to escort them backto thewaiting room. If anyquestions were asked theexperimenter
either repeated the relevant section of theabove recruitment information or stressed that the
question could notreally be answered asthismightinfluence the experimental results.
Three similar yetdistinct training programs were used to establish baseline responding.
Three of the foursubjects were initially exposed to an increasing Geometric Ratio 2 schedule of
reinforcement with a maximum value of FR 250 (GR 2wI FR250max). Thisschedule requires
initial completion of a Fixed Ratio 2 for reinforcement and then geometrically increases, requiring
an FR4 during thenextcompletion, followed by the geometric increases to 8, 16,32, 64,128
andfinally reaching a terminal FR 250component. This schedule wasselected because it
provides a high density of reinforcement during the initial training period.
Thefourth subject, HN, was exposed to an arithmetic Progressive Ratio 3 (PR 3)schedule of
reinforcement which provides aneven higher density of reinforcement with the increasing ratio
series being arithmetic: 3, 6, 9,12,15,18, ... etc. Afterfivesessions underthisschedule during
which the highest progressive ratio reached required 150responses for reinforcement, HNwas
switched to a FR250 schedule of reinforcement butwith notimeoutoption yet introduced, for
twosessions. Subject ARwasalsoswitched to ananalogous FR250for threesessions after
being subjected to threesessions underthe initial GR2 wI FR250max. At thispoint for each
subject the time outcontingency was introduced on the heretofore inconsequential lefthand
response lever. Subjects PW& IN were exposed to onesession onlyof the GR2 wI FR 250max
andthenswitched immediately to the FR 250 condition withsimu~aneous introduction of the
timeout contingency.
Verbal Introduction of Time Out. Pilot research intoperformance underincreasing
ratio schedules suggested thatperhaps an implicit instructional demand towards vigilance in
responding wasin part responsible forpervasive andcor.sistent high response rates, even under
thehighest values examined andextinction.. Several procedures have beenincorporated into
thecurrent study to circumvent or at least minimize anysuch implicit influences given the inherent
bias of sitting infront of a telegraph key. Forexample, thefinancial arrangements provide a base
rate of payfor simply sitting in theexperimental room regardless of relation to performance. This
may implicitly suggest that appropriate behavior might simply include sitting inthe experimental
room.
Theverbal introduction of thetimeoutcontingency wasexplicitly designed to minimize any
demand characteristics of theexperimental situation towards responding evenin those cases
26
where minimal or no reinforcement is forthcoming. To thisend, subjects were instructed that, if
so desired, they could stopresponding andtakea break by engaging in a specific response.
Baseline sessions underthe Fixed Ratio 250 schedule of reinforcement began withthe
verbal introduction of thetime outcontingency onthe leftmost keypriorto experimental session
2 for subjects IN & PW, session 7 for AR andsession 8 forsubject HN. Thetimeout relationship
was introduced as follows:
Starting today, if youeverfeel likestopping for awhile youcan press thiskey(point to leftkey) andall of theequipment willshutoff andthe lights willgo out. When youpress it again everything will resume where you leftoff.Thesession length will remain at 30 minutes regardless ofwhether youtumoff the equipment or not.
Experimental Design. Theessential task in this experiment wasto establish a stable
basal levelof performance on thesame schedule of reinforcement for thefoursubjects andthen
to increase the requirements for reinforcement bydifferent values across SUbjects andobserve
anydifferential effects on schedule performance. In thisdesign the initial schedule change
following baseline is a critical component. In thisexperiment thebaseline levelwas incremented
by a factorof 2,4, 8, & 16forsubjects AR, IN, PW, & HN, respectively. If nostrain wasseen each
subject wasrununtilthestability criterion wasmetat the newschedule value which wasthen
increased again. Theoccurrence ofstrain occasioned a reversal to a value of the schedule
sustaining performance. Responding was reinstated at these !owe:- schedule values by the
verbal instruction, "Points will be easier to get, today" priorthe thefirstsession of reversal.
Following stability, a second different incremental regimen towards thedisruptive FRvalue was
undertaken, requiring stable patterns at each intermediate value prior to further increment.
Criterion for Stability and Strain. Sidman (1960) reports that a stable or steady state
of behavior maybe defined asoneinwhich thebehavior in question does notchange its
characteristics overa period of time. A steady state of behavior is particularly useful asa baseline
fromwhichto assess theeffects of anexperimental manipulation. In theprocess of changing
from one steady stateto another following sucha manipulation, behavior passes through a
transition phase or state. The two are notcompletely separate. A stability criterion may be useful
in experiments examining steady state performance in the provision of a method to distinguish
transition states fromsteady states.
27
Theselection of a criterion forstabilny is by nature an ideographic process withthe utility of
thecrlterlon defined by the reliabilny andvalidity of thedataobtained by its use. Sidman argues
that if the steady state criterion yields orderly and replicable functional relations, it maybe
accepted as adequate. It need not, in fact, represent the ultimate stable stateof behavior in
question. If thecriterion of stability selected is one through which the behavior must inevitably
passon it'swayto an ultimate steady state, thenthe obtained datawill be orderly andmeaningful.
Thecriterion should select reproducible andgeneralizable states of behavior.
Optimally, a stability criterion might be developed from a longtermdescriptive analysis of
behavior undera steady state. However, a number of events maypreclude such an analysis
including limitations in resources or, as in thiscase. tnsparticular expeiimental question being
asked. Forexample, research tasks mayspecifically examine theeffects of a manipulation on
steady states during which a criterion of stability maybeparticularly useful, or may focus onlyon
transition states during which a stabilny criterion maybeof lessuse, or some intermediate aspect
thereof.
Experiment I posits ratio strain to be primarily a transition phenomenon related to theprogram
of schedule increases. Thisanalysis of ratio strain bynecessity dictates an examination of those
transition states where performance is deteriorating aswellas those states where performance is
being maintained, or is minimally ina transition statesuggesting anellor approaching steady state
performance maintenance. Therelatively laxstability criterion incorporated in thisexperiment
therefore is notselected asonewhich will portray an ultimate steady stateof perlormance but
rather onewhich suggests a likelihood of a similar ultimate steady state.
In examining response frequency, three pointsof reference are minimally necessary to
suggest stability. More arepreferable. Visually andstatistically these should portray an absence
of tendency in behavior to be increasing or decreasing in rate. That is, the third point should be
within the range established bythe first two,or the range of the previous points if morethanthree
reference points exist. Thedegree of variability within this range to betolerated in a definition of
stability will ultimately be determinecl by theconsistency of the functional relations soobtained
andby the degree of experimental control thatcan be obtained.
Optimally inthis experiment, the range of variability defining thesteady state would be
visually minimal. Statistically, in those cases defying visual concordance, the maximum acceptable
variability proposed consists of each of three minimum reference points falling within a range of
20%above or 20% below the mean response rate of those three points. Forexample, the mean
of the three consecutive datapoints, 4.5,5.5, and5.3 responses persecond is 5.1. A range of
28
20%above and below would incorporate response rates from 4.1 to 6.1, therefore thevariability
represented bythese three points would be within the limns of acceptability. Additionally, the
thirdpointfalls within the range established by the firsttwomeeting the requirements suggested
earlier.
Thus, stability is defined in thisexperiment minimally bytheoccurrence of noupward or
downward trends in response rate over a minimum of three sessions, with variability of response
rate for anygiven point lessthan20% above or below the mean response rate of those three
sessions. Again, thiscriterion is notmeant to reflect anultimate steady state of performance but
rather to suggest the absence or occurrence of ratio strain. The crneria forassessing ratio strain,
then, are inextricably linked to this definition. Visual observation of performance may allow more
stringent analysis of stability overthese minimum statistical requirements, where feasible.
Ifwe assume this stability criterion, onesetof criteria for assessing strain arise logically.
Previous discussion has suggested thatthe termratio strain has been utilized asa rather
picturesque metaphor to suggest changes, notably decreases inperformance, undercertain
conditions of ratio reinforcement schedules. A criterion suggesting theoccurrence of strain
thuswould include a cleardownward trend in response rate during the transition phase following
anexperimental manipulation and/or steady state performance ata substantially lower response
rate than thatof thebaseline level. A quantification of this decrease in mean response rate as
greater than 20% from the mean baseline levelis commensurate withthepreviously established
stability criterion.
In evaluating the effects of different sized schedule or program increases, performance was
examined under the control of the baseline FR250schedule aswell asthe higher FRschedule
values of 500, 1000,2000,4000, & 8000. In addition, theperformance ofsubject IN was
examined during an extended period of extinction. Table1 shows the exact size of
incrementation of the fixed ratios for each subject insummarizing theorder ofschedule
presentation and the number of sessions spent responding to each value of theschedule. As
canbe seen, reading down, subject AR spent sixsessions at FR 500afterfourbaseline sessions
of FR250. Thiswasthenconsecutively doubled to FR 1000,2000, &4000 for3,5 &7 sessions,
respectively. These lastthree fixed ratio values comprised the initial values ofschedule increase
for the othersubjects. That is,subject IN initially wasexpoG~ (0 an FR1000 following baseline,
PWwasexposed to FR2000, while HN wasexposed to FR4000. Notice, forsubject PW, that
afterthis innial exposure to FR 2000 following baseline, a reversal to FR500 occurs for five
sessions atwhich time the incremental regimen is analogous to and replicates thatof AR. Onthe
29
otherhand, the fixed ratio value is increased further forsubject IN from FR 1000 to FR4000, FR
8000and thenthesubject is observed under23 sessions of extinction.
Table 1
Order and Incrementation Sizeof the Regimen of Increasing Fixed Ratio Schedules andthe Number ofSessions (In Parentheses) SpentResponding to EachFR by EachSubjectin Experiment I.
AB ~ fW .I:Iti250 (4) 250(6) 250(7) 250 (5)
500 (6) 1000 (5) 2000 (3) 4000 (4)
1000(3) 4000 (5) 500(5) 250 (1)
2000 (5) 8000(7) 1000 (4) 500(3)
4000 (7) Ext (23) 2000 (4) 4000 (4)
Theapriori contractual arrangement with subject HNwasfor considerably shorter participation
thanthe othersubjects andconsequently afterthe foursessions of responding underFR4000,
a single sessions reversal to FR 250occurred afterwhich timetheschedule was increased to FR
500for 3 sessions, the minimum necessary to establish stability.
RESULTS
A number of performance measures wereemployed in the analysis of program increases and
strain. The various response patterns generated bythose fixedratio schedules programmed on
the mainresponse leverwere empirically analyzed intermsof responses persecond. Cumulative
records of these patterns were collected providing a visual representation, aswell. Although
these two measures weretheprimaiY toolsof analysis in thisexperiment, the frequency of
responses to both thecenterkeyandthetimeoutkey, aswell as theduration of timeengaged in
timeout,werealsoexamined.
Theeffects of varying the increment sizewhen programming schedule increases on stable
performance and strain canbeseenin Figure 2. This figure provides a general overview of
Experiment I in terms of themainkey responses persecond for each subject. Responses per
second are indexed along theordinate while experimental sessions are represented on the
abscissa with thechanging experimental conditions indicated by vertical bars.
30
Subject AA~
~D- n .-a~~ ~g.g ..CI..,_
f-f-I- FA 250 FA SOD FA 1000 FA 2000 FA 4000~-.
76543Z1o 7 8 !) 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 252627 41 42 43 44
Subject IN6~
S C)o9 n .a..a432 FR 250 FA 100010
e 9 12
f-
~FA 2000
- a-" n a--" n n n n 0-9 9 • ~n .n...-d=I-- FR 250
~ FR 500 FR 1000 FR 2000 FR 4000I-
"'CCooQ)
CIJL.(1) Subject PW0.o 6CD 5o 4c: 3o 20.1o 0CD S 6 7 8 9 10 11 12 13 14 lS 16 11 18 19 20 21 22 23 24 2S 26 21 28
a:Subject HN
e54
321o
f-
f-~FR 4000 n a.-- c III
f-
f-
~T251°f- FR 250 FR 500-8 9 10 11 12 13 14 lS 115 17 18 19 20
Sessions
Figure 2. The effects of varying increases in the sr.!!el1l.lll! Clf rp.inforcement onfrequency of responding to the main key by all SUbjects during Experiment I.
31
Theresponse rate index wascalculated by dividing the total number of responses to the
main response lever by the number of seconds in a 30minute session minus thetimeout
duration. Responses to the main response leverduring a timeoutwere calculated separately and
arenot included in this index. Response rate under ratio schedules of reinforcement mayrefer
to either the local rate, which is the rate of responding minus the postreinforcement pause and
usually reflects a higher rate of responding, or to anoverall rate which incorporates the PRP into
the calculation. In thepresent research theamount of timespentpausing afterreinforcement
was generally very brief (less than three seconds) by allsubjects. As a result these two indexes
were very similar andconsequently onlyoverall response rate will be reported.
Ascanbe seen in Figure 2, the baseline response rates of the foursubjects arequitehigh,
in the range of 4-6 responses persecond. These rates remain quite highandeven increase
through the experiment for subject AR, with the exception of the temporary decrease seenin
session 21. Note the interruption of consecutive sessions occurring with regard to ARbetween
the thirdandfourth session under the FR4000. Other experimental arrangements, these being
the subjects of experiments" & III,were introduced and withdrawn during this period andthese
wi;1 be detailed anddiscussed further during each ensuing experiment. Forthepresent
purposes, the reversal nature of the experimental design allows thecurrent comparison. Subject
AR was exposed to thesmallest series of schedule increases.
Thesame general pattern is seen forsubject IN in thata high consistent response rate is
generated during the more abrupt ascending schedule series of FR250,1000,4000, and8000.
In thiscase, response rate decreases slightly, yet consistently, witheach schedule change.
However, overall performance forIN and AR canbecharacterized similarly as represented bya
generally highand persistent rate of responding on themain leveracross the two regimens of
increasing schedules.
Thisis in contrast to thepatterns generated during extinction byIN andthose generated by
subjects PWand HNunderthe largest schedule increases of FR2000 and4000, respectively.
Two clearinstances of strain as identified both visually and empirically by the aforementioned
criteria occurinthe lattercases. Following thehighrate baseline performance ofsubject PW, ratio
strain occurs when the schedule is increased to FR2000. Following thisdisruption performance
wasreestablished at the lower schedule of FR500 and the high response rates were immediately
reinstated by theverbal instruction, "Points will be easier to get, today" priorto session 12. The
factthat these high rates were maintained during the following 16 sessions while the schedule
value was increased gradually to FR 4000 ina fashion analogous to thesmall increases reported
32
for AR lends support to the suggested importance of program influences in the presence or
absence of strain.
A second instance of performance deterioration following these larger schedule increases is
seen in the case of HN. Strain canbeobserved to occurduring theschedule increase to FR4000
but notduring thebaseline or under thefollowing lowerratio schedule of FR500. Following the
fourth session of FR 4000 responding wasagain verbally reestablished, as reported for PW,
underonesession of FR 250witha subsequent increase to FR500. As reported in the
Procedures section forHNit wasnotpossible to further continue witha gradual program of
schedule increases to the FR 4000 value which produced the response ratebreakdown. It might
be noted, however, that each of thethree othersubjects showed sustained performance at that
value following a lesser series of increasing steps.
Response patterns during extinction were examined withsubject IN. During FR 8000 the
subject was onlyreceiving onereinforcer persession because of thetime commitment involved
incompleting the ratio requirement, yeta high sustained rate of performance wasobserved. IN
wastherefore run under anextended period of extinction to examine whether if indeed it wasthe
response-reinforcer relation thatwas maintaining thisperformance. During 24sessions of
extinction several severe instances of performance disruption areseen, increasing in frequency
of occurrence oversessions, asduring sessions 26,36, 46& 48. Response rateis lower during
the lasthalfof theextinction condition thanduring the initial sessions suggesting thatthe
process of extinction is Inprogress, yet incomplete. Theagreed uponcessation datewasmet
aftersession 48 andit wasnotpossible to continue running further extinction sessions withIN.
However, it does appear thatthe high rate patterns observed under the FR 8000 were indeed
maintained by the minimal reinforcement obtained.
CumulatIveRecords.
A more detailed analysis of response patterns may begleaned from cumulative records
representative of thevarious conditions for each subject. A cumulative record is a graphic record
of responding whereby a recording pen moves along theabscissa of therecord from leftto right
withpassage of timeand along theordinate each time a performance occurs (Catania. 1968). The
rate of performance is therefore demonstrated bythe slope of thecurve, withincreased rate
reflected by increased slope.
33
Figure 3. Sample cumulative records portraying response patterns generated during eachcondition of Experiment I by Subject AR. A complete collection ofallcumulative recordsgenerated during this experiment is included in Appendix I.
34
~-""========C\I-=====-':==-g: -=========--- I
0--====:=====-;~
35
Figure 4. Sample cumulative records portraying response patterns generated during eachcondition of Experiment I bySubject IN. The upperfour records represent responding underthe increasing series of fixedratio schedules while the lowersix records show the firstthreeandfinalthreesessions of extinction. A complete collection of all cumulative recordsgenerated during thisexperiment is included in Appendix I.
PR 250 PH 1000
36
•••• » i ».»» i; » .; I I ••• i»'
PH 4000
Bxt1l1ctton 1 ot 24
Ext1no1:1on :3 ot 24
Extinotion 23 ot 24
Bxt1Dct1on 2 ot 24
Extinotion 22 ot 24
Bxt1nct1on 24 ot 24
37
Figure 3 shows representative cumulative records of the experimental condltlons
experienced by subject AR. Two recording pens wereused in obtaining these records. One
pen. referred to astheresponse marker was linked to the main response lever of the
experimental panel andrecorded each response by moving successively along theordinate. The
response marker automatically reset when reaching thefurthest point from theabscissa asshown
onthese records bya line exactly perpendicular to the abscissa drawn asthepenreturned toward
the abscissa. Approximately 450 responses were required to increment the response marker to
thepoint of reset. Additionally. when the ratio requirements of the fixed ratios experimentally
programmed on themain keywere completed, the response marker engaged ina brief downward
deflection corresponding to the point of reinforcement availability. These reinforcement hatch
marks canbeseen periodically yet regularly along thecourse of the response marker.
The second penreferred to asthe reinforcement marker, simply moved along theabscissa
withthepassage of timemaking a downward deflection corresponding to theavailability and
collection of the points used asreinforcement. The linedrawn bythis pencanbe easily used to
observe reinforcement density under each condition of theexperiment. Each uninterrupted
cumulative record represents a 30 minute experimental session. Ascanbe seen in the
cumulative records shown forAR in Figure 3, a continuous highrate of responding reflected by
the steep slope with no interruptions occurs under all conditions. Decreases in reinforcement
density canbe observed overthecourse of theexperiment, to the point where onlytwo
reinforcers areobtained during thesession under FR4000. Nopostreinforcement pausing
occurs at anyschedule value.
Similar patterns canbe seen bysubject IN in the fourconditions priorto extinction
represented in Figure 4. Responses to thetimeoutkey are alsorepresented on thecumulative
record. These are indicated bya premature reset of the response marker and two such
occurrences areobservable during thefirstschedule completion of FR4000 byIN in thethird
record. The timeouts portrayed areeach less than onesecond induration and an immediate
resumption of main keyresponding canbeobserved. Thesecond schedule completion
continues withnofurther timeouts. A considerable PRP occurs following the completion of the
FR 8000 asseenin thefourth graph of Figure 4. This pause is approximately twominutes in
length, lasting untiltheendof thesession. Ingeneral, IN exhibited verylittle pausing after
reinforcement asshown in theother records andthisparticular example is likely a function of the
historical relation under FR8000 whereby IN could consistently onlycomplete oneratio during
the30 minute session.
38
Figure 4 also shows thefirst three and final three sessions of extinction for IN. A variety of
response patterns canbeobserved. The fifth, seventh, and ninth cumulative records of Figure 4
all showthesame high rate response patterns exhibited earlier when reinforcement wasavailable.
The initial disruption of this steady pattern canbeseen during the second session of extinction,
shown by the sixth record. This session is characterized by a number of brief timeouts
interrupting high rate performances asat A, several substantial pauses occurring independent of
anytimeoutas atB anda second distinct lower rate of responding atC. Anequipment
malfunction occurs at 0 obscuring thebehavioral pattern portrayed. A time out is engaged in yet
the response marker does not reset completely untila laterpoint. A pause is reflected, however,
it is notcertain asto whether this is theactual behavior occurring here or not.
A distinctive pattern of responding is seen in theeighth graph of Figure 4, the 22nd session
of extinction. Hundreds of resets are shown representing consistent responding onthe timeout
keyaswellasthe main response key. The finalsession of extinction isshown in thefinal record
of Figure 4. A much lower rate of responding is observed without the repeated occurrence of
timeout responses. A good dealof variability occurs in this lower rate of responding withsome
shortbreaks of lessthan 1Q seconds occurring.
Cumulative records from each condition experienced bysubject PWcanbe seen in Figure 5.
Thefirst and lastfour cumulative records represent responding underthe FR250baseline and
the gradual program increase of FR 500,1000,2000& 4000, respectively. Theresponse
patterns reflected by these records represent a high andstable rate of responding withno PRPs
exhibited. This is in contrast to the second, third andfourth records which portray all three
sessions of the initial abrupt schedule change to FR2000. These threerecords arecharacterized
by a break andrunpattern withan increased amount of time spent pausing during the final two
sessions. A number of timeoutsoccur during thefirstsession butonlyone is observable during
the lasttwo. Only one reinforcement is obtained early Inthefirstsession, while the ratio
requirement is notcompleted during the finaltwosessions at FR2000, thusa functional
extinction of sorts Is observed.
In Figure 6,cumulative records forsubject HN canbe seen. The firstand fifthof these
records represent responding before andafterthe initial large schedule change andportray
response patterns generated underthe baseline ratio of FR250. Post reinforcement pausing is
clearly observable in these records. Notice that in each of these twosessions a timeoutoccurs
briefly andthatin each case thetimeoutoccurs during the PRP.
39
Figure 5. Sample cumulative records portraying response patterns generated during eachcondition of Experiment I by SUbject PW. The second, thirdandfourth records present acomplete record of responding overthreesessions of the initial presentation of FR 2000. Acomplete collection of all cumulative records generated during this experiment is included inAppendix I.
n 250 n 2000 1 ot J
40
n200020tJ
n 2000
.. 1000
n 4000
/
41
Figure 6. Sample cumulative records portraying response patterns generated during eachcondition of Experiment I by Subject HN. Thesecond, third and fourth records showresponding during the first. second andfinalsessions of FR4000, respectively. A completecollection of allcumulative records generated during thisexperiment is included in Appendix I.
FR 250 PR 4000 1 ot 4
r "'-T • , , , i » » i' » , , » » » , , i » » iii» » • i , , , »
PR 4000 2 at 4
FH 250
I
PR 4000 4 ot 4
PH 500
,. -.- , • » • » , • » » • , • » • , ,
..,.I\)
43
Thesecond, third, andfourth record portray 3 of the 4 FR4000 sessions for HN(thethird
session of FR4000 is notshown). During the initial session of FR4000 a number of shortbreaks
areseenaswellas a good number of short timeouts, however, overall response rates remain
high although variable. Responding decreases in frequency during thesecond session
following an initial burstanda number of time outscanbeobserved, oftenof briefduration.
Towards the endof the session pausing isevident. Bythe lastsession of FR4000, mainkey
response frequency hasdecreased even more andwhen occurring the rate is lowerand more
variable. A highfrequency of timeout responding is also seen. Aswith PWunder the initial
exposure to FR 2000, HNonlyobtains onereinforcer during the first session of FR4000withno
further schedule completions during the following three sessions.
Thefinal record shown in Figure 6 portrays the final condition of FR500. Response patterns
aresimilar to the baseline patterns, with PRPs evident. Visual observation suggests a slight
increase in PRPs withthe schedule increase from FR 250to FR500, however anyincreases are
negligible.
Time Out Infonnation
A more detailed analysis of performance interms of the interaction of mainkeyresponding,
timeout responding and responding to the thirdcenter keyat times otherthanwhen
reinforcement wasavailable canbe seen foreachsubject in Figures 7, 8, 9, & 10. These
response patterns areshown for subject AR in Figure 7. Thetop graph of this figure reproduces
the main key response rate information presented earlier in Figure 2,while the middle two graphs
portray timeout responding. The lowergraph represents frequency of responding to the third
center key. Recall that responding to thiskeywas inconsequential at times otherthanwhen
reinforcement wasreadily available assignalled by theillumination of theyellow lights. Hereafter,
thiskeyshallbe referred to asthe extra key.
Since the response patterns to the main response keyhavealready been described earlier,
thecurrent discussion will focus on the ancillary timeoutandextrakeypatterns. Thefirstsix
sessions, andparticularly thefirst three, in Figure 7 show the evolution of response patterns
generated by the main keycontingency everpatterns to the otherlevers duringtraining. Notice
thedifferences of ordinate value for each graph. At session 7 the timeout contingency was
introduced to thesubject andthe formal baseline wasbegun.
SubjectAR
f0o- I~
~~
~7~I~~
FR 250 I FR 250No r. o. I with FR 500 FR 1000 FR 2000 FR 4000.
I Tjme Out •123 456 7 e 910111213141516171819202122232425262741424344
44
9 1011 12131415161718192021222324252627 4142 43 44Sessions
Figure 7. Response patterns to the main key. time out key. and extra key by subject ARin all conditions of Experiment 1, and duration of time spent engaging in time out.
SlJljecl IN
4DO (374):0-
~ (275)
I..II. 6
~ 4
.! 2I-
0
3 6 9 12 15 18 21 24 27 30 33 36 39 42 4S 48
Sessions
Figtre B. Frequency of responding to main key. time out key. and extra key by subject IN In Experiment I.
45
46
If one examines the frequency of responses to the timeout key before(sessions 1-6)and
after (sessions 7-44) the introduction of the time outcontingency, it is possible to observe that
the contingency actually serves to restrict overall frequency of this response. Veryfew time outs
by AR occuroverthe course of this experiment. Thefew that do tendto occur, occurwithin the
first twosessions aftera schedule change, butareof relatively briefduration.
Notime outsare engaged in duringthe FR 250baseline andthe first occurs during the
second session of FR500. Withthe increase to FR 1000, two timeoutsoccur, whilethe greatest
increase in timeout duration occurs withthe schedule at FR2000. Thisparticular timeoutoccurs
during the session wherethe mainkey response patterns havedecreased in rateandthe
engagement in timeout maybe partially responsible for this drop. It is possible to suggest at this
pointthat the magnitude of time out responding in termsof frequency andduration is increasing
with increasing schedule change, however the magnitude of the time out measures at FR4000
andthe generally negligible aspects of all timeoutsalsosuggest that this conclusion maynotbe
warranted.
The response patterns to the extrakeyseenin the bottom graph of Figure 7 for ARarealso
fairlynegligible. In general, very few responses areseento this inconsequential key, however
periodically overthe sessions a burstof responding mayoccuronlyto subside in subsequent
sessions. A higher frequency of responding occurs to this key thanto the timeout key. It is not
applicable to graph frequency of responses to the extra key during sessions 41-44as a prior
experimental condition influencing the probability of theiroccurrence was undertaken.
The responses of IN to eachof the threekeyscan be seenin Figure 8. Again, the topgraph
reflects the previously discussed mainkey response pattern. Thecentergraphshows
responses to the time out key. Although notgraphed 1124responses occurto the timeout key
priorto the verbal introduction of the timeout contingency in session 1. Veryfewresponses to
this keyoccurafterthe introduction of the timeout contingency untilextinction when in particular
two substantial response increases occur in sessions 26 & 46. More responses to the timeout
keyoccur laterrather than earlier in the extinction condition. Of thetimeoutsthatoccurpriorto
extinction, they occurduring the first session of a schedule increase in two instances. The
amount of timespentengaging in each timeoutwasverybriefby IN, typically lessthan1 second
per engagement, andtime outduration is notgraphed.
Perhaps of moreinterest in the caseof IN is the distribution of responses across the
experimental sessions to the extra keyportrayed in the lowergraph of Figure 8. During the first
two experimental sessions a relatively highfrequency of responses occurto this key, however
47
Subject PW
..5II•C
Q'O 4a,C.. QII 0 3a: II
(I)
"" 2II N~&C 1
..-4II
0~
I-
~ "- " .S' nn n n~I- -- ~D
I- FR 2000
~(FR 250 L FR 500 FR 1000 FR 2000 FR 4000
I
2 4 I) 8 10 12 14 16 18 20 22 24 26 28
I) 8 10 12 14 16 18 20 22 24 26 282 4
4
2
O'-ioIIIIII'~_""""'''''-''''''''''''_.-...II...._w.......................ue;:
500
..QI..CClQ,litQI
It>. 30QI~
1lI 20.....)(w 10
0
-(457)
~ ,4OIl'\ 1-l-
eI-
I~ ~~ 0.... a..... ~2 4 Ii 8 10 12 14 1G 18 20 22 24 2G 28
Sessions
Figure9. Frequency of responding to main key. time out key.and extra key by subject PW in all conditions of Experiment I.
48
Subject HN-0
>.cGI 0 FA 250 FA 4000 FA 250 FA 500:l'g 6om......~
• &='< t:~in Key~GI 4~Q, D.-C
05" Extn KeyGI2c .. --... c
0 0
z::~ 0GI
II 10 11 12 13 14 15 16 17 18 19 20Ill: 8
>.u 400c FA 250GI;:,13 300GI
'"u, 200II
~ 1000Q,
• 0GI~ 8 9 10 11 12 13 14 15 16 17 18 19 20
~:lCl
200GIlIi""'-0 150~c
0~oo GIl 100erno e 50......~• 0...:lQ
I- FA 250 FA 4000 ~A 25b FA 500
l- N~
j-~ ~
8 9 10 11 12 13 14 15 16 17 18 19 20
Sessions
Figure 10. Response patterns to the main, extra, and time out keys by
HN in Experiment I and the duration of time spent engaging in time out.
49
these decrease overthe next fewsessions asthe main key response pattern becomes strong. A
temporary increase in responding to theextrakeyoccurs withthe schedule increase to FR 1000
andagain when increased to FR 4000. In both of these cases tnepattern returns to a lowlevelby
the thirdsession of each condition. A slight increase is witnessed withthechange to FR 1000.
Thispattern is notobserved withthe increase to FR8000, however, a higher frequency of
responding to this keyoverthe entire condition is observed.
During extinction an increase in extra keyresponding is observed aswell. Aswiththetime
out responding, the occurrence of extra key responding occurs withgreater frequency later
rather thanearlier in the extinction condition and an initial increase canbe observed during the
second session. When comparing main keyand extra key response patterns there is, in general,
a converse correspondence. That is,the lowest main key response rates areseenin sessions
26,36, 44,46, & 48,whereas thehighest extra key response patterns areobserved during the
same sessions.
Responses by subject PWto each of the three keysduring the course of thisexperiment
can be seen if Figure 9. Very fewresponses occur to eitherthe timeoutkeyor the extra key
except duringthe main keyperformance deterioration generated by the abrupt schedule
increase to FR2000. Notice thedifferences in theordinate value on each graph. Although both
timeout andextrakeyresponses occurduring the initialpresentation of the FR 2000, by far the
larger response increase is to theextralever. Incomparison timeout responding is negligible.
However, giventhe baseline andsucceeding frequencies of the timeout responses it is
suggested that those conditions producing strain are likewise generating the increase in timeout
responding. Theamount of timespent engaging in timeoutwasagain verybriefby subject PW.
A unique pattem of responding canbe seen by subject HNin the top graph of Figure 10.
Response patterns to both the main keyandextra keyareportrayed inthisgraph. Themainkey
pattern hasbeen previously discussed. Theresponse rateto theextrakey, alsocalculated in
responses persecond, hasbeen plotted on thesame graph andas can beseenis quite similar in
nature. Thissuggests andvisual observation confirms that thesubject was responding
simultaneously to bothresponse levers andthatthispattern wasadventitiously reinforced by the
fixed ratio contingency programmed onlyon themain lever. As might be recalled, subject HNwas
theonlysubject to consistently show PRPs andthe additional response effort of this two handed
pattern, might be onecontributing factor.
Th~ lower two graphs of Figure 10show time outdatafor HN. Ascan beseenin thecanter
graph the freqtJ/:mcy of responding to the timeoutkey increases withthe advent of the FR4000
50
contingency and reverts to baseline levels upon withdrawal. The final session of FR4000 in
particular shows a drastic increase in response frequency. This pattern of time out increases
during strain and reversal upon reinstatement of main keyresponse patterns issimilar to that
displayed in thecase of PW in Figure 9,withthe exception of the disparity in overall frequency of
timeoutsas indicated byordinate values on each graph.
Themeasurement of time outby duration shows thissame general pattern, as shown inthe
lower graph of Figure 10.An increase in the amount of timespent engaging intimeoutoccurs
during the FR4000 overboth the FR250 baseline or schedule increase to FR 500. Theanalysis
of timeoutduration alsosuggests thatthetimeouts engaged inby HNwere very brief.
DISCUSSION
Human operant responding andconversely thedisruption of performance or ratio strain was
observed to be distinctly influenced bythe regimen of schedule increases selected for usewhen
increasing ratio size. Ratio strain occurred when large increases in response requirements were
programmed and notwhen more moderate increases were scheduled. That ratio strain is notthe
outcome of some tenninal ratio size incapable of sustaining performance is evidenced bythefact
thatsubjects were quiteable to sustain responding at ratio sizes considerably larger than those
inducing strain when moderate increases were programmed. When a relatively moderate
program of schedule increases wasfollowed, responding persisted bythe human subjects even
under the largest fixedratios examined and initially at least under extinction foronesubject,
providing strong evidence thatundertheconditions in thisexperiment thecritical feature of
strained performance was theschedule increase regimen.
In the present research responding waspersistent under FR4000 for subjects ARandPW
and FR 8000forIN following the moderate series of schedule increases of FR500, 1000, 2000,
4000 (AR andPW) and FR 1000,4000, and 8000 (IN) from thebaseline conditions of FR250.
Larger increases frombaseline were shown to produce performance disruption when increasing
theschedules to FR2000 for subject PW and to FR 4000 for HN. The performance patterns
generated in thecase of PW are particularly illustrative ofthe relation between ratio strain and the
procedural regimen incorporated when increasing response requirements. Immediate and
pervasive performance disruption wasobserved following the initial schedule increase from
baseline to FR2000, however following reinstatement of response patterns at a lower FRand an
ensuing moderate series of schedule increases, performance was strong andpersistent at the
same FR2000 andthen even at the larger ratio of FR 4000.
51
These results suggest thatthe ratio size perse is notas important of variable in ratio strain as
is the environmental context within which thatcontingency requirement is placed. When, as in
thisparticular case, a substantial response requirement is placed within the context of a series of
increasing response requirements, the specifics areattainable whereas without a history of
gradual increases, performance breakdown occurs. To some extent this history influences
performance by providing variability in one aspect of the controlling contingency complex or
cthcrwcc ~!o~g oneparticular dimension of thecontrolling stimulus, lesstechnically along one
particular dimension of the rules for reinforcement, thatof frequency requirements. With a
gradual series of increases a history of change along thisdimension is established facilitating
transfer of schedule control overresponding underconditions of schedule change to new,
increased frequency requirements. Conversely, without such a history the behavior of subjects
may notfulfill requirements of this newrule for reinforcement frequently enough for this
contingency to establish formal control overresponding andparticularly in cases of extremely
large ratio increases, behavior maydecrease in frequency showing "strain" even though the
frequency requirements arephysically attainable asevidenced behaviorally following a gradual
series of increases.
This suggests that ratio strain in thiscontext may besimilar to theperformance disruption
which occurs during transition phases following a distinct change inschedules controlling
performance. Earlier it wassuggested that each schedule of reinforcement actsas a
discriminative stimulus in thateach schedule selectively reinforces certain response patterns.
When schedule requirements arechanged a disruption in previously stable pertormance OCCUiS
during a temporary stage of response instability referred to as thetransition phase. Stable
patterns reflecting control by the newschedule are likely to evolve following the transisiton
phase. Hence, during thetransition phase the discriminative properties of the newschedule do
notyetcome to control response patterns and the response patterns during thisphase may
initially reflect a predominance of thattopography which hasbeen historically reinforced. Later in
thetransition phase thepatterns may come to approximate more andmore those patterns
ultimately selected in the future steady state performance. During some intervening point,
however, thebehavioral patterns occuring willbeunderminimal schedule control, that is
performance breakdown or inconsistencies during the transition phase occuras a result of the
inconsistencies in stimulus control inherent in a schedule change. Ferster andSkinner (1957)
describe indetail anexperiment examining performance during thetransition phase when a
schedule change from a VR360to an FR360 is undertaken with pigeons. Prior to and for a
52
period following thechange, typical stable high rate VR performance occurs. However, toward
the endof theexperimental session performance decreases to thepointof taking 45 minutes to
complete one360 ratio run. Several pauses of 5-15 minutes occur during this runanda long
pause occurs nearthestartof the next fixed ratio segment. Overthe nextfivesessions response
sequences acquire thetypical break and run FRpattern characteristic of animal performance.
It is suggested thatthe performance decreases andinconsistencies referred to in the
present experiment as ratio strain are essentially similar to those instabilities of performance
witnessed during tranistion phases. Ineach case performance previously understrong schedule
control is disrupted by a change in thecontrolling schedule. Ratio strain is invoked asoccurring
when performance does notcome under subsequent "strong" schedule control. When control
by a subsequent schedule is reestablished thedisruption is referred to as transitory. Ratio strain
is in this circumstance a schedule or stimulus control phenomena influenced by procedural
arrangements enhancing schedule control.
When schedule control deteriorates and particularly when extremely high ratio requirements
aredemanded, a functional extinction of sorts maybeobserved. Thisoccurred in the present
experiment iii insiances of ratio strain with subjects PWand HNfollowing the large schedule
increases from baseline to FR2000 & 4000, respectively. In each of thesecases substantial
decreases in responding occurred during the sessions Immediately following the large FR
increase. These decreases Inperformance were Incompatible withcompleting the new
significantly larger schedule requirements. Only oneschedule completion wasfulfilled by each
subject in the firstexperimental session following theschedule increase. Thereafter response
rates of each PWandHNwere so lowthatnofurther reinforcers were obtained overthenext
several sessions even though responding wasto some extent maintained overthose sessions.
It is unlikely that thispersistence would be maintained indefinitely, for without reinforcement
performance is likely to cease or decrease to minimal occurrence asunderextinction. Therefore a
functional extinction maybe observed with strained ratio performance as in the present case.
Thissituation is referred to asa type of functional extinction in that formal requirements of
extinction arenotmet. Extinction refers to a procedural arrangement whereby reinforcement of a
previously reinforced operant is discontinued. The usual effect of extinction is to decrease the
frequency of performance. A critical feature in theuseof thetermextinction andonenotmet in
theexamples of ratio strain cited is thatuse is specifically limited to theprocedure of discontinuing
reinforcement (Ferster, Culbertson & Boren, 1975). In the current casethecontingency
requirements remain intact, albeit increased, and reinforcement is available as evidenced byeach
53
of thesubjects initial schedule completion, hence formally this is nota case of extinction.
However, the response patterns emitted by PWand HNwere behaviorally far removed from the
contingency requirements anda situation hadcome to exist where practically itwasas if the
contingency hadbeen broken.
It is important to useextinction as a description of the procedure of discontinuing
reinforcement ratherthanas a description of achange in frequency ofperformance since on
occasion frequency mayactually increase upon extinction. Forexample, behavior is frequently
observed to markedly increase upon the initial discontinuation of reinforcement (Reynolds.
1975). In another example behavior may be occurring at a lowrate when reinforced under a
schedule of Differential Reinforcement of Lowrates or DRL. In thissituation the behavior may
actually increase under extinction returning to a baseline rate actually higher than thatoccurring
when reinforced underthe DRL (Ferster & Skinner, 1957). Ferster, et al. (1975) point outthat if
wedefined extinction otherthanprocedurally in these cases wewould be, "inthe unfortunate
position of saying, 'he performance wasextinguished but it did notextinguish' ". Thus, froma
formal perspective extinction describes a procedural condition which is notcurrently met.
The formal conditions of extinction weremetandperformance patterns generated during
extinction were examined in the current axperiment in thecase of subject IN. Sustained high
rate performance by IN on FR8000wasobserved underconditions of verylittlereinforcement.
Oneobjective of examining performance underextinction wasto evaluate whether or notit was
the response-reinforcer relationship maintaining thishighrate behavior. IN wasexposed to 24
sessions of extinction priorto a mutually agreed upontermination point. In general responding
during extinction waspersistent although a cleardownward trend in response ratewasobserved
overthe course of the extinction period, suggesting an extinction curve in progress. Itmight be
noted that. in thecontext of thepreceding discussion. an initial increase in responding byIN
occurred during the initial session of extinction.
Although responding byIN was prolonged throughout the 24-session extinction period,
evidence suggests thatthis course of responding underextinction is commensurate with
extinction curves generated in operant research with infrahuman subjects following a similar
previous history of intermittent ratio reinforcement. Forexample, Hutchinson, Azrin and Hunt
(1968) report sustained responding bysquirrel monkeys overa 41-day period of extinction
during which 4·hourdailysessions were run. By Day 41 of extinction, the rate of barpressing was
significantly decreased. Notably, an approximate decrease of 50%wasobserved by Day 20. This
is somewhat similarto thepattern sean inthe case of IN in thepresent experiment and onemight
54
speculate that a best fit line through theextinction datagenerated suggests an extinction curve
which might becomplete in ananalogous 40 or50 sessions. Lindsley (1960) likewise reports
thatextensive histories of intermittent reinforcement of human operant responding lead to
extensive patterns of responding during extinction. Inthecase reported anextinction period of
approximately 100sessions wasnecessary to effect a nearcomplete cessation of responding
following 260sessions of variable interval reinforcement.
Thepatterns observed during extinction bysubject IN suggest that indeed previous high
rate and high requirement patterns of responding under the FRswere maintained by
contingency requirements andthe reinforcement of suchpatterns. Anessential difference in the
patterns exhibited by IN during extinction andthose extinction-like patterns generated by PW
and HNfollowing large schedule increases liesin theabruptness of frequency decrement.
Responding substantially andimmediately decreased underconditions of strain whereas
frequency of responding decreased more gradually underconditions of extinction. However,
these differences maybe related to schedule history differences in thatIN had experienced 24
sessions of ratio reinforcement priorto extinction whereas subjects PW andHN experienced 8 &
12sessions, respectively, priorto the large schedule increases.
A number of interesting performance pattems and relations areobserved when examining
timeout responding in thecurrent research. Asmaybe recalled previous research suggests that
timeoutsare reinforcing under conditions of increasing FRrequirements in thatthey increase as
a function of these increasing requirements (Azrin, 1961; Appel, 1963; Thompson, 1964, 1965).
Thesame authors alsoreport thattime out responding is mostlikely to ooeur during the PRP or
early inthe response run. Forthe most part these results arenotreplicated in thecurrent
research. Timeout responding wasseen to occurpredominantly during conditions of schedule
instability following schedule increases andpartirolarly duringmain keyperformance deterioration
following large schedule increases and during extinction. These results agree in large with
pattems of timeout responding reported by Zimmerman and Ferster (1964).
As reported, Zimmerman and Ferster did notobserve an increasing overall incidence of time
outasthe main keyFRwasincreased, rather theywitnessed an increased incidence of time out
during thefirstorsecond session following a schedule increase. This change wastransitory and
wasreported to decrease by about thefourth session following a change. Furthermore, the
frequency of time out responding wasseen to decrease markedly following a schedule change to
a smaller rather thanlarger FA. TIme outs were observed to occur during thePRP, however,
these weregenerally verybriefandresponses to thetimeout keytended to occurin even
55
numbered groups of 2 or4 responses, etc. These response bursts onthetime out key resulted
in time inconditions being reintroduced immediately or shortly afteranoddnumbered group of
responses. Thus, theduration of timespent engaging intimeoutwas generally minimal. These
results and others obtained lead the authors to propose that time out responding occurred as a
relatively weak alternative behavior becoming predominant undercertain conditions of instability
in the referent main keycontingency rather than as escape from reinforcement.
Similar patterns were observed tooccur inthecurrent research. For example, duration of
timeoutwasgenerally negligible with responses usually occurring in even numbered bursts, as
reported byZimmerman and Ferster, quickly returning conditions to time in. Additionally, time
outs were notobserved to increase systematically with increases in theratio requirements when
ratios were raised ina fashion notdisrupting main keyperformance, as inthecases of AR andIN
andduring the lattergradually increasing ..egimen of fixed ratios programmed for PW. However,
timeout responding did increase with increasing fixed ratios when those increases were large,
concommitantly producing strained main key performance, as in thecase of HNandduring the
initial schedule increase of PW. Thus timeouts seemed to be related more to conditions of
instability than increasing FRs, perse.
Time outs were often observed to occur during the post reinforcement pause as reported in
previous studies. Forexample uncler conditions of response maintenance, cumulative records of
subject HNshowtime outs occurring exclusively duiir.gbrief post reinforcement pauses.
However, timeoutsoccurred at times other than during the PRP aswell. Cumulative records from
subject IN showtime outs, albeit brief, occurring wellwithin ratio runs forexample. Whereas
subject HN tended to emit timeouts exclusively during the PRP, theothers showed no
consistent pattems of timeoutdistribution with time outsoccurring both Inandout of the PRP.
Generally, although instances of PRP related time out responding were observed, these results
were notconsistent enough to support theclaim of timeoutexclusivity inthePRP or early ratio
run. The reader is invited to examine location of timeout responding asportrayed in the
cumulative records for each subject located inAppendix I.
Time out responding wasseen to substantially increase during the initial stage of extinction
andthen latterduring extinction under conditions of variability in the main keyperformance
patterns in thecaseof subject IN. A large initial incidence of timeout responding during
extlnction has likewise been reported by Zimmerman and Ferster (1964), although these authors
report a decline in prevalence with thecessation of main keyperformance. Unfortunately, the
relationship of timeout responding during thefinal stages of extinction could notbe assessed in
56
the current study. The results reported herein andthesimilarity of thecurrent results to those
previously reported by Zimmerman and Ferster argue strongly against the theory proposed by
Azrin (1961) andothers of timeout as reinforcing under increasing fixed ratios. Infactevidence
fromthe current research suggests that thetimeout contingency actually serves to restrict
responding acting as a punishment procedure.
Theevidence for this is twofold. It may be recalled thata verbal introduction of the timeout
contingency was introduced afterthe initial training session(s) for all subjects andduring the initial
training session(s) thetime outkeywasfreely availabla and responses to this keywere recorded
although no timeout contingency was in place. A substantial decrease in the incidence of
responding to the timeout key occurred for all subjects afterthis verbal introduction of the time
out contingency. Inthe case of subject AR, for example, asseenin Figure 7, thetimeout
contingency wasactually introduced priorto the seventh overall session afterwhich minimal time
out responding occurred, whilepriorto this introduction responding wasactive to that response
key. The timeoutcontingency was introduced priorto session two forsubjects PWandIN
duringwhich 0 & 2 timeoutsoccurred, respectively, while 33 & 1124 timeout responses were
recorded duringtheprevious session, respectively. The average number of responses to the
time out key engaged in by subject HNduring the five sessions priorto the introduction of the
time outwas 2550 whileand average of 29timeout responses occurred during the fivesessions
following the introduction of the timeoutcontingency. Therefore onelineof evidence
suggesting that thetimeoutcontingency actually restricts responding derives from analysis of
the pre-post contingency response rates.
During the initial training sessions response frequencies to the nocontingency timeout key
andto the extrakey were approximately similar. Thisis notparticularly surprising given thatthese
two keyswereessentially equivalent during thispartof theexperiment in having noprogrammed
consequences (except for when the extrakeyserved as the key for a consummatory response
following completion of the schedule requirements). In comparing the frequency of responding
to thesetwo keysafterthe introduction of the timeoutcontingency, it can be seen thata much
higher frequency of responding occurs to theextrakeyfor all subjects. It is proposed that this
difference is primarily dueto a decrease in response frequency to thetimeout keyresuUing from
the imposition of thetimeout contingency. Thus a second lineof evidence suggesting thatthe
timeoutservesasa punishment of responding to thatkeycomes fromcomparing theoverall
response frequencies of theextraandtimeoutkeys. When examining the response patterns to
these keysall foursubjects in this experiment showed a much higher frequency of responding to
57
the extra key. Additionally, increases in timeout responding werevery likelyto be paralleled by a
symetrical increase in extrakey response patterns with substantially largerfrequency increases
occurringto the no contingency extrakey. Hence again evidence suggeststhat the timeout
contingency may actually serve to restrict responding.
A presumption is madein this analysis that the response rates to thesetwo keys mightbe
similarif no time out contingency waspresent. Caution in this premise is warranted in that an
increased response frequency to the extrakey mightariseas a spuriouscorrelation or response
generalization which results from the consummatory response requirement intermittently
programmed on that samekey. However, several factorsargueto the legitimacy of the premise.
First, a distinctarrayof stimulus lights served as theoccasion for theconsummatory response
potentially enhancing discrimination. Secondly, the behavioral patternsobserved to the
extra/consummatory keyduring the absence andpresence of the stimulus lightssuggestthat
two separateclassesof responding coexisted. In the presence of the lightsa single response
often occurred to this key whereas duringthe absence of the stimulus lightsresponse bursts and
extended runswere observed. Visual observation confirmed distinct response patterns. In the
case of a schedule completion subjects often relaxed, sittingback in the chair upon illumination of
the lights then reaching out and casually "slapping" the consummatory response key thereby
incrementing the pointscounter. In the absence of such lightsthe subjectsdisplayedintensity of
responding and behavioral patterns similarto thoseemittedduring main key response runs.
Giventhese indications of stimulus controlby the consummatory occasion and conversely by the
absence of such occasion, the premise of functional equivalence of the extrakey and the
pre-contingency time out key is strengthened. A secondpointof caution in accepting this
premise lies in the physical arrangement of the experimental apparatus. As can be seen in Figure
2 and as discussed in the apparatus section the centerextra/consummatory keywas physically
closerto the subject, perhapsfacilitating responding.
Someevidence existsin the timeout literature as well, suggesting that the actualtime out
contingency servesto limit responding to a second key. Appel(1963) in examining the
increasing incidence of time out responding under increasing fixed ratios reports that a similar
relationship is seen whenstimuluschange aloneresults as a consequence of the response to a
secondkey. In fact, Appel reports the duration of time spent engaging in theconditionof
stimuluschange but no time out to increase as an exponential function of the FR size whereas
the durationof time spentengaging in timeout was alsoseento increase with increasing FRsbut
at a less acclerated rate. This resulted in a substantially greaterdurationof timespentunderthe
58
stimulus change condition relative to thetimeoutcondition at the highest FRtested. This
interpretation of the literature andevidence fromthe current research thensuggests that
although as initially formulated the mechanism of timeoutperformance wasoneof reinforcement,
in actuality it seems thatthe phenomena andrelatcns involved may be considerably different and
at least from oneperspective the effect of a timeoutcontingency maybe perhaps more akin to
punishment. This perspective of theeffects of timeout is more in linewiththetraditional position
(Ferster, 1957) on which clinical practice is based (SUlzer-Azaroff & Mayer. 1977).
Response patterns to boththe time out key and to theextra key seem to predictably occur
under conditions of schedule variability in the main keycontingency. This variability is most
obvious under those conditions producing strain andduring extinction. However. increased
responding is likely to be observed to the extra and timeoutkey following a schedule change as
well and. as mentioned. a higher frequency of responding is seen underthese conditions to the
extra rather than to thetime outkey. This is particularly apparent in thecases of subjects IN and
PWwhere response increases areseen in the firstsessions following a change, subsiding bythe
lattersessions. Increases in response patterns to these two keys areoften observed in the
second or thirdsession following a scheduIechange for subject ARaswell. Subject HN's
idiosyncratic pattern of responding simultaneously to the main keyandextra keyprevents a
similar analysis of responding.
In summary. it seems likely thattheresponse pattems to thetimeoutkeyin this experiment
occur as a result of variability in performance under thecontrol of the main keyschedule of
reinforcement and those conditions producing schedule variability. rather thanasan escape from
aversive conditions. and that the frequency of time out responding underthese conditons is less
thanto a second no contingency extra key, leading oneto suspect that thetimeout contingency
in thissituation mayitself be procedurally classified aspunishing Inthat it restricts response
patterns to the timeoutkeyunderthese conditions of contingency Induced variability.
Inan experimentally arranged situation. a behavior isdefined as operant when it is shown to
beunderthecontrol of consequences given a particular occasion. The relationship of the
consequences to the behavior. or theschedule of reinforcement, is critical in controlling and
predicting or otherwise "governing" theobserved patterns of behavior. Periodically, other
behaviors may beobserved to be systematically related to a referent contingency. yet notunder
operant control by thatcontingency in thesense of being defined by consequences. A behavior
is saidto be "adjunctive" (i.e., to the schedule) or "schedule-Induced" when it varies
systematically or it'spresence or absence is systematically related to a particular schedule of
59
reinforcement and when it is notspecifically under theconsequential control of thatschedule.
Patterns of responding to the time outandextra keys in thecurrent experiment share certain
commonalities wnh such schedule-induced behavior. Adjunctive or induced behavior maynot
necessarily havedefined or apparent consequences andyetmaybe observed to occur
systematically or predictably asa result of contingency arrangements focussing on not
necessarily related referent pattems of behavior. Thus, "schedule-induced" behaviors can be
distinguished from"schedule-governed" or reinforced operant behaviors.
Thefieldof schedule-induced behavior wasbrought to attention by theworkof Falk (1961)
who reported polydipsia or increased waterconsumption as a systematic effect occurring under
specific conditions of traditional lever press-food reinforcement schedules. Under certain fixed
interval (FI) condnions ratswere observed to drink excessively, up to one-half of theirbody weight
inwater, overthecourse of an experimental session. Since this initial research, a variety of
schedule-induced behaviors, many of clinical relevance, have been reported in animals including,
but not limited to; alchohol consumption (Falk, Sampson, & Winger, 1972); defecation (Rayfield,
Segal &Goldiamond, 1982); wheel RJnning (King, 1974); air licking (Mendleson &Chilling,
1970); aggression (Yobum & Cohen, 1979); self-administration of various dRJgs (Oei, Singer,
Jefferys, Lang & Latiff, 1978); wood chewing (Roper, 1978); andpica(Villareal, Note 2). In
experimental snuations a variety of adjunctive or schedule-Induced behaviors have been
reported withhumans aswell,for example; pacing anddrinking (Kachanoff, Leveille, Mclelland &
Wayne, 1973); eating, drinking and grooming (Fallon, Allen &Butler, 1979); smoking (Wallace &
Singer, 1976), aggression (Frederickson & Peterson, 1974); locomotor activity (Muller, Crow&
Cheney, 1979); andvocalization (Porter, Brown & Goldsmnh, 1982).
In general theobserved condUions likely to induce adjunctive patterns areessentially similar
to those initially described by Falk (1961) andinclude exposure to temporal schedules of
reinforcement withgenerally constant interreinforcement intervals, hence, fixed interval or fixed
time (whereby reinforcement is delivered at a fixed timeregardless of anyrelationship to behavior)
schedules. Level of deprivation with regard to a referent contingency often influences the
degree of schedule induction asmaythe schedule parameter Ineffect. With regard to polydipsia
a bltonc function of schedule-induction hasbeen reported onsimple FIschedules wnh maximum
consumption occurring at FI values of 32, 64, 128seconds while lowerlevels of consumption
occurat larger and smaller schedule values (Falk, 1964). Behaviors have been reported to be
induced underotherconditions aswell. A number of articles reviewing thisarea andinherent
research intricacies areavailable (Falk, 1981; Roper, 1981).
60
A second general class of contingency induced behaviors might be deemed "contingency
change" induced behaviors. Inthissituation behaviors adjunctive to a given referent
contingency have been reported underconditions of change in thatcontingency. The
contingencies involved in these relations arenotnecessarily temporal. Forexample, Hutchinson,
Azrin andHunt (1968) report increases in biting responses by monkeys upon the introduction of
high FRrequirements. Gentry (1968) likewise hasreported attack andaggression byonepigeon
upon another as an induced pattern uponintroduction of ratio schedules. Ingeneral, the
changes involved producing adjunctive pattems are in thedirection of increased response
requirement or decreased reinforcement density. At oneextreme, aggression andattack have
been reported by both humans (Kelly & Hake, 1980) andinfrahumans (Azrin, Hutchinson & Hake,
1966) under conditons of extinction in a referent contingency.
Thegenerating conditions of extra andtime out key response patterns andthese response
patterns in thepresent experiment arecommon withobserved contingency induced patterns
reported in the experimental literature. Hutchinson, Azrin andHunt(1968) examined the bar
pressing performance of squirrel monkeys under a variety of ratio schedules including CRF, both
gradually andabruptiy increasing fixed ratios, FR reversals to lowerrequirements, andunder
extinction. In addition, the authors examined the occurrence of a second concurrent response
which had noprogrammed consequence. Thissecond response consisted of bitinga rubber
hose. Thissecond biting response wasobserved to occurat an increased frequency undera
number of experimental conditions. When extinction wasprogrammed following barpress
responses under lowratios, biting increased substantially from a zerobaseline levelceasing
immediately uponthe reintroduction of reinforcement.
Biting was also seen to occurupon increases in thefixed ratio requirements, withbiting
initially occurring throughout the main response interreinforcement interval but latterstabilizing
andlocalizing in thepostreinforcement pause or early response run. Theoverall frequency of
biting wasobserved to increase with increases in the ratio values anddecrease upon
reintroduction of lower values, although these changes were notnecessarily spontaneous.
Finally afteran extended history of FRperformance extinction was undertaken withonemonkey
overa 41 dayperiod during which biting occurred at an initial high frequency decreasing gradually
overthecourse of the extinction period yet remaining pervasive throughout.
It is suggested thatpatterns of timeout responding asseen in thecurrent research andfor
that matter patterns of extra keyresponding areactually induced patterns brought about by
degrees of contingency change. These behaviors aresimilar to timeoutpatterns previously
61
observed by Zimmerman and Ferster (1964) in thattheyappear to be relatively weak (in thesense
of frequency) alternative performances increasing in frequency underconditions of schedule
instability, which may be brought about by contingency change. To thedegree thatthe patterns
of responding currently observed aresimilar to, or arecommensurate with, those reported in the
contingency induced literature oneagain questions theoretical relevance of historical explanatory
mechanisms suggested by the timeout frompositive reinforcement literature.
Time out responding has, in fact, beenreported to occurexplicitly as an adjunctive activity
undertypical fixed interval inducing conditions by Brown andFlory (1972). They report that, as
withotherFI induced behaviors, timeout responding occurs as a bitonic function initially
increasing andthendecreasing under increasing fixed interval schedules andclearpatterns of
thiscan beseenin twoof five pigeons exposed to FIschedules ranging from FI30 to FI960sec.
A number of procedural differences prohibit a direct comparison of these pattems withthe
previous patterns of timeout responding seen underratio schedules or with thecurrent
research, however the authors conclude thatpreviously reported increasing functions of FR
response requirements andthe amount of escape (or timeout responding) was notobserved
andthattheir obtained results lend support to thenotions proposed by Falkof a class of
adjunctive or schedule-induced behaviors. One of the notable procedural differences in the
Brown and Flory analysis of timeout responding concerns the timeoutcontingency in effect
during a response to the escape or timeoutkey. Under theconditions reported a distinct
stimulus change waspresent yet thetimerof the referent temporal based contingency continued
to time. Hence, thisstudy might properly report schedule-induced stimulus change responding.
A particularly striking result arising fromthepresent research concerns the fixed ratio values
shown to sustain human schedule performance. To some extentit is notsurprising to see
response maintenance underlarge fixed ratios giventhe current experimental arrangements of
minimal response effort andthe implicit experimental demand characteristic of a small room anda
lonely telegraph key. What is surprising is thatFR values of the magnitude examined herein have
notbeenreported in previous studies of human FRperformance. Forexample, analogous
research utilizing a similar button pressing response with human responders typically has
examined performance underthecontrol of schedules withmaximum ratio values of FR 10
(Weiner, 1970a, 1979, 1981a), FR 20 (Azrin, 1959), FR 30 (Frazier & Biletto, 1969), FR40
(Weiner, 1964b, 1967, 19n, 1981b, 1982), FR50 (Long, 1963;Weiner, 1964a), FR 60 (Repp
& Dietz, 1975), FR 80 (Wasik, 1969; Weiner, 1972)•.FR200(Holland, 1958), FR250 (Weiner,
1966), FR 300 (Poppen, 1982), andFR 1000 (Sanders, 1969). This is in contrast to the current
62
research utilizing an FR 250 baseline with increasing schedule regimens to FA4000 and FR
8000.
Human operant performance hasalso been examined under thecontrol ofplunger-pulling
schedules as large as FR20 (Lindlsey, 1S59), FR 120 (Duvinsky & Poppen, 1982), FR150
(Long, 1958), andFA300 (Hutchinson & Azrin, 1961). Additionally, Witters andBachrach (1965)
report aschedule of FR200 where a verbal response wasutilized astheoperant. As can be seen
typical FRresearch with humans heretofore hasbeen conducted atverylowratio values,
comparatively speaking. The large FRvalues studied in thecurrent research to some extent
parallel in magnitude those high ratios shown to sustain performance by pigeons (FR2000;
Reynolds, 1975) andmonkeys (FR 4000; Findley & Brady, 1965) and it is perhaps this relative
concurrence which hasproduced thedistinct humanlinfrahuman similarities noted in
performance with regard to time outand extra keypatterns.
Sanders (1969) in perhaps thestudy most closely related to thepresent interms of ratio
values examined, observed the button pressing performance of twohuman subjects over20
two-hour sessions. Theauthor specifically wasexamining concurrent fixed ratio andfixed interval
performance and reported sustained ratio responding at FRvalues of 50,100,150,300,500 and
1000. The largest ratio of FA 1000 wasreported to be in effect for 1.5 hours, and no strain or
postreinforcement pausing wasobserved at this value or overthecourse of the experiment.
Post reinforcement pausing is not reported to occurby human responders in those studies
examining performance at the large ratio values heretofore examined (Holland, 1958; Hutchinson
& Azrin, 1961; Poppen, 1982; Sanders, 1969; Weiner, 1966) norwas significant post
reinforcement pausing observed in the current study, although briefPRPs were observed in the
case of HN. This is incontrast to those aforementioned across-species similarities in thearea of
collateral response patterns. Analogous results lead Weiner (1966) to suggest that the
characteristic response patterns observed in infrahuman schedule research maynot necessarily
occurwithhumans.
Indeed, recent research andtheory in human operant behavior hassuggested thatmany of
the typical performance patterns observed in infrahuman (hereafter also called animal) research
maynotbe readily apparent when human responders areutilized (Matthews, Shimoff, Catania &
Sagvolden, 19n; Lowe, 1979; Harzem, 1984; Saunders, 1969). Some authors, notably Weiner
(1983), have argued thatthese results are likely a function of uncontrolled experimental variables.
histories andprocedures, while others have implicated theoccurrence of verbal behavior,
instructions, and self-instruction ascritical features of experimental differences (Lowe, 1979).
63
Much of thiscontroversy hasevolved from a series of research articles studying human
perfonnance under the control of fixedinterval schedules of reinforcement where the PRP has
remained elusive aswell. Although thetypical scalloped pattern of FI responding inanimal
research hassometimes been observed by human responders (Buskit, Miller, & Bennet, 1980)
and hasbeenproduced when nonexistent through experimental arrangements (Azrin, 1959;
Weiner, 1972; Buskit, Miller, &Bennettt, 1981), otherpatterns including high undifferentiated
rates (Weiner, 1969, 1970), variable rates (Harzem, 1984), andlowrates of responding with only
oneor a fewresponses occurring close to reinforcement ( Matthews, Shimoff, Catania, &
Sagvolden, 1977) have been reported aswell, sometimes in different subjects of the same
experiment (Weiner, 1969; Buskit, at aI., 1980).
The importance of issues arising from thiscontroversy has, at extremes, called intoquestion
the verynature of the relevance of the experimental analysis of behavior to thehuman condition.
To the extent thatacross species replicability provides evidence forthe extrapolation of
information from animal to human behavior (Skinner, 1953), the issue seems gennane. Ona less
grandiose level, discrepancies areat least interesting, at times perplexing, and usually worthy of
further examination.
Thecurrent studyserves to emphasize that a good dealof basic research stillneeds to be
undertaken before strong opinions canbe voiced on some of these issues. Forexample, human
behavior had not really been examined underthe control of most of the schedules incorporated
inthisstudy. Itappears that more realistic appraisals ofperfonnance will result asa function of
further variation in experimental conditions. In general, the results of the present research show
more similarity than dissimilarity to traditional infrahuman behavioral patterns (Ferster & Skinner,
1957) and in thecurrent analysis of behavioral disruption relations arising from research in the
experimental analysis of animal behavior seem as helpful Wnot more helpful than those relations
derived fromtheexamination ofhuman operant behavior. Ideally thetwoareas arecompatible.
A clueto thediscrepancy in the case of the absent post reinforcement pause might be
gleaned fromthatsituation in thecurrent research where brief PRPs wereobserved, thatbeing in.pattems by subject HN. As may be recalled subject HNengaged in a superstitious two-handed
response pattem pressing the main response leverand the extra key in tandem withthe right and
left hands, respectively. It is speculated thatthe effort involved in maintaining this response style
isconsiderably greater thana one-handed pattern allowing a switching of the hand inoperation.
Hence it seems feasible that PRPs maybe indirectly related to response effort. The influences of
response effort on human operant perfonnance areaddressed in Experiment II.
64
EXPERIMENT II
Responding wasobserved to besustained at high rates following a series of gradual
schedule increases in Experiment I, evert at fixed ratio values of 4000, 8000, andto an extent
during extinction, whereas strain wasobserved to occurfollowing more abrupt schedule
increases. At these higherratios responding wassustained at values much largerthan
previously reported in the human operant IUerature. It wassuggested that perhaps the nature
of the taskandinparticular the response effort requirements may have been partially
responsible for thisoccurrence.
Evidence suggests that indeed the behavior selected for analysis may becrltlcal in
schedule performance andsubsequently strain. Forexample, as reported previously Staats, et
at. (1963) suggest that the response effort involved Ina complex verbal, motorand
discriminative reading response waspartially responsible for the development of strain or
response breakdown on the lowFixed Ratios of 4 and7. Otherreports of response
deterioration at lowschedule parameters when complex response sequences are required
exist aswell. Ferster (1960) reports strained ratio performance wUh pigeons at the low ratio
values of FR95 & 47when theoperant selected wasa match to sample (Ferster, Culbertson, &
Boren, 1975) response. Likewise, Thompson andMoershboecher (1980) report ratio strain at
FR20when a chained sequential taskwasrequired of pigeons. Analysis of strain and schedule
performance in terms of response effortrequirements mayshed light onsome of the observed
patterns in thecurrent research andon discrepent results reported inother animal and human
schedule reports.
A variety of human behaviors have been selected asbehaviors of choice underoperant
schedule analysis inclUding button pressing (Sanders, 1969), plunger pUlling (Linsley, 1956;
Hutchinson & Azrin, 1961), andreading (Staats, et al. Note 1),among others. The response
efforts involved in eachof these behaviors may indeed substantially affect subsequent
schedule performance. Forexample, Hutchinson andAzrln report ratio strain at FR300with
pronounced post reinforcement pausing andwUhin runbreaks when plunger pulling requiring
a 300gm. force to pullwas utilized asa response, whereas Saunders (1969) reports sustained
responding at FR1000 (with no PRPs) when a 40gm.button pressing response wasutilized.
A similar button pressing response wasused in Experiment I andunder some conditions, at
least, similar response persistence wasobserved.
65
The question arises asto the effect of response effort on performance andparticularly in
this case on response decrement or strain. Azrin (1958) reports thathumans engaging ina
button pressing response requiring approximately 15gms. of response effort showed
persistent high rate response patterns under fixed interval schedules of reinforcement. This
particular pattern isdiscrepant withthe typical scalloped FIpattern observed in most laboratory
animal research. The author reports a decrement in overall response rate andthe emergence
of scalloped FI responding when the response effort wasincreased to a several hundred gram
force requirement.
The evidence reviewed suggests that response effort requirements do indeed influence
performance patterns andmay be a precursor to strain, particularly at the large fixed ratio values
observed in Experiment I. This particular experiment examines performance disruption as a
function of increasing the response effort requirements involved in engaging the main
response keyandit is hypothesized that response effort increases will produce strained
performance. Specifically, performance isobserved under large FRs when, ina reversal
design, theresponse effort is increased from a 100gram effort requirement to a 500gram
requirement.
METHODS
SUbjects
Three subjects participated in this experiment. Two of these three subjects, ARand PW,
alsoparticipated in theprevious experiment and corresponding subject information can be
seenin thesubject identification section of Experiment I. A thirdsubject was recruited using
the recruitment procedures specified earlier.
LD Subject LDwasa 35-year-old Caucasion female currently working as a
secondary school teacher andtutor. Shehad completed the B.S. degree in
education andhad taken some graduate coursework in special education, as well.
LD participated in 58sessions overa sixweek period collecting 3020 bonus
points for a totalearning of $88.20 and wasactually paid$97.15 or thecurrent
minimum wage of $3.35 per hour.
66
Apparatus
Theexperimental room, apparatus, andcontrol equipment usedin this experiment are
exactly thesame asdescribed in Experiment I. During some parts of this experiment, however,
the force required to close thecontacts of themain keyresponse leverwasincreased from the
heretofore used 115gm.,across .15em., to 500 gm.across 4.5em. These two response
forces arereferred to as minimum and maximum effort, respectively. In thisarrangement both
gram effort andthrow space Oi ~01 c~sure in centimeter wasvaried in orderto produce a
maximum change in response requirement given the limitations of the current experimental
apparatus.
Procedure
Unless otherwise specified the procedures used in this experiment canbe assumed
similar to those reported in Experiment I.
ExperimentalDesign. Thegeneral experimental design of thisstudywasthat of an
ABAreversal. A baseline level of responding wasestablished at a particular fixed ratio
schedule andwiththe minimum keyforce in effect, thenthe keyforcewas abruptly increased
to the maximum level and sessions were rununtil response patterns wereagain stable.
Following thismanipulation there wasa reversal in the keyforcereturning to the baseline level,
untilstability criterion weremet.
Fortwo subjects, ARandPW, thepartiQJlar schedule thatwasheld constant across these
three conditions wasan FR4000. ForSUbject LDtheschedule was FR500during thefirst two
components andthiswasincreased to FR1000during thethird, reversal, component.
Furthermore, in anexperimental design arrangement thatwill be discussed in detailduring
Experiment III, there wasa 9- and12-session interim between the maximum keyforceand
minimum reversal components for subjects AR and LD,respectively. During this time,other
experimental arrangements were introduced and withdrawn for each SUbject. Forthe sake of
clearpresentation, thesemanipulations will notbe presented here.1 Ontheotherhand, the
experimental arrangements andsequencing occurduring consecutive sessions for subject
PW, perhaps representing the mostrigorous experimental design.
1. As notedthese experimental arrangements willbe explicitly addressed as the subjectofdiscussion in Experiment III. Briefly, in the caseof subjectAR, a concurrent alternative reinforcementcontingency was introduced on the extrakeyand this alternative was available duringfour sessions(32-35) afterwhichtime the alternative waswithdrawn duringthe ensuing five sessions (36-40). Likewise
67
History. Each of these three subjects entered thisparticular experiment withextensive
experimental history. The nature of thisexperiment does not require as meticulous
documentation of history as Experiment I, however for thetwo subjects, AR andPW, who
participated in thatexperiment, this history Iswelldocumented. Bycomparing the session
numbers located on the abscissa of the respective subject graphs in Figure 11 withthe
respective graphs andsession numbers of Figure 2 on page 30, the reader canspecifically
examine the experimental history leading upto this experiment. Training datawasreported for
these twosubjects in Experiment I, as well.
Subject LO participated in33 experimental sessions priorto thisstudy, of which 10of
these sessions were exclusively devoted to the training procedures culminating in the FR250
baseline. Eight sessions of FR1000were runfollowing four FR250baseline sessions.
Following this a reversal backto the FR250 occurred for foursessions andthenan extended
number of sessions (12) were runat FR500, the last5 ofwhichcompose the baseline for the
present study.
RESULTS
Theresults presented in thisexperiment will include those measures previously discussed
in Experiment I. These include response rate indices to each of the three response levers and
cumulative records. Please refer to Experiment I for specific information relating to calculation
of response rate or analysis of cumulative records.
Theeffects of increasing the main key response effort fromthe minimum to maximum
setting on the mainkey response patterns for thethreesubjects in this experiment canbe
observed in Figure 11. Thethree graphs fromtopto bottom referto datafor AR, PWand LO,
respectively. Theoverall responses persecond seen across conditions ranged from5-7 for
subject AR, 2-5for PW, and1-3for LOas noted on each respective ordinate.
for subject LD an alternative reinforcement contingency was introduced for threesessions (49-51) andthenwithdrawn for four sessions (52-55). Priorto the Introduction of the alternative for subject LD(sessions 43-48)the main keycontingency requirements were increased from FR500 to FR 1000although no effectof this increase wasobserved. In general for eachof thesesubjects, main keyresponse patterns decreased substantially withthe introduction of the alternative reinforcementcontingency to the extrakey and likewise extrakey responding increased substantially although theseeffects wereobserved to reverse withthe reversal of experimental arrangements. The nature of theexperimental design andthe reversal of experimental effectpermits the present analysis.
68
SUbject AR
Subject lD
Subject PW
25 26 27 28 29 30 31 41 42 43 44
:~ .. n~
n n -l-
I-
- Min Max Min'-
- I FA 4000 JI • I
~
n/ g g .....g.
~ a ':I II .co.
I-n~
~Min Max Min
- I II FA 4000 I
1
6
5
4
3
2
1o
5
4
3
2
1
o
2.5I-~I
~2.0 '- I~~1.5 ~
Min I Max Min1.0 -.5 -I FA 500
1FA 1000
.0 I
34 35 36 37 38 39 40 41 42 56 57 58
-0c:oo0)
enr-CDCo(I)
CD(J)
c:oCo(J)Q)a:
Sessions
Figure l l, The effects of increasing the response force on the main key from .115 gm. to 500 gm. on responding by all three subjects during Experiment II.
69
Ascanbe seen in Figure 11 and for each subject. increasing the response effort on the
main lever results in a decrease in the referent response pattern witha corresponding increase
in rate upon reversal. This decrease. however, is notparticularly large, given thebaseline
values. Forexample. subject AR is responding at the high rate of 5 responses per second
even during the maximum effort condition. although this is a decrease of approximately 1
response persecond from the baseline levels. Forsubjects AR andPWthe response rates
decrease below the baseline levels across all4 sessions during the maximum effortcondition.
while thisoccurs on three outof foursessions for LD. The reversal of thiseffect suggests that
it is indeed the experimental manipuiation that is responsible for thedecrease in rate.
In Experiment I a criterion for assessing strained performance wasdeveloped targeting
either a cleardownward trend in response rate during thetransition phase following an
experimental manipulation anellor a steady state performance at a response ratedecreased by
greater than20%froma previously established baseline level. In thecurrent experiment only
one instance of strained performance canbe observed utilizing thiscriterion, that in thecase of
subject PW. Theaverage main keyresponse rate of subject PWduring thepremanipulation
baseline was3.65responses persecond. This figure was derived as the mean of the four
individual sessions comprising thebaseline. Theaverage response rate during the ensuing
fourexperimental sessions was2.65 or otherwise a decrease of 27%in mean responses per
second. Mean responses per second wereobserved to increase to a level above the initial
baseline during thepostexperimental baseline averaging 4.4responses persecond.
Subject AR responded to the main key at anaverage rate of 6.4 responses persecond
during thepremanipulation baseline period of thecurrent experiment. whereas responding
decreased approximately 16%to anaverage of 5.4 responses persecond during the
experimental manipulation. Baseline responding increased to an average level of 5.8
responses persecond following reversal of theexperimental procedure. Average responding
for subject LDoverthese threeexperimental periods was2.2. 1.8& 2.3responses per
second. respectively. with rates falling off approximately 18% during the experimental
manipulation. Neither of these performances would be deemed strained according to the
established criteria of a 20%decrement.
70
Figure 12. Sample cumulative records portraying response patterns generated duringeachcondition of Experiment II by Subject AR. The upper and lowerrecords represent the preandpost manipulation baseline sessions, respectively, while thecenterrecord shows respondingduring the maximum response effort experimental manipulation. A complete collection of allcumulative records generated during thisexperiment is included in Appendix I.
PR 4000 Minimum response effortA
PR 4000 Maximum response effort
71
c
PR 4000 Minimum response effort
72
Figure 13. Sample cumulative records portraying response patterns generated during eachcondition of Experiment II by Subject PW. Theupper and lower records represent the pre andpostmanipulation baseline sessions, respectively, while thecenter record shows respondingduring the maximum response effort experimental manipulation. A complete collection of allcumulative records generated during this experiment is included in Appendix I.
PR 4000 Minimum reeponse effort
PR 4000 Maximum response effort
PR 4000 Minimum reepoDse effort
73
74
Cumulative Records
Cumulative records of these main key response patterns can be seen inFigures 12, 13,&
14. A description of the operation and interpretation of cumulative records isavailable in
Experiment I. The records of three representative sessions, onefrom each condflon, for
subject AR areportrayed in Figure 12. Theupperandlower records represent the pre and
post (experimental manipulation) baseline sessions, respectively, with the minimum keyeffort.
These records aresimilar to those produced by AR in thefirstexperiment andreflect a
consistent andhighresponse rate. At FR 4000, AR iscollecting two reinforcers a session. A
brieftimeout as indicated by a premature reset of the response marker canbeobserved at A in
the upperrecord anda loweryetconsistent response rate canbe temporarily observed at D in
the lower record. Otherwise, these records portray a very steady rateof responding.
Thecenterrecord of Figure 12shows responding during the maximum effort condhlon.
Theoverall record is generally similar to thebaseline sessions. Forthe most parta highand
steady rate can again beobserved withonebrief timeoutoccurring during the initial ratio
completion as evidenced by the premature reset. However, a couple of distinctive features
suggest the influence of the experimental manipulation. Two instances of rough grain or
temporary andabrupt rate changes occur at B andthese represent more variable rate changes
thanthat at D. Also, it is possible to observe theoverall decrease in response frequency by
comparing the interreinforcement intervals andatwhatpoint in the session thefinal reinforcer is
delivered, at C,with the baseline sessions. During the response effort condUion thedecrease
in rate is reflected by the laterdelivery of reinforcement within the session.
Cumulative records showing typical response patterns bysubject PWarepresented in
Figure 13. Aswith the record of AR, the upperandlower cumulative records showpatterns
generated during the pre andpostbaseline periods, respectively, whilethe center record
portrays responding during the maximum effort experimental condition. The experimental
manipulation is seento decrease the overall response rate, as indexed by the placement of the
reinforcer on the record within the session, andalsoproduces within ratio pausing. These
pauses areof relatively briefduration (5-10 sec.) andarespaced regularly through the session,
suggesting perhaps a counting function. A slight decrease in response rateoccurs
immediately following reinforcement butthis increases shortly therafter. The baseline records
showsteady high rate performance withnosubstantial postreinforcement pausing by PW. In
the lowerrecord two ratios arecompleted in contrast to the single reinforcer obtained in each of
the uppertwo records.
75
Figure 14. Sample cumulative records portraying response patterns generated during eachcondition of Experiment II by Subject LD. Theupper andlowerrecords represent the pre andpostmanipulation baseline sessions, respectively, while thecenter record shows respondingduring the maximum response effort experimental manipulation. A complete collection of allcumulative records generated during this experiment is included in Appendix I.
PR 500 Minimum response effort
iR 500 Maximum response effort
PR 1000 Minimum response effort
76
77
A lesssteadyrate of responding can be observed by subject LD in Figure 14. Theupper
record of thisfigureportrays pre manipulation baseline responding undera FR500. Ascan be
seen, a fairlysteadyrate of responding occurs, as notedearlierin the range of 2 to 2.5
responses per second, with a number of pauses duringthe ratios. For the mostpart these
pauses areshort, however, occasionally a longer pause occurs as at A. Several substantial
PRPs canbe observed andone brieftimeout occurs late in the session as indicated by the
premature resetof the response marker. A similar pattemoccursduringthe post manipulaton
baseline sessions underFR 1000shownin the lowerrecord. No time outsoccur during the
session shownandpausing is lesssubstantial.
Visual observation of the subject's behavior duringthesebaseline sessions provides
some information on the natureof the within ratio pausesby LD. Thissubjecttendedto
altemate pressing the mainresponse leverwith individual fingers of both hands. Forexample,
the subject might initially presswith the fifth digiton the righthandfor 75 responses andthen
switchto the fourthfor 100 responses, the third for 60 responses and so on, workingthrough
all fingers andthumbs on both hands. A briefpause wasthen likelyto occurwith eachswitch.
The centerrecord of Figure 14showsresponse patternsduringthe first sessionof the
experimental manipulation for LD. An increase in the number of time outsas evidenced by
cumulative record resets can be observed, however for the mostpart theseareof brief
duration. A timaout occurs following reinforcement at B, at whichtime mainkey responding
resumes during the timeout and continues untilC, when a second response to the timeout
lever returns the subject to time in. Main key responding thencontinues to the end of the
session. Response rate remains at a levelcommensurate with the baseline levelsduring this
session.
Time Out and Extra Key Infonnation
In addition to the mainkey response pattems portrayed earlierin Figure 11 and the
aforementioned cumulative records, Figures 15,16, & 17 showadditional response patterns
to the timeout and extra keyfor subjects AR,PWandLD, respectively. Subject AR, as shown
in the middle graphof Figure 15 responds minimally to the time out key overthe courseof this
experiment. The uppergraphof this Figure presents the previously discussed mainkey
response rate, whilethe lowershowsextra key responding. No effectsof the experimental
condition canbe observed with regard to the timeout or extrakey response patternsin this
experiment, although a higherfrequency of extra key rather thantime out responding by AR is
78
SUbject AR
7
654
3
21o
:~ n n n~
n-
r- Min Max Minr-,.. I FA 40DO I
II
I
25 2G 27 28 29 30 31 41 42 43 44
:-.QcII
3:laII..
l&. 2..,:l0 1II•.... 0~
25 26 27 28 29 30 31 41 42 43 44
wCDwe 300Qa.wII 200II:
:-.CD
100:IeII not applicable..~
0I&l 25 2G 27 28 29 30 31 41 42 43 44
Sessions
Figure 15. The effects of increasing the main key response efforton frequency of responding to the main key. time out and extra key.
SUbject PW
79
25 26 27 28 29 30 31 32 33 34 35 36 37
5
4
3
2
1
£I
~0c:GI:l 217•...
II...,:lQ
GIi
0....~
IIIII..e0 15Do..IIIII: 10~GI~
5•...~ 0lW
l-I'!- g g
D""'",g
I-~
I-~
~ Min Max Min1-' FR 4000 I
I I
I-
\I-
.. .. ..25 26 27 28 29 30 31 32 33 34 35 36 37
2S 26 27 28 29 30 31 32 33 34 35 36 37
Sessions
Figure 16. The effects of Increasing the main key response efforton frequency of responding to the main key. time out and extra key.
80
SUbject lD
5857
Min
FA 1000
4240 4139
o
36
Min
34 35
2.52.01.51.0
.5
.0 ~~~~~_~_~..:-~_~~~~~~_~_~.....
UI~III:lUICC:
ailia.QUI... 111~a:
34 3S 36 37 36 39 40 41 42I
56 57 58
5857
not applicable
563836 3734 3S
50
40
20
O~~~~~~~~~~~_~_~_~.......~_~_~ ..>'CIIII III~ ..
c:CIa100 a.~ ..x III1IJa:
ggI 0
III
I- _ I~I
..00
300
200
100
o34 3S 36 31 38 39 41 42 56 51 58
34 3S 36 37 38
800 ~
800
400
200o
IIII
~I39 40 42
I 0
56
o
57 58
Sessions
Figure 11. The effects of increasing the main key response effort on frequencyof responding to the main key. time out and extra key, as well as duration oftime out and resecnses to the main key during the time out.
81
observed. These results reflect a pattern of responding to each leversimilar to that portrayed
by AR in Experiment I andrepresents a rather high andsustained rate to the main response
leverwith little variation or systematic relation of performance to theothertwo response levers.
The response patterns to each of the three levers bysubject PW. on theotherhand, does
showan interrelationship to the experimental conditions as shown in Figure 16. Thedecrease
in response rate in the topgraph is conversely paralled by an increase in timeout responding
andextra keyresponding in the second andthird graphs, respectively, during themaximum
effort condition. albeit this increase represents an exceedingly minimal change in the case of
timeout responding. Thisoccurrence of twotimeout events insession 29 represents thefirst
occurrence of a timeoutby PWsince theoccasion of strain during the first experiment in
session 9. So even though thisdoes notreflect a substantial increase. it is interesting bysake
of the conditions underwhich it occurs. Thatis, subject PWwasonlyobserved to engage in 11
timeoutsoverthe entire course of Experiments I & II,or otherwise overa totalof 37 sessions,
the first of which occurred shortly afterthe introduction of the timeoutcontingency. Of the
remaining 10timeouts. eight were engaged in during theconditions of strain reported during
sessions 9 & 10of Experiment I with the remaining timeouts occurring presently in session 29.
In addition to thetimeouts, an increase in extra key responding overbaseline levels canbe
clearly seen in Figure 16 in two of thefoursessions undertheexperimental condition.
Responses by subject LDto each of thethree response levers during thecourse of this
experiment can beobserved in the topthree graphs of Figure 17. Nosystematic relation can
be seenwith regard to the frequency of timeout responding overthe experimental conditions,
although in looking at thefourth graph a substantial increase induration of timeout occurs
during the firstsession following the response effort manipulation. The lowergraph portrays
the frequency of main key responding during the initiated timeoutsandthesimilarity of this
graph to the graph of timeout duration suggests thatthe timeoutsengaged in were active
rather thanpassive. That is, the subject continued to respond even during the timeout(s). It
maybe recalled thatthispattem was observed inthecenter cumulative record of Figure 14. In
thiscase. although the experimental manipulation did notparticularly influence the frequency
of timeout responding or the general response rate to the main key, a temporary increase in
theduration of timeout andof otherpatterns engaged in during the timeoutcanbe observed
to occurduring thesession following the experimental change.
As seen in thecenter graph of Figure 17.a slight increase in the frequency of extrakey
responding overbaseline levels canbeobserved during twosessions, the first and fourth,
82
following the increase inresponse effort aswellalthough this is notsubstantial by anymeans.
Thefinal baseline of extra key response frequency has notbeen graphed as a further
experimental condition influencing theprobability of occurrence wasundertaken during
sessions 43-55. notshown. andas mentioned thiswill be discussed further in Experiment III.
As with theothertwo subjects of this experiment a higher frequency of responding occurs to
the extrarather thantimeoutkeyover all experimental conditions.
DISCUSSION
Experiment II examined the influence of response effort onfixedratio schedule
performance andconversely on performance deterioration or strain. Human operant
responding underthe control of large fixedratio schedules wasexamined underconditions of
minimum andmaximum response effort given the response effort constraints of the current
experimental situation. Increased response effort systematically decreased response rate for
each of three subjects andthese rates were observed to increase uponreversal of the
response effort requirements. Thereversal and pervasiveness of this effect across three
subjects each withdistinct experimental arrangements suggests a strong systematic replication
of the effect (Sidman 1960) however these performance decrements were notsubstantial.
approximately one response persecond. During conditions of maximum response effort
responding was observed to be sustained, in the case of ARat the rateof5 responses per
second. and generally around two responses per second in theothercases. In onlyoneof
thesecases was response deterioration substantial enough to meet the somewhat arbitrarily
established criteria of strained performance proposed in Experiment I. Hence. response
deterioration undertheconditions of increased response effort in Experiment II wasnot
commensurate withthedegree of strained performance observed following the large schedule
increases of Experiment I. Ontheotherhand. increased response effort systematically
disrupted previously established levels of responding as revealed by response rate analysis
andexamination of cumulative records.
It maybe thatthe degree of performance deterioration resulting from response effort
manipulations is related to the amount of response effort invoked during an increase. The
current experimental apparatus provided a limitation inthis regard. It maybe possible thatlarger
disruptions anddecreases in performance would result from similar experiments examining
83
largerchanges in effort. Forexample if a plunger pull(Lindsley, 1960) or, at an extreme,
swinging a baseball batwere utilized as anoperant it might beexpected thatstrainwould occur
earlier undersimilar ratio conditions particularly if increasing degrees of resistance orweights
were attached to the plunger or bat. However under the current conditions, this experiment
suggests that established human operant responding is quite persistent even under
conditions of increasing response effort, although response effort doesinfluence
performance.
Analysis of response patterns to the extra and time outkeys in Experiment IIsuggest a
continuity of effectandpatterns observed in the first experiment. Observation of the response
patterns to thesekeys in Experiment I suggested thatthe patterns were in largepart induced
underconditions of variability in the referent main keycontingency andthat responding was
more likely to be seen to the extra rather thanthetimeout key. The experimental manipulation
in the current experiment produced lessdisruption in mainkeyperformance than that reported
earlier and as a result onewould expect to see inthisexperiment a smaller degree of response
variability andcontingency induced patterns to theextra and timeout key, as is the case. By
andlarge noconsistent increases infrequency occur to either thetimeoutor extra key asa
result of the experimental manipulation for the twosubjects showing the leastdegreeof main
key experimental effect, thosebeing AR andLD. Ofthe responding observed to the ancillary
keys by these subjects, a greater frequency of responding was observed to the extra rather
thantime out key.
Thegreatest observed mainkeyexperimental effect andtheonlycase in the current
experiment of performance decrement fullfilling theutilized strain criterion occurred in thecase
of subject PW. Response patterns to theancillary keyswere conversely observed to Increase
overbaseline levels during the increased response effort experimental manipulation
suggesting a pattern of induced responding to thetimeout and extrakeys similar to that
reported in the first experiment. Again responding was more frequent to the extrakeyrather
thanthe timeout keyalthough frequency of performance to each was low, particularly when
compared to the persistent patterns generated to the mainkeyby thatcontingency.
Although response frequency patterns to the ancillary keys by subject LDshowno striking
experimental influence, a more detailed analysis of theseandotherperformance measures
suggest otherwise. As reported, nosystematic increases or decreases in response frequency
wereobserved to occurto the extra or timeoutkeyover thecourse of this experiment,
although on close examination it mightbe noted thatthe two highest frequencies of extra key
84
responding do occurduring the increased effort manipulation. Thiseffect is not large.
however. A second measure of timeout responding. thatof duration, suggests a systematic
relation to the introduction of the experimental condition aswell. Time out duration as revealed
in Figure 17 increases substantially above baseline levels during the initial session of
experimental change returning the nextsession to baseline levels. Figure 17alsoreports the
response frequency of a heretofore unreported behavior, thatof responding to the mainkey
during a timeout. Although response frequency of thisbehavior wasmonitored andrecorded
throughout this entire series of experiments it'spresence hasnotbeen reported otherthan in
this instance, asoccurrence wasgenerally negligible andnonsystematic. However. in the
present casean increase and decrement parallel to that reported for timeoutduration occurs
during the initial session of the increased response effort. Thisparallel pattern is explained in
thatthe cumulative record of thissession confirms the subject engaging in a sustained period
of main key responding during a lengthy time out. The significance of this point is in
suggesting thata variety of otherbehaviors. not necessarily even monitored, maybe
systematically related to changes iiime referent contingency relation. Heretofore, this pattern
of sustained responding to the main keyduring a time outby subject LDwas notreadily
apparent suggesting on a theoretical level thatenvironmental changes in oneparticular
contingency relation mayproduce fairlynovel yet related behavioral changes.
A second point of theoretical significance brought to lightby this experiment relates to the
present inducing conditions. In Experiment I behavioral patterns to the ancillary keys were
brought about by schedule changes andparticularly large schedule increases resulting in a
decrement of reinforcement density. In thecurrent experiment theschedule, perse,wasnot
manipulated. The inducing conditions consisted of a change in the response effort toward a
greater behavioral requirement while reinforcement density wasgenerally maintained although
sometimes slightly decreased. Thus, procedural change notnecessarily related to a direct
schedule change maybesufficient to produce changes in alternative behavioral patterns. The
significance of alternative behavioral patterns andtheirrelationship to ratio strain is the topic of
Experiment III. Generally, induced patterns seem to be more likely when procedural or
schedule changes result in lessfavorable environmental conditions, meaning an increase in
the response requirements or a decrease in the reinforcement density.
85
EXPERIMENT '"
Responding waspersistent in Experiment II at large ratio values and underconditions of
increased response effort. Fortwoexperimental subjects responding under FR 4000
contingencies onlyoneor two schedule completions occurred per30 minute session, hence
reinforcement density wasquitelean. It seems paradoxical that responding might be maintained
underthese conditions, in onecase at the rate of overfive responses persecond. It is possible
that response maintenance and conversely response cessation underthese conditions, given
theseresults, might bea function of the alternatives available. Thatis, responding may be more
likely to persist underunfavorable conditions, as in the current research, when the organism or
subject hasa restricted range of alternative behavioral opportunities inwhich to engage inand
which subsequently might be reinforced.
Thatthe availability of alternatives inwhich to engage inmaybe important in the analysis of
operant behavior andschedule performance has been suggested theoretically (Goldiarnond,
1975; Hursh, 1980) aswellasempirically (Catania, 1969; Holz, Azrin, & Ayllon, 1963). For
example, the breaking point of progressive ratio performance is influenced by the availability of an
alternative response. In thissituation responding will be maintained forgreater periods of time
when; a) there is no reset alternative available (Hodos & Trumble, 1967); b) when responding to
thealternative is punished (Dardano & Sauerbraun, 1966; Dardano, 1968), and;c) when the
response frequency requirements on thealternative are increased (Hodos & Trumble, 1967).
Thus clearly the availability of an alternative response andthecontingency requirements involved
when responding to thisalternative are important to the understanding of progressive ratio
schedule performance andsubsequent analysis of response cessation.
Similarconclusions aredrawn fromananalysis of punishment procedures on human
schedule performance. Holz, Azrin, andAyllon (1963) andHerman andAzrin (1964) examined
the effects of twodistinct punishment procedures on human responding maintained by interval
schedules of reinforcement. When an alternative operant response wasavailable which
procurred reiriforcement, mildpunishment of theinitial andprimary response completely
eliminated thatbehavior. However, responding waspersistent even under conditions of
punishment when no alternative response wasavailable, albeit at a lowered rate.
In thecurrent series of experiments SUbjects areseated in a small room fora fixed period of
timewith nofurniture otherthanthe experimental equipment, and little space. Under these
86
conditions it seems likely that responding may be maintained at highrates through the restriction
of opportunities to engage in otherbehaviors inherent in thissituation andthis restriction may
influence the occurrence of ratio strain. Experiment III examines me influences oi an available
alternative behavior onschedule performance andschedule disruption. It is specnically
postulated that strain, asused in the context of decreased frequency of performance, is more
likely to occur withtheadvent of an available alternative response, particularly underthe
unfavorable reinforcement conditions observed in Experiment II. Inthecurrent experiment the
alternative consists of thespecific introduction andremoval of a concurrent (Catania, 1969)
reinforcement contingency on a second existing leverwhich hasa higher reinforcement density
thanthatcontingency progranlrTled io ihe maln lever,
METHODS
SUbjects
Subjects LD and AR served asthe participants in thisexperiment. Subject AR also
participated in Experiments I & " andidentifying information maybefound inthe Subjects section
of the firstexperiment. Subject LDparticipated in Experiment II andidentifying information may
be found there.
Apparatus
Theexperimental room, control equipment andexperimental apparatus remained during this
experiment asdescribed in Experiment I, withoneexception. During the experimental
manipulation of this stUdy the SUbjects could engage in an alternative behavior to effectively
access reinforcement contingencies otherthanthe referent fixed ratio programmed on themain
response lever. Response performance onthe mainkeywasthenexamined as a function of this
relation to anavailable alternative. Thisalternative consisted of a concurrent highdensity
reinforcement schedule brought into effect during theexperimental manipulation on the
heretofore inconsequential center response lever, previously referred to as theextrakey. This
centerlever, then,served twofunctions. A conditional relation wasin effectwhereby when the
yellow reinforcement lights wereilluminated signalling the availability of points (see the Apparatus
section of Experiment I for details), thecenter lever served asa "consummatory" occasion
whereby a response incremented the bonus point total. When theyellowlightswere not
illuminated, andonlyduring the manipulation phase of this experiment, the centerkey served as
the occasion for the effective alternative pattern. During the baseline conditions of this
87
experiment, thiscenter key had nospecific programmed consequences andwashence, the
extrakey. The response effort in effect on the main response leverduringthis entire experiment
wasthatof the maximum setting or 500gm. The response effort onthe time out andcenterkeys
approximated the minimum effort or 115gm.
Procedure
Experimental Design and History. Theexperimental design usedin this experiment
wasthatof an ABA reversal, similar to thatused in Experiment II. The current design wasactually'
embedded in a larger ABCBA experimental design (Hersen & Bellack, 1981) andis represented
by the BCB components. It can be recalled that in Experiment II, the experimental sessions were
notconsecutive for AR andLDfollowing the response keyforce manipulation andpriorto the
final baseline. This interruption wasdueto thecurrent experimental investigation. Thatis, in the
larger context of the A1 B1C1 B2A2 design, Experiment II focussed on theA1B1A2 reversal while
the current investigation focusses on the embedded B1C1B2 reversal. As a result of this, andin
contrast to Experiment II, the current experiment consists of consecutive experimental sessions.
Additionally, the relevant experimental history of each LDandARcanbedetailed by examining
Experiment II andExperiments I & II, respectively.
Theexperimental manipulation in thisstudy involved introducing a second concurrent ratio
schedule on a second response leverwhich had a significantly higher density of reinforcement
than thefirst and examining the relation of thisavailable alternative to response patterns on the
main key. No instructions asto the presence or nature of this contingency weregiven. The fixed
ratio on the main response leverremained constant throughout thisexperiment at FR 1000for
subject LDand FR4000 for AR.
Following sixsessions of baseline responding withno alternative, an arithmetic Progressive
Ratio 3 (PR3) schedule of reinforcement (3,6,9,12, 15, ...etc.) was introduced on the center
response leverfor subject LD. The PR3 remained in effect for three sessions afterwhich the no
alternative baseline conditions were reestablished. The highest progressive ratio completed
during thesethree sessions byLDwas PR 177.
Subject AR responded underfoursessions of no alternative baseline, foursessions of
manipulation where an alternative waspresent andthen five sessions of baseline following the
withdrawal of the alternative. The alternative present during thefirstsession of the experimental
manipulation wasanarithmetic Progressive Ratio 3 which held constant oncethe increment
88
nun lfieySubject LO C-<J
Extra/Alt. Key>. -......Q1
~i::Jr--FA 1000Q1 c: 3.5 G:> 0 NO
"P4 (J 3.0 AItern8tive+J Q1 5Ill(/)
2.5c: \.0 4100 Ql 2.0Q1 C.+J 3~
'" 1.5-e Ql 2
'" 1.01:1 ce 0 .5 1
ID C.
c: '" .0 4 4 4 4 4 4 0"P4 III 4344 45 4647 48 495051525354 55ilia:1:
SUbject AR
FA 4000I I
~Witll .r-AlterMt1ve
I110 cl~ 110
Altt!rMtivlI
Alt1!rnlltive
.. I ,I~I4 4 4 .... II I
2829 30 31323334 353637 383Q 40
>. 20 80CIII
\:::lcr 15 6QI~
II.... 10 4:::l0QI 5 2EI- 0 0
I
I
1III
III~:v
d II I I2829 3031 323334 35 36 37 38:m 40
28 29 3031323334 35363738:m 40
I-01
.-~IIIIQ.-.Q....gI
D n g g I CI n g g g
II• 00
2.00
1.00
3.00
$11. 00
43« 454647 4849505152535455
~
~ a- Cl g g
.00
2.00
1.00
3.00
$ 4.00
CIt01C
C~
IIIWCo..CIt1IIen
Sessions Sessions
Figure 18. Frequency of responding to the main key. the extra/alternative key. andthe time out key by subjects LD & AA in all conditions of Experiment III. and totalearnings per session. The left hand column of graphs portrays data for subject LD.while data for subject AR can be seen in the graphs of right hand column.
89
reached 12(3, 6, 9, 12,12,12, etc.). For allpractical purposes thiswasan FR 12. Thealternative
available during the finalthree sessions of the experimental manipulation consisted of a
Geometric Ratio 2 witha terminal FR500 component (2,4, 8, 16,32,64,128,256,500,500,
500, etc.). Criteria for stability andstrain remained during thisexperiment asdefined previously
during the first experiment. Otherprocedures were asspecified earlier.
RESULTS
Theperformance moasures presented in discussing the results of this experiment include
cumulative records and graphs of response frequency to each of the threeresponse panel
levers. Additionally, graphed dataon monetary earnings persession will be presented asa
measure of reinforcement density in the baseline and experimental conditions.
Figure 18presents these graphed datafor bothsubjects, LDandAR. Data for subject LD
aregraphed inthe lefthand column, while dataforARareto the right. Thetopgraph in each
column represents mainkeyresponse rate for each subject, while also reflecting responses to
the alternative/extra key. Response rate to bothof these keysis graphed in responses per
second. The middle graphin each column portrays response frequency to the timeoutkey
during each condition while, finally, thelowergraph presents monetary earnings per session over
thecourse of this experiment for eachsubject.
The results of this experiment aresimilar for each subject. When presented with the
alternative, the heretofore persistent high rate main keyperformance quickly decreases, while
conversely, the previously lowrate and intermittent response patterns to theextrakey increase
substantially as a result of the attached accessible highdensity reinforcement contingency.
When thissecond contingency is removed a gradual yetsubstantial reversal of effect is observed
with an increase in the frequency of time outsalsoaccompanying thisdecrease in reinforcement
density. Thedecrease in main key response pattern as reflected in the top leftgraph of Figure 18
is immediate and, for all practical purposes, totalforsubject LD. Thisrepresents a complete
switch fromresponding on the main lever to responding on the alternative lever. This is in
contrast to the pattern evident in the top right graph. Subject ARcontinued to respond, although
at a lowerrate, on the main lever while substantially increasing response rates to the alternative
keyas well. During the experimental manipulation, in both of these cases, main key response
rate decreases infrequency to a pointsubstantially below the proposed 20%strain criterion.
90
Figure 19. Sample cumulative records portraying response patterns generated during eachcondition of Experiment III by Subject LD. Thecenter record shows the initial session ofresponding during which a concurrent alternative schedule of reinforcement was available,while upperandlower records represent thepreandpostmanipulation baseline sessions,respectively. A complete collection of all cumulative records generated during this experimentis included in Appendix I.
l!'R 1000 110 Alternative
PR 1000 with .u ternative
)~---------i-IB --.I.
91
1)),n,,11 "1,,",, n t ) , ) ) , ;
PR 1000 110 llternative
92
Visual observation reveals thatthispattem of reponding by ARwasessentially a procedural
artifact in thata superstitious pattern of responding was reinforced wherein the subject would
alternately press the mainkey and thenthe alternative key in high rate succession
(main-alt.-main-alt., etc.), andthispattem was reinforced witheach completion of the alternative
keyschedule requirements. If thetwo keyshadbeenarranged to be independent, either
physically in space or behaviorally through programming, it is suspected that the realized
response patterns would parallel those of subject LD. Theresults of this inefficient response
pattern is reflected in a loweroverall reinforcement density for ARwhen contrasted to LD,during
thefinal threesessions of the experimental manipulation in the lowergraph of Figure 18.
Cumulative Records
Sample cumulative records fromeach of the three conditions of this experiment for SUbject
LDcanbe seenin Figure 19. The pre and post(experimental manipulation) baseline patterns are
shown in the upperand lower records, respectively, andto a large extent theserecords
represent schedule performance by LDsimilar to that portrayed and discussed inExperiment II. A
number of briefpauses occurduring the ratio runs andbriefPRPs occuras well. An extended
pause occurs during the initial portion of the lowerrecord andan estimate quiteconsistent with
thegraphed datawouldbethatduring this pause, thesubject wasresponding to the extrakeyas
a result of the historical andfavorable reinforcement conditions. Although speculative it is likely
thatduring this pause the subject wasresponding to theextrakey(some post-manipulation
responding is evident in Figure 18), sortof "feeling out"potential environmental conditions and
returning finally to the triedandtruemain keyupon environmental acknowledgement of a lackof
contingency.
Thecentercumulative record of Figure 19shows the first session of the experimental
manipulation andthe immediate decrease in response rate to the main key is evident. Initially, at
A, a number of responses can beseen, however responding to this leverquickly stops. At B a
timeout is engaged in asshown by the reset. During thiscondition, only, the reinforcement
marker along thebottom of the record wasattached to thealternative keyandmarked everytime a
reinforcer waspresented in thecontext of the alternative contingency. As can be seen
reinforcement is presented consistently throughout the session, representing sustained
responding to the alternative key. The interreinforcement interval canbe observed to increase
steadily as would be expected under the progressive ratio schedule.
93
Figure 20. Sample cumulative records portraying response patterns generated during eachcondition of Experiment IIIby Subject AR. Theupperleft and lower right records represent thepre and postmanipulation baseline records, respectively. The upperright record shows thefirst of four sessions during which a concurrent alternative schedule of reinforcement wasavailable, whilethe lowerleft record shows thefinal session of when the alternative wasavailable. A complete collection of all cumulative records generated during this experiment isincluded in Appendix I.
..
..
94
....0
•. ~..~ .... •..
E•§ .... .... ;0 ..• •~ ~.. .... ..
95
Cumulative records from representative sessions for subject ARcan be seen in Figure 20.
Thefirst andfourth records represent the preandpostbaseline patterns of responding while the
second andthirdportray the firstandlastsessions of the experimental manipulation, respectively.
Responding underbothbaseline sessions canbeobserved to be of a high rateandconsistent
nature. Thispattern characterizes thefirsthalfof the initial session where the alternative was
present, as well.
Thedevelopment of response patterns to the alternative key in thesecond graph can be
observed by examining the reinforcement of responses to thealternative keyduring this
experimental condition. Following the initial schedule completion of the FR4000contingency
programmed on the main response lever, an initial alternative schedule iscompleted atA. Main
key responding remains in effect at the previous rate, however, untilC,when it deteriorates to a
much lower frequency, breakand run pattern. At B, andeven while the mainkeypattern is
strong, responding begins to occur more consistently to the the alternative key as reflected by
the reinforcement marks on the bottom of thecumulative record. This marks the beginning of the
aforementioned two-handed alternating pattern whichbecomes fully developed by D. The
period of C through D reflects a transition phase priorto the development of this two-handed
response pattern andthe period of the greatest strain with regard to themain key response
pattern.
The cumuiative record from the final session of the experimental manipulation forARshown
inthe thirdgraph of Figure 20shows a more or lessstable pattern of responding, at the rate
evident at D in theprevious record. A fewbrief timeoutsearly inthe session andseveral brief
pauses and/or intermittent rate changes areobservable during thissession, but for themostpart
the rate is consistent. The reinforcement marks on the bottom of the record suggest consistent
response patterns to the alternative key aswell. In addition to reinforcement available fromthe
alternative contingency. an instance of reinforcement fromfulfillment of the referent FR 4000
contingency can beobserved at A.
DISCUSSION
Experiment IIIexamined thepersistence of human operant performance as a function of
alternatives. Under conditions of high response requirement. lowreinforcement density,
increased response effort, andtimeout arrangements explicitly designed to minimize excess
96
responding, performance wassustained for two subjects when accessibility to a reinforced
alternative wasnotpresent. Mainkeyperformance wasobserved to immediately and pervasively
decline to levels defined asstrained upon the introduction of an accessible alternative
reinforcement contingency andthis effect wasshown to be reversible upon removal of the
alternative. Hence, response decrement of human subjects wassubstantially influenced by the
availability of a response alternative.
The leveland immediacy of performance decrement in this situation suggests that the
. provision of an alternative is an effective method in reduction of behavior. Main key ref,pondir.g
wasnearly completely eliminated in thecase ofonesubject andsubstantially reduced I.:'y th::l
second subject. Thatthissecond pattern wasnotcompletely reduced was attributable to a
procedural artifact. Thebehavioral changes observed anddefined asstrained in thisexperiment
areremarkable when compared to performance decreases in Experiments I & II in themagnitude
andimmediacy of effect andthe ensuing stability of decreased performance levels. Hence, ratio
strain when examined from these perspectives wasnotcommensurate withstrain observed in
earlier experiments, although aspects of commonality areapparent.
It wassuggested in Experiment I that ratio strain was if notprimarily a transition phenomenon
thena stimulus control phenomenon closely resembling patterns observed during transition
phases. Thegreatest degree of performance disruption observed in the current experiment
appeared likewise as a transition phenomena, thatbeing in the case of subject ARasrevealed
briefly in the first session following the altemative introduction aswitnessed by cumulative record
in Figure 20. However, bythe endof thissession performance hadstabalized at the rate and
pattern observed throughout the remaining manipulation period. Inothercommonalities the
magnitude of observed effect was most closely related to that seen in Experiment I while later
stability wasparallel to patterns of Experiment 11.
Response patterns to the time outkeywere clearly observed to be related to contingency
change conditions influencing reinforcement density andsimilarpatterns wereobserved for both
subjects in thecurrent experiment. Time out responding wasseen to increase when overall
experimental reinforcement density decreased. Thiseffectappeared transitory in nature
supporting previous suggestions of timeout response patterns under ratio schedules as being
induced in nature. In particular timeout responding increased substantially during thefirst two
sessions following the experimental reversal and thenreturned to baseline levels. Thisincrease
paralleled a substantial decrease in overall reinforcement density which hadbeen elevated bythe
experimental manipulation. TIme out response frequency waselevated above baseline levels
97
during the final three of four experimental sessions forsubject ARaswell, however this elevation
wasconsistent withthe current analysis in thata substantial decrease in reinforcement density
occurred fromthe initial experimental session wherein the subject earned approximately $11.00.
Increases in timeout responding did notoccurfollowing contingency change conditions
towards favorable experimental arrangements but rather onlytowards decreases inoverall
reinforcement density, in this particular experiment. When examining the present conditions
inducing response patterns to the ancillary timeoutkey, it is interesting to note that nochanges
occurred duringtheentire course of this experiment in the defined referent main key
contingency, although behavior patterns underthe control of thiscontingency were seento be
substantially effected. In this situation an analysis of the inducing context must necessarily
examine inter-contingency relations, for contingency induction waseffected by the introduction
andremoval of a concurrently arranged experimental contingency. Analysis of behavior in terms
of intersecting sets of contingencies (Goldiamond, 1975; 1984) certainly complicates traditional
behavioral formulations suchasthose in the case of ratio strain or schedule-induction, however
such analysis is apparently necessary to a thorough functional analysis.
Goldiamond (1975) presents a discussion andframework for thedescription andanalysis of
interacting alternative setsof contingencies. Current methods of analysis examining behavior
onlyas a directfunction of an influencing referent contingency arepresented as linearsystems of
analysis and contrasted withnonlinear analyses. Under nonlinear analyis, observed behavior
patterns are examined in the context of a jointfunction of the referent contingency andoneor
more alternative and influencing contingencies. Rationality of seemingly irrational behavior is
defined in thecontext of available alternative behaviors, contingencies, and jointeffects.
Forexample, as reported by Goldiamond, nondiscriminated avoidance responding is
paradoxical when examined linearly in the traditional methods of operant analysis. Behavior
underthe control of nondiscriminated avoidance tends to occurwith noobservable
consequence in that responding serves to postpone anotherwise imminent shock. Hence
linearly in the context of the three-term operant contingency, given theoccasion of the
experimental situation a response is followed by no discernible event. Onthe surface this
appears to be "irrational" behavior from aconsequential perspective andin appearance this
imitates extinction, yetavoidance responding is quite likely to persist for longperiods of time,
evenafterformal extinction procedures areundertaken (if, for example, the machine dispensing
the aversive consequences of not responding wereunplugged). When examined froma
nonlinear perspective this behavior is understood or "makes sense" in that responding is a
98
function of alternatives. Onehasto lookat theconsequences of not responding as wellas of
responding to seethis rationality. That is,altemative contingency arrangements, inthiscase
shock, become probable whenno responding isobserved andthisstateof affairs more or less
forces the occurrence of an "irrational" behavior (that is, responding for noobservable
consequences), asdefined linearly.
A procedural limitation exists inthecurrent experimental analysis of ratio strain. Results of
this experiment clearly suggest the importance of available response opportunities andthe
contingency relations thereof in theanalysis of behavioral persistence anddecrement. However
the procedures undertaken to make thispoint areperhaps notthose most relevant to the
examination of ratio strain. Ratio strain when examined inthecontext of a myriad ofavailable
alternative behavioral opportunities refers to conditions which depotentiate or make less
probable a referent contingency relation andpresumably make more probable an alternative or
number of alternative existing contingencies. These depotentiating conditions would include
the current topicsof investigation among others.
Ratio strainwith regard to a referent contingency would refer to theconditions andpointat
which a freely available alternative was undertaken, andthis is the basis of the present procedural
limitation. In thecurrent experiment thealternative wasconcurrently introduced and later
withdrawn to affect response disruption. Ideally, onewould hope to examine ratio strain as a
function of existent alternatives andthose conditions influencing referent performanca versus
the engagement of an alternative. Forexample, it might be interesting to examine at which point
of increasing effort or requirement a particular alternative contingency relation becomes
prepotent when a number of topographically andfunctionally distinct altematives arefreely
available to a subject performing under schedule control. Ratio strain with regard to the referent
contingency in thissituation would referto the switching pointwhere analternative was
undertaken rather thanto the complete cessation of performance. Although analysis of
performance decrement in the context of freely available alternatives wasnotundertaken in the
current study, preliminary examination of thisarea suggests the present findings hold in that
responding is more likely to deteriorate or show strain in thepresence of a freely available
alternative than in theabsence of such alternative givenan increase in schedule requirements.
That is, performance is morelikely to besustained when noaltemative is present. The interested
reader is referred to Appendix II fora preliminary single-case inquiry into thisarea.
99
GENERAL DISCUSSION
Ratio strain, or thedisruption of established patterns of operant perfonnance, wasexamined
in a series of three experiments specifically observing theconditions under which performance
decrement occurred. Performance of human responders wassubstantially affected bythe
degree of increase in a regimen of increasing fixed ratio schedules withperformance
deterioration onlyoccurring with relatively large schedule increases (Exp. I),by thepresence or
absence of alternative behavioral opportunities (Exp. III), andto a lesser extent bythe response
effort requirements of the operant involved (Exp. II).
In each of these three experiments patterns of response decrement deemed to be strained
were qualitatively andquantitatively different. Performance decrement in Exp. I was immediate
andprogressive inthatpatterns of responding became variable withthe advent of larger
schedule increases. Thisvariability, specifically response frequency decrement, occurred
increasingly overensuing sessions. In Exp. III perfonnance deterioration, although immediate
andsubstantial, reflected a stable level ofdecrement. Deterioration in Exp. IIwaslesssubstantial,
although immediate, andagain represented a stable levelof decrement.
Thediversity in experimental conditions producing behavioral decrement and thediversity in
the behavioral topographies produced under these conditions suggests thatstrained
performance canbe a function of multiple variables. These experiments have attempted to more
closely examine conditions producing ratio strain and these results leadto conclusions similar in
nature to those reported in a recent study by Peden andTimberlake (1984) whoconcur that ratio
strain "is nota unitary phenomena".
Often times researchers andclinicians implicitly imply thatratio strain isa unitary
phenomenon. It canbe seen upon closer scrutiny of theuseof ratio strain as a behavioral
descriptor, that theterm is readily accepted andused bythis population, and further thatthisuse
is specific to thedescription of performance decrement. However, thecurrent research and
literature reviewed suggests thatthissingularity of useat bestcamouflages a number of
experimental andbehavioral relationships andthrough precedent of terminological acceptance
precludes ongoing analysis of performance decrement relationships.
Thistraditional perspective of ratio strain is based ontheperspective of strain asa
construct-like phenomenon. That is, ratio strain is seen assomething that occurs. Anorganism
stops responding because it has "ratio strain". In actuality perfonnance either isor is notoccurring
100
andthis responding or lackof responding canbe analyzed as a function of controlling variables. It
is thisdetailed analysis that is precluded by theevocation of strain.
The importance of controlling variables andenvironmental control in performance and
conversely, in response cessation is implicitly yetdistinctly acknowledged in thecurrent usage of
thedescriptor ratio strain bytheconditions underwhich it is used. Forexample, ratio strain
universally refers to performance decrement underthecontrol of ratio reinforcement schedules.
Often times, but notexclusively, strain is saidto be occurafterresponse decrement when the
ratio requirement is high, or afteran increase in the schedule requirements or perhaps aftera
decrease in reinforcement density. It is through these commonalities of performance decrement
conditions that ratio strain is reified to the extent that it is accepted asa construct influencing
behavior.
The purpose of thecurrent research hasnotbeento suggest that ratio strain is anantiquated
or useless term, for thecurrent widespread usage canbe considered onebarometer of value, but
rather to directattention in this usage to those specific conditions influencing the performance
decrement or disruption indicated by thedescriptor. Inthis sense ratio strainmight be thought of
more as a concept thana construct. That is,asa concept, ratio strain canbe defined bya
description of effect under the conditions of usage (Markle andTiemann, 1969). Inborrowing an
example from Markle andTiemann, a number of objects varying in topography might be
conceptualized as, or called, a chair. In actuallity, a chairis distinguished fromother sitting-objects
(stools, loveseats, beach lounge chairs, etc.) by onlythree defining attributes. That is, a chairhas
a backsupport, seats a single person, and sitting occurs at an appropriate (knees bentat a 90
degree angle) level. In defining these critical characteristics present when people label "chair",
we are not onlyableto identify irrelevant attributes thatmaycloud perception, but can promote
generalization and discrimination in useof thedescriptor "chair". Hence, we areapproaching the
essence of theconcept chairthrough definition of environmental conditions present upon
attribution of the concept. When analyzed in this regard ratio strain canbe observed to be a
performance decrement metaphor defined by theconditions present uponattribution. This
analysis directs theconfirmation of ratio strain towards the identification of generating or in this
case, de-generating conditions.
Ratio strain inthe context of the present analysis andexperiments refers to thebehavioral
effects of a specific class of operant decremental operations. Onlyasubset of alloperant
decremental operations are relevant to ratio strain. Forexample operant decrement brought
about through theuseof punishment, extinction, conditioned ernosonat reaction procedures,
101
andphysical restraint areusually conceptualized as independent of ratio strain. Additionally, it was
earlier suggested that response decrement brought about by fatigue was not relevant to ratio
strain, norwas decrement resulting froma lengthening of the PRP.
Onthe otherhand the present experiments clearly identify several conditions of operant
performance decrement which maybe relevant to ratio strain. Certainly the present experiments
examine performance decrement underratio schedules, a precursor to thedevelopment of
strain, andspecifically underhigh ratio schedules, which have beenimplicated as being related to
ratio strain by a number of authors. As may begathered operant decrement, perse, maynotbe
directly relevant to thatwhich hastraditionally been conceptualized as ratio strain. However, an
understanding of what is mostrelevant to ratio strain mayperhaps bebestdelineated through
comprehensive examiiiaiioiiof decremental operations andthrough an ensuing process of
procedural classification.
The present experiments thenclearly identify several conditions of operant pertormance
decrement. Undoubtably, a number of otherconditions leadto performance disruption aswell
andfurther research in these areas will provide a comprehensive analysis of ratio decrement and
hence strain. In Exp. I the influence of program increases in performance decrement was
observed. Large increases wereshown to produce strain whilesmaller step increments
produced undisrupted response continuity at highratios. Significantly, strained performance was
notobserved at very high ratio values when response alternatives werenot present tc the
responders. Thisexperiment suggests thatthe program usedto establish responding is a critical
element in the generation of strained performance. This experiment alsosuggests that the
ultimate ratio values incorporated maynotbe important undercertain program conditions, that is
in response situations where no response alternatives exist. Exp. IIIshowed performance
decrement at these same highratio values when alternative response opportunities were
available to the responders. Thissuggests thatanother critical element in examining
performance decrement lies in the available response alternatives.
Although not explicitly examined in thisseries of experiments. reinforcement magnitude also
appears to be a variable influencing performance disruption underratio schedules (Peden and
Timberlake, 1984). Forexample the authors report that pigeons' responding wasmaintained
underconsiderably higherratios when theywere working for unlimited access to grainunder a
closed experimental economy, whereby all obtained food wasprocured duringexperimental
sessions as reinforcement. Under a similar experimental arrangement Findley reports that
pigeons will maintain responding at FR20,000 (asreported in Whaley and Malott, 1971)1 In a
102
second experiment Peden andTimberlake report that strain was more Iikey to occurwhenthe
control of reinforcement access by a timerproduced smaller rather than larger reinforcement
magnitudes underincreasing FRs. Thus. strain has been postulated to occurdifferentially under
increasing FRswhen reinforcement density isdirectly manioulated,
A final response decrement operation examined in the current research pertains to the
response effortrequirements. Exp. II examined performance decrease occurring as a function of
increased response effort requirements. Under theconditons examined in the current research
response effortmanipulations didnotconsistently produce strained performance. While
response effortmanipulations consistently produced performance decrement. onlyoneof three
subjects showed a decrement substantial enough to meet the designated 20%response
decrement level selected for identifying strained performance.
Thedecreased response levels observed underthese procedural conditions were notthen
commensurate with those seen inthe otherexperiments of thisseries. These results suggest
that although response effort variables are important in perfonnance decrement and in selecting
operant levels of performance. theymay notbe specifically implicated in the occurrence of strain
astraditionally formulated. at least under theconditions observed in thisexperiment.
Performance decrement occasioned by response effort increases showed a stabilization of
effect. Thisstabilization leads one to suspect thateffortmanipulations perhaps produce
decreases in performance through theprovision of physical constraints on the frequency of
behavior ratherthanthrough other means. when a relatively high rate operant selected for
analysis is examined as inthe present case andwhen relatively low response forcerequirements
are incorporated. again as in thepresent case.
Although ratio strain hasbeen defined rather precisely in theory. by a number of authors. the
preceding review suggests thatthese definitions areinconsistent. In practice. strainis implicitly
defined asanyof a number of performance decrement situations traditionally summarized as
those situations in which the ratio is "too large" to sustain performance. It is currently argued that
a greater specification of behavior-environment relations during perfonnance decrement will lead
to a more complete understanding anddefinition of the concept of rntlo strain.
103
Given this, a definition consistent withthisthesis might be that ratio strain refers to the
behavioral effects of a speciflc class of operant decremental operations whereby ongoing
performance decreases in frequency under fixed or variable ratio schedules of reinforcement asa
function of:
a) the increment program utilized in increasing ratio requirements;b) alternative contingency arrangements;c) arrangements of reinforcement magnitude anddensity;d) response effort requirements (although not explicitly suggested by the
present results); ande) presumably otherevents currently unspecified.
Ontheotherhand, ratio strain is exclusive of the PRP, that is, it does notreferto decrements in
response probability occurring as thePRP increases, nordoes it referto physical fatigue or other
currently defined performance decrement operations (stimulus change, extinction, restraint, CER
procedures, etc.).
When examined and defined in thismanner, andalthough a defining characteristic of strained
performance isoccurrence underratio schedules, it is suggested that similar procedures will affect
interval schedule performance aswell. However, to date theconcept of "interval strain" is
nonexistent. Thisis notanaltogether undesirable stateof affairs inthat investigators examining
interval schedules areforced to more closely define the procedures affecting performance rather
than envoking strain. It is suggested that a similar courtesy bepaidto those conditions of operant
decrement underratio schedules of reinforcement.
As it isunlikely thatthiscourtesy will be respected, ratio strain as a descriptor is perhaps most
appropriately reserved for description of the typeof performance decrement observed in the first
experiment herein, thatoccurring as a result of program increases. Thereasons for thissuggestion
aretwofold. First this is in linewithcurrent usage whereby ratio strain is posited to occur "when ratio
requirements areveryhigh or are raised abnJptly" (Sulzer-Azaroff andMayer, 1977). Secondly, this
targets aneasily identifiable procedural condition, whereas for example, "too large of ratio" is as
discussed influenced by a host of procedural arrangements. In thiscontext strain describes a
transition/stimulus control phenomenon whereby performance decreases occuras a function of
changes in stimulus control withperformance maintenance affected byprocedures enhancing the
transfer of stimulus control during schedule changes.
The nature of response patterns occurring to thetimeoutkey in thecurrent research suggest
thatthe aversive emotional components oftenattributed to strained performance mayneed to be
further researched and documented. Ratio strain, perhaps more thananyothercurrent concept in
104
behavior analysis, is heavily anthropomorphic in use. Thatis, ratio strain suggests in usethe
uniquely human attributes of burnout, frustration, depression, etc. Thistendency results from
traditional formulations attributing aversive emotional components to highratio schedules (Azrin,
1961). Currently observed patterns indicate thattimeout responding is more of a
contingency-induced phenomenon and lesslikethe traditional escape from positive reinforcement
formulation. It is suggested that thisanthropomorphism is unwarranted.
Inthe current research timeout responding wassystematically observed to occurunder
conditions of increased variability in performance underthe control of the referent main key
contingency. To the extent thatperformance variability wasgenerated following a substantial
increase in ratio requirements, thecurrent results werein agreement withtraditional timeout
analyses. That is, increases in timeoutswereobserved withincreasing schedule requirements.
However, when theprogram of schedule increases wasgradual showing sustained main key
performance theseresults were notseen. In fact. otherobserved conditions of timeout
responding in the current experiments suggest that this behavior is undercomplex control andthat
the mechanisms governing timeout responding aresubstantially different fromthe postulated
escape fromincreasingly aversive conditions associated with increasing ratio requirements. The
introduction of thetimeout contingency in the current research was consistently observed to
decrease responding to the keyonwhich thatcontingency was imposed suggesting thatthe time
outwas in fact a punishment procedure, a point supported bycomparative examination of response
patterns to each thetimeout andextra keys.
It is suggested that response patterns observed in the current research to boththe timeout
key andthe ancillary extra key were undercontingeny-induced governance. That is, responding to
these keyswas induced by a variety of change-conditions in the referent main keycontingency. A
variety of generating conditions were observed in the present experiments including
contingency-change in the form of increasing schedule requirements andparticularly those large
increases producing strain (Exp. I),extinction (Exp. I), increases in response effort (Exp. II),and
nuances in inter-contingency relationships, specifically decreases in overall experimental
reinforcement density produced by extinction of a concurrent contingency (Exp. III). Ingeneral this
analysis of contingency-induced responding is consistent with Zimmerman and Ferster's (1964)
analysis of timeout responding. They postulated thattimeout responding occurred asa relatively
weakalternative behavior increasing in frequency underconditions of schedule variability.
105
Patterns of timeout performance by human responders weretherefore consistent with those
reported by Zimmerman andFerster whoexamined infrahuman response patterns. Ingeneral the
human operant response patterns currently observed weresimilarto patterns evolving out of the
animal laboratory, although performance wasobserved to be sustained at quite highvalues with
littlepost-reinforcement pausing, a finding previously acknowledged by invesitigators examining
human operant performance (Sanders, 1969). An interesting avenue of research with regard to
post reinforcement pausing issuggested by Experiment IIIbpresented in Appendix II. Substantial
PRPs wereobserved when an alternative wasavailable to the subject. Theexperimental analysis of
alternative response opportunities is particularly germane to understanding human performance in
thateveryday environmental conditions typically provide a widerange of alternatives. The narrow
range of alternatives inherent in the laboratory studyof human operant performance is undoubtably
onecontributing factorto the large ratio values shown to sustain responding in thecurrent
experiments. These large FRs areparticularly notable in lightof thefactthattypical studies of
human operant responding examine performance at values considerably lessthanexamined
herein. Forexample. studies of human FRperformance haveexamined behavior underthe control
of ratio schedules ranging anywhere from 1-1000 responses whereas FR values of 1000, 2000,
4000. & 8000were currently programmed. Further examination of theeffects of alternatives on
behavior andof human performance underlarge schedules is warranted.
There area number of clinical implications of the present experimental analysis of ratio strain.
Forexample, Ferster (1973) hassuggested thatstrainandschedules of reinforcement are
important in the analysis of clinical depression. That is.depressed individuals may be suffering from
"ratio strain". Thecurrent research suggests several pertinent points. Thefirstof which is the
recognition thatstrain is a multicomponent phenomena andthat depression astheoretically tiedto
ratio strain is undoubtably likewise, in thatdifferent "types" of depression are likely to share
commonalities withdifferent "types" of strain. In otherwords, it is likely thatdistinctions in cases of
clinical depression will arise asa result of distinct behavior-environment relations. Thepresent
research thensuggests that individual cases of depression should be individually examined to
ascertain the specific environmental conditions influencing the depressed behavior. Tothe extent
that ratio strain is but onecontributing component to clinical depression as suggested by Ferster, it
is likewise important thateach case of clinical depression be approached asa particular ideographic
set of contingencies to be addressed rather thanthe currently acceptable division of depressed
disorders intomanic, reactive, bipolar, cyclothymic, dysthymic, unipolar, or various permutations
thereof.
106
Given this ideographic approach, the current research suggests at least two procedural
conditions producing depressed behavior or affect andasa result suggests assessment and
treatment approaches. As mentioned earlier andasexamined in Experiment I, ratio strain can occur
as a result of changing environmental conditions producing a lossof stimulus or schedule control. It
might be expected that the verbal behavior of patients in thissituation may reflect this
environmental condition in that theymay"notknow what is expected of them" or otherwise may not
be ableto describe controlling schedules of reinforcement. In this regard a confusion over
appropriate behavior mayabound as in the case of thedepressed individual exclaiming things like,
"I don'tknowwhatyou expect from me," or "things usedto besosimple," or "justtellmewhatthe
(expletive) you wantmeto do". Thisview of depression to some extent approximates the learned
helplessness theoryin thata historical program of development leads to behavior notunder
defined stimulus control.
Depression mayalso be linked or arise as anoffshoot of the alternatives available to an
individual. Thecurrent research suggests thatanexamination of alternatives is critical in
understanding behavioral persistence anddecrement. Although topographically onemight argue
that depression or depressed behavior is relevant to the reduced frequency of behavior observed
upon the introduction of an alternative in actuality it seems more likely thatdepression may arise in
the converse situation where noalternatives arepresent, or depression itselfmaybethe only
available alternative. Inthissituation thecultivation of altematives andconstruction of behavioral
repertoires through skill-building interventions seem warranted.
Several points of clinical relevance to theelimination andconstruction of behavior evolve from
the current research. The persistance of behavior observed underconditions of extinction and
response effort suggest that these arenotveryviable procedures to eliminate behavior when that
behavior iscritical in a setof altematives, that iswhen noalternatives areavailable. Ontheother
hand the introduction of an accessible alternative wasseen to substantially eliminate behavior and
this suggests a valuable clinical procedure. The construction of alternatives as a specific
intervention in theelimination of disturbing behavior hasbeen termed theConstructional Approach
by Goldiamond (1974). Finally, theprogramming process utilized when developing behavioral
repertoires wasseento becritical in thecurrent research. Significant repertoires were developed
whenprogram increases weresystematic andgradual andnototherwise.
Several research areas aresuggested when looking toward potential future extensions of this
program of inquiry. A number of these have been suggested overthiscourse of discussion.
Perhaps first and foremost, further quantification anddelineation of operant decremental
107
operations iswarranted in a continuing intensive analysis of thatwhich hastradltlonally been
summarized as strained performance. Secondly, a number of interesting contingency change
relations andcontingency-induced patterns havecome to lightin thecurrent study. It would be
interesting to pursue these issues andexamine theenvironmental conditions generating and
maintaining historical and novel induction. It has traditionally been assumed thatstimulus change is
onecondition likely to produce performance disruption (Azrin, 1959) andhence set the occasion
for behavioral variability. Thecurrent study suggests thatperhaps such change-induction is a more
pervasive phenomena, withstimulus change procedures being onesubset of a larger
contingency-change phenomenon. Forexample, schedule changes have been shown to induce
ancillary keyresponding in thecurrent research ashave response effort changes. It is possible that
each individual element of a contingency relationship issusceptible to generating induced
behavioral patterns upon change. Such elements might include deprivationallevels, schedule
relationships, stimulus control relationships, consequences, behavioral requirements, ambient
stimuli, instructional andabstractional requirements, etc., for example. In addition thecurrent
research suggests thatdifferential patterns of induced responding are related to the alternatives
present andthis areaof research maybeenlightning.
It hasalsobeen suggested thatan analysis of alternatives andhuman operant performance
mayprovide information on post reinforcement pausing by humans. Itwould be worthwhile to
systematically replicate the single case preliminary inquiry presented inAppendix II examining
human performance, ratio strain, andthe influence of alternatives. The lackof research with human
responders at large ratio values suggests a further area of stUdy.
Finally, an area of research notaddressed bythecurrent study concerns theanalysis and
assessment of verbal behavior, private events, andsubjective states by human subjects under
similar experimental control. Current experimental procedures andcontrols prohibited intensive
examination of these phenomena yet informal observation suggests that relationships between
public andprivate events would be aninteresting area of analysis. Numerous investigations in this
areaare possible however with regard to the current focus it would be interesting to examine
subjects verbal report andsubjective impressions of contingency control underconditions of
response maintenance andstrain. Thistype of study may provide valuable information asto, for
example, the purported aversive nature of highratio schedules or feelings of helplessness, etc.,
underconditions of lossof schedule control and may bea particularly valuable adjunct in the
extension of laboratory studies to the clinic.
108
APPENDIX I
Cumulative records of sessions 1-44for subject ARcanbeseen onpages 109-112. Subject ARparticipated in Experiments I. II,& III.
Cumulative records of sessions 1-48 for subject IN canbeseen onpages 113-116. Subject IN participated in Experiment I.
Cumulative records ofsessions 1-37forsubject PWcanbeseenon pages 117-120. Subject PWparticipated in Experiments I &II.
Cumulative records of sessions 1-20 forsubject HNcanbeseen onpages 121-122. Subject HN participated in Experiment I.
Cumulative records of sessions 34-58 for subject LDcan beseenon pages 123-125. Subject LDparticipated in Experiments II & III.
Cumulative records of sessions 1-38 for subject LMcan beseenon pages 126-129. Subject LMparticipated in Experiment IIIb.
On the following pages. cumulative records andresponse frequency information to each of
the three response levers exists aswellas timeoutduration information foreach session of
participation by thesix experimental subjects observed overthecourse of these experiments.
When examining the lower portion of individual cumulative records analphabetical and
numerical legend exists describing relevant subject, condition, and response information.
Information is presented in the following manor:
Subject Initials; Session Number; Experimental Condition;Response Frequency to the Time OutLever/Duration of Time Out;Response Frequency to the Extra Key; andResponse Frequencyto the MainResponse Lever.
Forexample a legend: AR 1 GR2 FR 250 233 484 885. would read: Thefirst
session of theexperiment for subject AR responding undera Geometric Ratio 2 schedule witha
terminal component of FR 250(see text for specific details of experimental conditions). Subject
AR responded to thetimeout key233 times, to the extra key484times and to the main key885
times. Since noduration of timeout is reported. wecanassume thatthe time outcontingency
hadnotyetbeenintroduced to thatkey. If a duration of timeoutwere present the legend might
looklike this:PW3 FR250 1/2.6 4 7261, meaning thatsubject PWengaged in 1 time
out of a 2.6second duration.
All 3 CIl Z~ Z50 0 17 lCJ721 All. ra Z50 no tID Cl& " • II • Hn
, ,
AIlS Fa:ZS1 no tID Cl& 0 11 10101 .• All. ra Z50 no tID Cl& • 20 .11 ,"1
~ .All 11 ft 500 Q • 7 "rno All 12 PIl 50Q ' . 0 Q 10M2
All Z7 nl ~OOO Q . .:lS1 . 1l)(QJ All ZI ra - _.. Q . Q ~4
• rI .
All U III 4000 _ Q JI",. .Ajl JQ 1Il_ _ Q I f1~
r
. -All JS III 4000 _ Ql Z 1Il5GO ZI'.5 5190 S6IZ All JI III _ _ VI. I J4~ 7151
112
AP. U f'Il 4O')Q Q ItT 10100 AP.4t n 4QQQ Q:I4' • lQJ4lI
· JlI 11 1111000 0 0 M75 ~ IZ 1111000 0 J MZO
'"J11 17 Do 4000 . • 0 2 1_' JII II ftl 1000 0 0 . 103
.. JII,' ralooo • 0 • 4G2 1371 JII20 ft 1000 1/.' JS4 1199
--J11 D III 8000 1/.7 _ . 1m -.. ZI III 1000 0 1 1001
JII Z5 _
t.
-3J1 _
0" JJ _
o
o
o
·0
o . I44C
o lUG
.0Jll21 111&.
JII ZI E4, .
JII 10 111& o o 1,"
115
V.7 ".llIJi 111& 1017
r
JJlf:ll 1:01: o .0 . no.
116
J
. Jlf J' .1:01: 1/1.7"Jlf 40 1:01:
,.., ..041
·JlI4\ 1:01: 51'11.a 454 '!170
1I
JlI4J - • 1/.3 500 ..II JlI44 - ..".' 111:1 44Z'
~I
L.1.oM1. ,~u.~1•. JlI 45 1:01: 1/1.1 . . lIS POZ •• JlI .- - :I14/2Ot.7 11Ol1 1I2t
Jlf 41 _ o 0 7171
Jlf 1 c:a Z .lll Z:lO ID~ _ U 13 Jlf Z III Z50 O.
Jlf] III ZSO
Jlf 5 III Z50
1/2.' .
o .
Jlf 4 III Z50
1..
III t III ZOOO ~.J 113 mt
Jlf 11 III ZOOO ·0 ZI 1]50 Jlf IZ III 5GO o 1 1113
IIf 13 n.soo. 7m.
1
IIf 15 n 50Q a a
a a 71"
aaIIIZ4 nzaaa
I
IW ZS ,.4000
IWZS ,.4__
1W11 ,.4__
Q
7
.".
431Q
5411
IW a ft 4000
III J2 ,. 4000 ... Q
Q 2
119
1431
47111
z,
rJ
o o --
120
•,;
•!1
121
., .. 3 m "' m .Z "3. - 1701 1m
r~,
Jz
,,1..3 .. 3 1AO 1703 "'4 " •• I'lL 3 3511 2112 zt07
.S .. 3 30'
., 111250 "·.,0 111250 .7130
122
.. 11 .. %SO 1/.' 1021 II" .. IZ. lIl.Z50 ~07., 7451 1571
. .
ICIZIU.' SUO .IZlI .. 14 III _.. . 1~5t•• %120 ~
.. IS III _ 3Q14'.' ,UI JUS .. ,. III - 43~1J.7 475 '154
. . SIt.' 1m "11 .. ,. "'00 0 9011 ~
.. ,t lit 500 0.,30 tolI1 • 20 .. SOO 0 I'" toG, .
UI JI 1Il 500UI Jt 1Il 500 _
UI 40 1Il 500 """
U1U nt 500 """
UI 44 1Il laoo """ 11'.1 I
4011
Jln
UI II 1Il 500 """
UI U 1Il 100II"""
UI 4S 1Il laoo """
111.5
SIZ.t 1
JCIIlO
al ~ nt 10Cl0 _
al ~ nt lOOl1 _
al 50 nt 10Cl0 _ .. 3 11.5
15
5179
30Cl0
...
al.7 ,. 1ClOO ....
al.' ,. lClOO .... fa 3
•I
al 51 ra lClOO.... fa 3
10 3141
124
al 51 nt 10Cl0 ....
al 54 ft lClOO _
17flU.. 131'
m
11M
1Z407
al 53 ft 1000 ....
XI) 55 n IClOO ....
125
til,. 11l 1000 13/1'.Z Zit 3105 ID 57 1Il 1000 Vl7.' 123 U4t
til ,. 11l 1000 112
= 126
~. l
--til I CIl2 nt 15O m 104 1173 tIl2 ca 2 llt 15O. no 1m 1454
til 11 nt 250 lllI Ale 11S'.. 3' 1424 tit 12 n. 250 Ale 1/212 2lI 10000·
tI' ZJ nt 1000 ... Al.e 7"'" W 5000 tI' 2.' nt IllOQ ue 21'105 0 . 2llOQ
til ~ rlt'ooo .u~ 1/15~ 0 1000 til 30 tIl'ooo ... .ut tI/12.' 34 4281
~ ;
f~ ,I
til JS III 250 ... A1~ 0 to 716' til JI til ~O .ut 111240 4 ~OO
129
til 31 m 250 "" Ale WI'I1.9 14 3501
130
APPENDIX"
Theinfluences of an available alternative behavior on schedule performance and schedule
disruption were examined in Experiment III. Inthat experiment the alternative consisted of the
specific introduction and removal of a concurrent reinforcement contingency on a second
response leverwhich wasof a higher reinforcement density thanthatcontingency programmed
to the main lever. This wasseen to have an immediate andsubstantial effect on the referent main
key response pattern suggesting that indeed alternative response opportunities andthe
contingency relations thereof areimportant in theanalysis of strain.
Although Experiment III examined the influence of alternatives in thecontext of the
introduction andremoval of a concurrent contingency, ideally onewould wantto examine ratio
strain asa function of existing topographically and functionally distinct alternatives. Thecurrent
experiment (1IIb) explores performance relations withsuch analternative. Specifically, referent FR
performance is observed underbothbaseline andincreased schedule conditions withand
without the presence of a topographically, in terms of response requirement, andfunctionally, in
termsof outcome, distinct alternative.
METHODS
SUbject
Onlyonesubject. LM, parlicipated in this experiment. Therecruitment procedures utilized in
contacting the subject were the same as specified in Experiment I. The relevant subject
information is asfollows:
LM Subject LM was an 18-year-{)ld Caucasian male high school student, living at
home. At the timeof participation thesubject wasunemployed although hehad
previously worked part-time asa computer programmer. Subject LMparticipated
in 38 sessions overa fourweekperiod collecting 3600 bonus points tor a total
earning of $74.00 andwasthuspaid$3.89 per hour.
Apparatus
Theexperimental apparatus, control equipment, andexparimental room were thesame as
described inthe firstexperiment. A commercially available hand heldMatteiElectronics Footbal/2
131
electronic game, model number 1050-0320, wasadditionally incorporated intothis experiment as
an alternativo Independent of theexperimentally programmed reinforcement schedules.
Procedure
Unless otherwise specified theprocedures used inthls expsrmentwere similarto those
reported in Experiment J.
Experimental Design and Training. Thisexperiment examined response patterns
withandwithout an available alternative underbaseline conditions, an increase in the schedule to
FR 1000, anda return to the FR250baseline. Within this ABA design the effects of the
alternative were examined by randomly, on a .5probability basis, introducing a handheld
electronic game in anexperimental design similar to that of themultiple schedule (Hersen &
Barlow, 1976).
Undera multiple schedule. twoor more schedules of reinforcement arealternated either
consecutively, randomly. or onsome otherbasis wnheachschedule containing an explicit
exteroceptive stimulus. The reliable occurrence of differential response patterns underthe
occasion of each stimulus suggests differential experimental control by each component. In the
present experiment the presence or absence of the electronic gal!1e wasthe exteroceptive
stimulus. If indeed anavailable alternative is an antecedent to performance decrement,
differential responding should develop during these alternating conditions. Specifically, one
would expect the presence of such an alternative to make the referent contingency les-s potent
and influence response rates underconditions likely to produce strain, and lessso underother
conddons.Ina training procedure similar to thatreported earlier with subjects AR, PW, andIN, subject
LMinitially responded undera Geometric Ratio 2 schedule of reinforcement witha maximum
schedule value of 250for a totalof three sessions. At that timethe Time Outcontingency was
verbally introduced andthe following 15sessions werespent underthe FR250baseline. During
the final 14sessions of baseline there wasa .5probability thatthealternative would be
introduced during anygiven session. Thealternative wasactually introduced on 6 of these14
occasions.
Theensuing 14sessions were runwith the Fixed Ratio increased to 1000. The alternative
was introduced on 4 of these occasions, again on a .5probability basis. Finally, a return to the FR
250baseline occurred. This final baseline wasonlysixsessions in length asthe previously
132
specified termination period wasmet. Thealternative was randomly introduced on three of these
sixoccasions.
Verbal Introduction of the ARernatlve. The alternative was initially provided during
the fourth baseline session. Prior to thistimethe electronic gamehadbeen available for the
subject to play in thewaiting room between sessions andit continued to be ~vailable at this time
during the entire course of the experiment. Thesubject engaged in thisactivity on a regular basis
during the between session break.
The hand held electronic game was chosen as a convenient alternative foruseas it generally
requires two hands to effectively play. Itwas incorporated aspartof a mutually exclusive
alternative contingency relation. However, pilot research suggested that everingeneous
subjects wereableto engage in both schedule andelectronic game activities to varying degrees
of efficiency by, for example, hitting the response leverwithan elbow whileplaying the game
with the hands. Forthis reason it was necessary to establish mutual exclusiveness through
instruction andthis procedure will bedescribed.
A second procedural issue in thedesign of thisexperiment involved obtaining anobjective
andempirical index of theamount of timespent engaging in the alternative behavior. Since the
existing control equipment andelectronic game couldnotbe easily adapted for thispurpose, a
decision was made to verbally linkthe instruction (onthe mutually exclusive nature of the
relationship between the game andthe main response lever), to the TIme Out leverwhich was, in
turn, linked to the existing control equipment. Under thisarrangement thesubject was required
to engage in a TIme OUt when playing thegame andto TIme Inwhen notplaying the game. This
allowed precise quantification of the timespent engaging in the alternative pattern. This
procedural deviation, however, necessitates caution when interpreting andcomparing the TIme
Outdata underthose conditions when the electronic game is present to when thegame is not
present or to Time Out relations of otherexperiments.
Thefollowing instruction wasused upon the initial introduction of thealternative:
During some sessions youwill havetheopportunity to playthiselectronic game. If youwantto playsimply hitthiskeyto shutoff theequipment andhit it again when you arethrough (point to left key) to tum theequipment backon.You arenotallowed to interact withthe experimentalpanel inanyotherwaywhile you areplaying. You wiii risk losing points jj youviolate these rules. However, feel freeto playor notplaythe game at yourdesire.
133
SUbject lMNo Alt.
o1/ Alt..
9 12 15 18 21 24 27 30 33 36 39
FA 250
..
FA 1000FA 250",lolI~
GA 2QlG!o 8~ "0cQI
4cOrn.la.., ....J:iIIQ1 2llI:a,
03 6
Sessions
Figure 21. Overview of responses per second to the main responslJlever during all conditions of Experiment Jllb by subject LM.
134
Although subject LMcould freely engage in a time outat any time during thosesessions
whenthe alternative waspresent yet notactually engage in playing the electronic game, visual
monitoring revealed thatthis never occurred. Conversly, LMcouldengage in the alternative
without initiating a timecut. This didoccuronce duringthesession when the alternative was
initially introduced but did notoccuragain afterthesubject was reminded that he risked losing
points by the violation, Hence thisprocedure effectively allowed quantification of response
patterns while establishing themutually exclusive nature of the experimental tasks.
RESULTS
As in priorexperiments, response rates to each of thethreeresponse levers wereexamined in
this experiment andcumulative records of pertormance to the mainkeywerecollected. Specific
procedural information regarding these measures may befound in earlier experiments. The
duration of timespent engaging in the available alternative wasalsoexamined.
Forthe sake of clarity, results undereach component of the multiple schedule design shallbe
discussed separately. Under the necessities of this arrangement, experimental sessions maynot
be in consecutive orderduring presentation of the respective results withandwithout the
alternative present. However, anoverview of the experimental results showing specific sequencing
arrangements canbe seen in Figure 2~. Responses persecond to the main leverare examined
underthe differing conditions of the experiment. The general ABAdesign of FR250, followed by
FR 1000with a return to FR250 canbe observed, along withthe first threesessions showing the
GR2training period. Sessions, plotted along the abscissa, during which the alternative waspresent
aredemarcated onthe graph bya triangle rather thana square. The squares represent sessions
during which the alternative was absent. Ingeneral, it canbe observed thatduring the initial FR250
baseline similarrates of responding canbe observed in each alternative-present/absent design
component whilerates separate when the schedule is increased to FR 1000. Lower response
rates areseen inthosesessions when the alternative is present. Aswill be discussed, upon
reversal to FR 250 a reversal of effect canbeobserved overthe initial threesessions but notover
the final three. A detailed discussion of this figure will notbe undertaken, in lieuof which Figures 22
& 23 are introduced. Response patterns under eachdesign component of the multiple schedule
design maybe observed more clearly from these.
Subject LM
135
76
::.,100" 5IIII11C~~g 4c 5!.! 3...t _v.G~... 23:Q:CII
Q. 1o
I- FR 250 FR 1000 FR 250
-~--
\l-
I-
I-~ g g
7 8 12 13 14 18 24 29 31 32 33 36 37
7 8 12 13 14 18 24 29 31 32 33 36 37
::.,
g-:: 2000 ...---------..,.-----..,.-----~'"~0 ....3~ 100 1500c< ."lI.I II C~ ~ 0 1000C .... 0GI~ GIQ. G mmE: c 500lD III '"e~~:( 0 ... ... .... ...
r:...t
>.oCGI::Ir:t•...II.>.II~
•...~
XlI.I
60
SO40
30
20
10O... ... .....iOIiI.,_..
7 8 12 13 14 18 24 29 31 32 33 36 37
Sessions
Figure 22. Frequency of responding to the main key and the extra key duringthose sessions when an alternative was available. as well as the duration oftime spent engaging in the alternative by SUbject LM in Experiment IIlb.
136
Subject LM
FA 250GA 2 FA 250
J~1 2 3 456 9 1011151617 19202122232526272830343538
7554321o--~ -- ~ ~~!'"'""'~ - ......
>. 1000
~0e
~OOQI::lr:rCD '°1..LI.
~60 No
::l40 iime0
CIl Out::I 20....I- 0
1 2
>.Cl
~ 2000::lgo 1600
~ 1200
: BOO~
• 400.. a)(w
1 2 3 4 5 6 s 10 11 15 1617 19 20 21 22 23 25 2527 28 30 34 35 38
Sessions
Figure 23. Frequency of responding to the main key. the time outkey, and the extra key by subject LM during those conditions whenno alternative was present in Experiment IIIb.
137
Figure 22 shows the frequency of responding to the main keyandthe extrakey during those
sessions when thealternative wasavailable, aswellas the duration of tirnsspent engaging in the
alternative by LM. Thetop graph of Figure 22shows responding to the mainkeyduring the
presence of the alternative underthe three experimental conditions. During the FR 250 baseline
the response rate is in the range of 4-6persecond. A large decrease in rateoccurs withthe advent
of theschedule increase andthis is maintained through the condition. The rate increases to
baseline levels duringthe first session of the reversal, however, this rate is not maintained during
the final two sessions.
The middle graph of Figure 22 shows the duration of time in seconds spentengaging inthe
alternative activity undereachexperimental condition, and an inverse relationship of high
conformity can be observed between thisduration andthe mainkeyresponse rate. This
suggests that the subject was likelyto beengaging in thealtemative behavior when not
responding to the mainkey. So,for example, thisconformity suggests that during the final two
sessions of the FR 250 reversal, subject LM wasengaging in thealternative when not
responding to the mainkey. The lowergraph of thisfigure shows response frequency to the
extrakeyduringthis experiment. Although theoverallfrequency of responding is low, it is
interesting to notethat when an altemative is available, as in this experiment, frequency of
responding to this key is seento decrease withthe schedule increase ratherthanto increase as
seen in previous experiments undersimilar conditions of a schedule change to lessfavorable
conditions. However, this lackof increase maybedueto the priorsubsidence of extrakey
iesponding oversessions following the schedule increase, as the initial session with the
alternative present at FR1000 was, in fact, thesixth session of FR1000 overall. Extra key
responding increases with the return to the FR 250 condition, although decreases overthe next
twosessions.
Figure 23 portrays responding by LM to each of the three response levers during those
sessions of the experiment when no altemative was available. Themain key response rate
portrayed in the topgraphremains fairlyconstant across theexperimental conditions, alfl10ugh a
gooddealof variability in rate can be observed across sessions, particularly during the latter
portion of the FR 1000 condition. Thisgeneral continuity of a high rate is in contrast to the
pattern seenwhen the alternative waspresent as reported earlier. A slightdecrease in main key
response ratecan be observed with the increase in schedule requirement to FR1000, however
thisdecrease isnotnearly asdramatic asthat portrayed in Figure 22when thealternative was
available. A reversal of effect canbeseen during thefinalFR 250 condition in that main key
138
patterns tendto stabalize, although a lower rate canbe seen during the finalexperimental
session.
When the alternative wasavailable in thisexperiment the timeoutkeyactually served to time
theduration of the alternative behavior rather thanfunctioning as a timeout in the manner
previously used(see the discussion of this under Procedure). However, the experimental
conditions prevalent during the noavailable alternative component of the multiple schedule
design, asseenin those sessions portrayed in Figure 23,are indeed similar to the experimental
conditions of the earller experiments, with regards to the functions of the three response levers,
permitting dlscusslon of common relations, particularly with regard to thetimeout lever.
Withthis in mind, several findings observed earlier are replicated when examining the lower
twographs of Figure 23. Theintroduction of the timeoutcontingency at session 4 can be
observed to severely restrict or decrease theamount of timeout responding, whereas the
frequency of responding to the extra keygradually decreases asthe main key response patterns
become predominant. Given this lowerlevel of timeoutengagement, a temporary increase in
frequency canbe seen to occur whenthe schedule is increased. A parallel pattern of increased
responding with the schedule increase canalsobe observed to the extrakey in the lowergraph.
In both of these cases a decrease to baseline levels occurs overensuing sessions. In comparing .
theordinates of the twographs, it can be seen thateven though theoverall performance patterns
appear similar for timeout and extra keyresponding, thegreater frequency of occurrence is to the
extra key.
Each of the respective conditions portrayed in Figures 22 & 23maynowbe moreeasily
recoginzed when placed in theoverall experimental context of Figure 21. As noted with regard to
those conditions where the alternative waspresent, temporary andincomplete reversal of effect
wasseenwith theschedule return to FR250. In Figure 21,a decreasing trend ingeneral across
both conditions canbeobserved during this reversal. In an ABAdesign a reversal of effect
permits the mostpowerful conclusions. Hence, the interpretation of these experimental results
arecomplicated bythis incomplete reversal. Theimplications of thisandthe conclusions drawn
from these results will be presented further during thediscussion 01 thisexperiment.
Cumulative Records
Cumulative records from representative sessions of those conditions whenthe alternative
wasavailabe to the subject are shown in Figure 24. Records fromsessions when the alternative
wasnot present areshown in Figure 25. In each of these figures, theuppertwo and lowertwo
139
Figure 24. Sample cumulative records portraying representative response patterns generatedunderbothbaseline and increased fixed ratio schedule conditions when analternative wasavailable to subject LM in Experiment IIlb. Theupper two records portray responding duringthe initial FR250baseline andthe lower tworecords portray responding under similarconditions following a reversal. Thecenter twocumulative records portray responding underthe increased ratio condition of FR1000. A complete collection of all cumulative recordsgenerated during this experiment is included in Appendix I.
1
140
·• =•
= •=• e •
j : e.. :~
:; :5..~
:5.. :5•.. ..
•~
•0
0'"
'" .. '".. .. .... ..
• • ·~ :: ~..• • •f C C
· : :... ... :;•
~
:5 :5 :5~
..• •0 8
0.... --.. .. '"
===.. .. ..eo eo
eo ::
141
Figure 25. Sample cumulative records portraying representative response patterns generatedunderbothbaseline and increased fixed ratio schedule conditions whenno alternative wasavailable to subject LM in Experiment IIIb. Theupper two records portray respondinQ duringthe initial FR250baseline andthe lower two records portray responding undersimilarconditions following a reversal. Thecenter twocumulative records portray responding underthe increased ratio condition of FR 1000. A complete collection of all cumulative recordsgenerated duringthis experiment is included in Appendix I.
142
....-•e:....o•oN
••
"e:..a•
-= 3 -~•e•.-....a•
••=
•.. •=
.- e •• ": ..: .. :.. ..- • ....
I0 • 0•• §0 0
In'" .. ...• • •• • •
143
records portray response patterns generated during the preandpost manipulation FR 250
baseline conditions, respectively. Thecenter cumulative records are representative of those
generated underthe FR 1000 condition.
When comparing Figures 24 & 25 a striking difference canbe observed in response patterns
particularly during sessions when the schedule requirement is increased to FR 1000. When the
alternative is present fewerFixed Ratios are completed and longer post reinforcement pauses
occur, during which timethe altemative is engaged in asshown by thetimeout resets. Time outs
are more frequent during thesessions withthe altemative present giventheverbal linkage of the
timeoutsandthealtemative explained earlier. Ingeneral, the sessions where the alternative is
available seem to present less variability in response rate andthe response patterns givethe
impression of being more "organized". That is, responding to themain leveris eitherstrong, or
responding is to the alternative. More variability is seen in those records presented in Figure 25
in the sense thata variety ofshortpauses arepresent in thedifferent records particularly the third
andthese pauses don'tconform to anapproximately standard length. Responding, when
occurring does appear quite consistent however. Responding generally occurs throughout the
FR 1000sessions in contrast to those records portrayed in Figure 24.
Response patterns to themainleveroccurat a high and consistent rate in Figure 24. Time
outsoccurduring periods of no responding and visual observation confirms that at these times
thesubject wasengaging inthe altemative. The first andfifthgraphs showsomewhat similar
patterns of responding, l.e.,a highandconsistent rateoccursforthe majority of the session.
Towards the endof the session the subject stopsresponding and switches to the alternative
following a reinforcer. Theresponse rate of the first record is much higher than thatof the fifth
with40 reinforcers obtained during the former session, compared to 30 in the latter.
Thesecond cumulative record shows a similar yetdistinct pattern. During thissession 30
reinforcers areobtained with the alternative undertaken afterevery tenthreinforcer. Ingeneral
during this experiment, thesubject wasmore likely to pause following a "grouped" n!.!mber of
reinforcers such as 10,or 30, or 40,for example, than at othertimes.
Thethird,fourth, andfinal records of Figure 24showcomparable overall response patterns.
During each, thepredominant behavior portrayed is the ahemative. In each record an initial
schedule completion, or two in thecaseof thesixthrecord, is undertaken witha timeoutoccuring
afterreinforcement. It is during this timethatthe alternative is predominant. A second schedule
completion occurs about half-way throug.... the session of the third record, afterwhichthe
alternative is again undertaken.
144
Sample cumulative records of response patterns generated under each schedule conditon
when thealternative wasnotavailable arepresented in Figure 25. The first tworecords of
responding underthe FA 250show a highrate pattern for most of each session whh a substantial
pause occuring toward the end of the session. Thispause is particularly evident inthe second
record, occuring after the thirtieth reinforcer rather thanthefortieth of the first record. An
additional schedule completion occurs in the first record following the 2-3 min. pause whereas in
the second a timeoutof approximately 2 min. in length occurs afterabout 10 min. of no
responding. Some variability in response ratecanbeobserved, even during the period of these
two records when responding is strong, however for the most part, response patterns portrayed
areconsistent andof highrate, with littlepostreinforcement pausing seen.
Thethird and fourth records of Figure 25 show sample patterns during the FA 1000
condition. A break andrunpattern canbeseenduring the third record, particularly early in the
session. During thefirst two schedule completions a number of breaks canbe observed,
however the final three schedule completions showa steady response pattern of lower overall
rate thandlJring the FA 250baseline sessions. severalsubstantial PAPs canbe observed anda
timeoutoccurs at theendof thesession. A higher response rate commensurate withthe
baseline sessions canbe seen inthe fourth cumulative record. Thissession is characterized bya
highandconsistent response rate whh littlepostreinforcement pausing, although brief PAPs can
be observed following the finaltwo reinforcers of the session.
Thefifth andsixthrecords of Figure 25show responding following theschedule reversal to
FA 250. Thirtyreinforcers areobtained during thesession portrayed bythefifth record. In
general, a steady high rate response pattern with no PAPs canbeobserved. A lowering and
some variability in response rate occurapproximately one-third of theway intothe session anda
single within ratio pause occurs two-thirds of thewaythrough the session. A timeoutoccurs
following the final reinforcer.
Thefinalsession portrayed in Figure 25shows an anomolous record of responding. The
initial FA 250schedule completion does notoccuruntilover10 min. into the session. During this
initial ratio a number of timeouts andbreaks canbeobserved. A long PAP then follows.
Aesponding resumes latein thesession andfor the most part is then characterized as steady,
showing a highrate, andwith little post reinforcement pausing. However, some variability in rate
canbe observed during the finalfewschedule completions anda brieftime out is also seen to
occurduring the initial stages of oneof the latter ratio runs. An interruption of the drawn line
during the reset of the response marker occurs intwo instances onthisrecord. This is a
145
mechanical artifact produced by the reinforcement slash being drawn whilethepen is resetting,
and should bedisregarded.
DISCUSSION
Inthe current experiment ratio strain was examined in thecontext of an available alternative.
Heretofore, responding by human subjects wasobserved to be persistent at large fixed ratios
although instances of perfcrmance decrement were seen under specific conditions in
Experiments I & II. Experiment III suggested thatalternative contingency arrangements were
influential in understanding response persistence under thecontrol of a referent contingency,
although the specifics of that experiment were notnecessarily those mostappropriate to the
anlaysis of ratio strain. In thecurrent experiment procedural arrangements were modified to
better assess the effects of alternatives andstrain. and indeed performance disruption was
observed to occurto agreater degree under those circumstances where the subject wasgiven
the opportunity to engage inan alternative.
Specifically. performance was examined withand without theavailability of a hand-held
electronic gameunder a relatively lowfixed ratio of 250 and the higher FR 1000 with an ensuing
FR250 reversal. The electronic game was selected forusein thisexperiment because of its
topographical andfunctional independence relative to the established button pressing
response. Under the initial FR 250performance was seen to be persistent regardless of the
arrangements dictating the presence or absence of thealternative whereas preference for the
altemative wasclearly observed at thehigher fixed ratio. When thealternative was notpresent at
the higher FR responding was persistent suggesting that indeed the alternatives available to a
subject will influence theoccurrence of strain in thesense of performance decrement.
Thestrained performance examined herein is hence related to a largercontingency
ecosystem. In thiswell-defined experimental situation it is possible to clearly seethe influence of
alternative contingency arrangements onthe referent response pattern. In a less well controlled
environment. perhaps more akin to the world outside the laboratory, Itmaynotbe possible to
assess the presence or influence of alternative contingencies. Inthis situation performance
would be examined linearly rather than non-linearly in thatbehavior would be examined solely as
an effect of a specific occasion-behavior-consequence relationship rather thanin the context of a
host of suchcontingencies. Under linear analysis it is necessary to evoke theconstruct of ratio
146
strain to explain classes of performance decrement analogous to that presently seen which wasa
direct function of the introduced occasion to engage in an alternative behavior. Conclusions as
to specific ratio sizes capable of sustaining human operant performance would varyconsiderably
under linear andnon-linear assessment giventhepresence and/or absence of alternatives.
An incomplete reversal of experimental effect wasobserved in thisexperiment. Following
theexperimental series of FR1000 aschedule reversal to FR250wasundertaken forsix
sessions afterwhich time the experiment wasterminated under a previous temporal agreement
between subject andinvestigator. A reversal of effect was observed during the initial three of
these sessions in thatperformance was persistent at a rate consistent with the initial FR250
baseline regardless of thepresence or absence of the alternative. Ideally, the strongest
experimental conclusions would have derived outof a continuation of thispattern overthe
remainder of the experiment. However, thiswasnotthecase. In the final three sessions a
distinct decrease in response rate occurred undereach of the twoalternative present andthe
single alternative absent experimental sessions.
It appears, while an initial reversal was present upon the reduction of the ratioto FR250, that
anoverall decreasing trend occurred during thefinal threesessions as presented in Figure 21. In
actuality it is hard to definitively assess thisdecreasing trend inthe no alternative session since
the lowered response rate observed during the final ;.0 aftemath:e session wasin line withsome
of theprevious FR1000 noalternative overall response rates. However, a distinct decreasing
trend wasobserved in those sessions where thealternative wasavailable, and indeed, the
subject wasseen to be engaging in thealternative during these sessions. Thiswas a definite
change in performance from the initial FR250 baseline.
It is difficult to interpret these results, in particular the lackof reversal in those sessions when
thealternative waspresent. A number of procedural limitations complicate the understanding of
these patterns. First and foremost is thatgiven thefactthatthisexperiment wasa preliminary
single-case inquiry, the response patterns of only onesubject were observed. This makes it
difficult to tease out anarnolous experimental results fromenduring experimental effects.
Secondly, thefinalFR250reversal wasonlysixsessions long whereas the otherconditions of
this experiment were 14sessions in length. Thisfinalbaseline maynotthenbe a representative
sample of thefinalcondition producing or notproducing a reversal.
Given these procedural limitations we areleftwith two potential states. Either the lackof
reversal is procedural in nature, in that a reversal would be observed undersustained
observation, or this is anactual sequential effect ofsome sort. Ifwe assume a trueeffect is
147
represented andthatan incomplete reversal is an enduring quality of this experimental procedure
then an interesting state of affairs exists. A particular contingency whose behavioral fulfillment is
of lowprobability (that being engaging in thealternative) becomes probable or potent through a
distinct and independent contingency manipulation (increasing the referent FRfrom 250to
1000) andthispotency isenduring even afterthe inducing conditions are removed (theFRis
returned to the lower value). Thiswouldbe interesting in-and-of itselfandworthy of further study
inthat it provides anexample of generation of novel patterns of behavior andof howcertain
schedule-induced patterns maybecome locked intoenvironmental contingencies andthus
sustatined. This might be referred to as the "How canyoukeepthemdownonthe farmonce
they've seengay Paree?" effect.
Onthe otherhand these results maybe related to conditions making the referent
contingencies lesspotent or probable. Evidence e)(ists as to thedepotentiation of the
experimental conditions and arrangements for subject LM overthefinal fewdaysof this
experiment, cognitively causing himto "lose interest" in the experimental monetary
contingencies andthiswill bediscussed. Under these conditions depotentiation of the referent
contingency would leadto an increase in the behavioral alternatives present. Theevidence for
thissortof explanation of the observed patterns is twofold. Objectively, the subject reported to
the investigator during the lastweekof the experiment that he hadrecently accepted andstarted
working at a part-time jobwhere hewas reportedly drawing a substantial paycheck. Hence, the
monetary contingencies which initially hadserved to sustain performance were perhaps
influenced by access to the monetary reinforcer outside of the sessions. In otherwords the
behavioral ecOnomy "opened up" (Hursh, 1980) or inthecurrent framework a behavioral
alternative presented itselfpotentially influencing thecurrent behaviors of investigation.
Onmore of a subjective level the subject presented a picture towards theendof the
experiment of someone whowasbored withthe experiment and. in fact.verbalized this on
several occasions. Thesubject was anxiously counting thedaysuntilparticipation in the
experiment was complete andthiswasonereason thatan extension of participation wasnot
negotiated so thata 14rather than6 session finalbaseline could be run. Hence, it seems
possible that the incomplete reversal observed in thisexperiment wasrelated to motivational
issues.
A replication of thisexperiment is warranted andhopefully would clarifythe nature of the
current results. Even though the incomplete reversal observed in this experiment complicates
the interpretation of these results, the current experiment none-the-Iess provides continued
148
support for the influence of alternative contingency arrangements in the analysis of strain and
suggests the need for an awarness andanalysis of alternatives in anythorough description of
behavioral changes.
It is interesting to note thatpost-reinforcement pausing is substantial by subject LM during
thealternative present experimental conditions, particularly during the higher FA. It is possible
that theoccasion foralternatives is an important element intre occurrence of post-reinforcement
pausing by humans and otherorganisms, for thatmatter. It is alsointeresting to note that
responding to the extrakeydecreased during the schedule increase to FR1000 during those
conditions when the alternative waspresent, while increasing when noalternative was present as
during previous experiments. Hence theoccurrence of alternatives may haveimplications for
bothanalysis of postreinforcement pause patterns andfor the selection andfrequency of
patterns induced under contingency-change conditions. Further research would bebeneficial in
these areas, aswell.
Ingeneral, with regards to patterns of ancillary key responding andgiven the aforementioned
frequency difference iii extrakey responding during sessions when the alternative waspresent,
thecurrent results support conclusions drawn earlier in Experiments I, II,& III. This isparticularly
apparent when frequency of responding is examined during those sessions when noalternative
is present. As revealed in Figure 23,timeout responding temporarily increases when the
schedule is increased fromFR250to FR 1000 yet notduring the schedule reversal to the lower
value. The same pattern canbeseen to the extra keyalthough on a comparable basis, extra key
response frequency is substantially higher leading oneto suspect that the timeout contingency
actually restricts response rate. Additional evidence for this exists when examining thepreand
posttimeoutcontingency introduction data. During sessions 1-3timeout responding occurs in
the600-1000 response frequency range while theposttimeout contingency range (sessions
4-38) is 0-80. Hence, patterns identical to those reported earlier areseen.
149
REFERENCES
Appel, J.B. (1963). Aversive aspects of a schedule of positive reinforcement. Journalof theExperimental Analysis of Behavior, 6, 423-428.
Azrin, N.H. (1958). Some effects of noise on human behavior. Journal of the ExperimentalAnalysis of Behavior, 1, 183-200.
Azrin, N.H. (1961). Time-out from positive reinforcement. SCience, 133,382-383.
Azrin, N.H., Hutchinson, R.R., & Hake, D.F. (1966). Extinction-induced aggression. JournaloftheExperimental Analysis of Behavior, 9, 191-204.
Barrett, J.E. (1976). Conjunctive schedules of reinforcement. III:A fixed-interval adjustingfixed-ratio schedule. Journal of theExperimental Analysis of Behavior, 25, 157·164.
Boren, J.J., Moershbaecher, J.M., & Whyte, AA. Variability of response location on fixed-ratioandfixed~interval schedules of reinforcement. Journalof theExperimental Analysis ofBehavior, 30, 63-67.
Brady, J.V.,& Hunt, H.F. (1955). An experimental approach to theanaysis of emotionalbehavior. Journal of Psychology, 40, 313-324.
Brown, T.G., &Flory, R.K. (1972). Schedule-induced escape fromfixed-interval reinforcement.Journal of the Experimental Analysis of Behavior, 17, 395-403.
Buskit, W.F., Miller, Jr.,H.L., & Bennett, R.H. (1980). Fixed-interval performance in humans:Sensitivity to temporal parameters when food is the reinforcer. The Psychological Record,3D, 111-121.
Buskit, W.F., Miller, Jr., H.L., & Bennett, R.H. (1981). Effects of instructional constraints onhuman fixed-interval performance. Journalof the Experimental Analysis of Behavior, 35,217-225.
Byrd, L.D. (1973). Effects of d-amphetamine onschedule-eontrolled keypressing anddrinkingin thechimpanzee. Journal uf Pharmacology and Experimental Therapeutics, 185,633·641.
Capehart, G.W., Eckerman, D.A., Guilkey, M.& StlJlI, R.L. (1980). A comparison of ratio andinterval reinforcement schedules withcomparable inter-reinforcement times. Journal of theExperimental Analysis of Behavior, 34, 61-76.
Catania, AC. (Ed.). (1968). Contemporary research Inoperant behavior. Glenview, IL:Scott,Forseman andCo.
Catania, AC. (1969). Concurrent performances: Inhibition of one response by reinforcement ofanother. Journal of the Experimental Analysis ofBehavior, 12, 731-744.
150
Collier, G.H., Hirsh, E., & Hamlin, P.H. (1972). Theecological determinants of reinforcement inthe rat. Physiology andBehavior, 9, 705-716.
Dardano, J.F. (1968). Reversal of preference underprogressive-ratio schedules of punishment.Journal of the Experimental Analysis of Behavior, 11, 133-146.
Dardano, J.F. (1973). Self-imposedtimeouts under increasing response requirements. Journalof the Experimental Analysis of Behavior, 19,269-287.
Dardano, J.F. (1974). Response-produced timeouts undera progressive-ratio schedule withareset option. Joumalof the Experimental Analysis of Behavior, 22, 103-113.
Dardano, J.F., & Sauerbrunn, D. (1964). Selective punishment of concurrent progressive-ratioperformance. Journal of the Experimental Analysis of Behavior, 7, 51-65.
Dews, P.B. (1958). Studies onbehavior. III:Stimulant actions of methamphetamine. JournalofPhamacologyand Experimental Therapeutics, 122, 137-147.
Duvinsky, J.D.,& Poppen R. (1982). Human performance on conjunctive fixed-intervalfixed-ratio schedules. Journal of the Experimental Analysis of Behavior, 37, 243-250.
Ellis, N.R., Barnett, C.D., & Pryer, M.W. (1960). Operant behavior in mental defectives:Exploratory studies. Journal of the Experimental Analysis of Behavior, 3, 63-69.
Falk, J.L. (1961). Production of polydipsia in normal ratsby an intermittent foodschedule.Science, 133,195-196.
Falk, J.L. (1964). Studies on schedule-induced polydipsia. InM.J. Wayner (Ed.), Thirst: Firstintemational symposium on thirstin the regulation of bodywater. NewYork: PergamonPress.
Falk, J.L. (1981). Theenvironmental generation of excessive bebavlor. In S.J. Mule, (ed.),Behavior in excess. New York: Free Press.
Falk, J.L., Sampson, H.H., & Winger, G. (1972). Behavioral maintenance of highconcentrationsc)fblood ethanol andphysical dependence in the rat. Science, 177, 811-813.
Fallon, J.H., Jr., Allen, J.D., &Butler, J.A. (1979). Assessment of adjunctive behavior in humansusing a stngent control procedure. Physiology andBehavior, 22, 1089-1092.
Felton, M.,& Lyon, D.O. (1966). Thepost-reinforcement pause. Journalof the ExperimentalAnalysis of Behavior, 3,131-134.
Ferster, C.B. (1957). Withdrawal of positive reinforcement as punishment. Science, 126,509.
Ferster, C.B. (1960). Intermittent reinforcement of matching to sample inthepigeon. Journal oftheExperimental Analysis of Behavior, 9. 259-272.
Ferster, C.B. (1973). A functional analysis of depression. American Psychologisf, 28, 15-45.
151
Fersler, C.B.,Appel, J.B.,& Hiss, R.A. (1962). Theeffects of drugs on a fixed-ratio performancesuppressed by a pre-timeout stimulus. Journal of the Experimental Analysis of Behavior, 5,73-88.
Ferster, C.B., Culbertson, S., & Boren, M.C.P. (1975). Behavior principles, (2nd Ed.).Englewood Cliffs, N.J.: Prentice-Hall.
Ferster, C.B., & Peele, D.B. (1980). The contribution of an added counter to a fixed-ratioschedule. Journal of the Experimental Analysis of Behavior, 34,93-105.
Ferster, C.B.,& Skinner, B.F. (1957). SChedules of reinforcement. NewYork:Appleton-Century-Crotts.
Findley, J.D.,& Brady, J.V. (1965). Facilitation of large ratio performance by useofconditionedreinforcement. Journal of theExperimental Analysis of Behavior, 8, 125-129.
Frazier, T.W., & Bitetto, V.E. (1969). The control of human vigilance byconcurrent schedules.Journal of theExperimental Analysis of Behavior, 12,591-600.
Frederiksen, L.W., & Peterson, G.L. (1974). Schedule-induced aggression in nursery schoolchildren. The Psychological Record, 24,343-351.
Gentry, W.O. (1968). Fixed-ratio schedule-induced aggression. Joumal of the ExperimentalAnalysis of Behavior, 11, 813-817.
Goldiamond, I. (1974;. Toward a constructional approach to social problems: Ethical andconstitutional issues raised by applied behavior analysis. Behaviorism, 2, 1-85.
Goldiamond, I. (1975) Alternative setsas a framework forbehavioral formulations andresearch.Behavion'sm, 3, 49-86.
Goldiamond,l. (1984). Training parent trainers andethicists in nonlinear analysis of behavior. InR.F. Dangle & R.A. Polster, (Eds.), Parent training: Foundations of research andpractice.New York: Guilford Press.
Green, R.J., Sanders, R.M., & Squire, R.W. (1959). SChedules of reinforcement anddiscrimination learning. Journal of the Experimental Analysis of Behavior, 2, 293-299.
Harzem, P. (1984). Experimental analysis of individual differences andpersonality. JoumaloftheExperimental Analysis of Behavior, 42,385-395.
Herman, R.L., & Azrin, N.H. (1964). Punishment by noise inan alternative response situation.Journal of the Experimental Analysis of Behavior, 7, 185-188.
Hersen, M.,&Barlow, D.H. (1976). Single case experimental designs: Strategies forstudyingbehaviorchange. NewYork: Pergamon Press.
152
Hodos, W.•&Kallman. G. (1963). Effects of increment size andreinforcer volume onprogressive-ratio performance. Journalof the Experimental Analysis of Behavior, 6,387-392.
Hodes, W.•& Trumble. G.H. (1967). Strategies of schecute preference in chimpanzees. Journalof the Experimental Analysis of Behavior, 10.503-514.
Holland. J.G. (1958). Human vigilance. SCience. 128, 61-67.
Holz, W.C.•Azrin, N.H.• & Allyon. T. (1963). Elimination of behavior of mental patients byresponse-produced extinction. Journal of the Experimental Analysis of Behavior, 6,407-412.
Hursh. S.R. (1980). Economic concepts for the analysis of behavior. Journal of theExperimental Analysis of Behavior, 34, 219·238.
Hutchinson, R.R.. & Azrin, N.H. (1961). Conditioning of mental hospital patients to fixed-ratioschedules of reinforcement. Journal of the Experimental Analysis of Behavior, 4, 87-95.
Hutchinson. R.R., Azrin. N.H.•& Hunt, G.M. (1968). Attack produced by inttermittentreinforcement of a concurrent operant response. Journal of the Experimental Analysis ofBehavior, 11,489-495.
Kachanoff, R.. Leveille. R., McLelland. J.P.•& Wayner, M.J. (1973). SChedule-induced behaviorin humans. Physiology andBehavior, 11.395·398.
Kelly. J.F.• & Hake. D.F. (1980). Anextinction-induced increase in an aggressive response withhumans. Journalof the Experimental Analysis ofBaha'ofioi, 14, 153-164.
King, G. (1974). Wheel running intherat induced by afixed timepresentation of water. AnimalLearning and Behavior. 2, 325-328.
Laursen. A.M. (1972). Post-reinforcement pauses and response rate of monkeys ona two-handFRschedule. Journalof the Experimental Analysis cfBehavior, 17,85·94.
Lindsley,O.R. (1956). Operant conditioning methods applied to research in chronicschizophrenia. Psychiatric Research Reports of theAmerican Psychiatric Association. 5.118·139. 140-153.
Lindsley. O.R. (1959). Reduction in rate of vocal psychotic symptoms by differential positivereinforcement. Journal of the Experimental Analysis of Behavior, 2, 269.
Lindsley,O.R. (1960). Characteristics of the behavior of chronic psychotics as revealed byfree-operant conditioning methods. Diseases of the Nervous System Monograph, '-1.61-78.
Long, E.R. (1963). Chained andtandem scheduling withchildren. Journal of the ExperimentalAnalysis of Behavior, 6, 459-472.
153
Long, E.R., Hammack, J.T., May. F., & Campbell, B.J. (1958). Intermittent reinforcement ofoperant behavior in children. Journalof the Experimental Analysis of Behavior, 1. 315-339.
Lowe. C.F. (1979). Determinants of human operant behavior. InM.D. Zeiler & P.Harzem, (Eds.),Advances inanalysis of behavior (vol. 1):Reinforcement andtheorganization of behavior.Chichester, U.K.: Wiley. 159-192.
Markle. S.M., & Tiemann. P.W. (1969). Really understanding concepts: Orin frumious pursuitofthejabberwock. Champaign, IL: Stipes Publishing Co.
Matthews, B.A., Shimoff. E., Catania. A.C., & Sagvolden, T. (19n). Uninstru{,~edhumanresponding: Sensitivity to ratio and interval contingencies. Journal of theExperimentalAnalysis of Behavior, 27.453-467.
McMillan. D.E. (1969). Effects of d-amphetamine on performance under several parameters ofmultiple FR FIschedules. Journal of Pharmacology andExperimental Therapeutics, 167,26-33.
McMillan, J.C. (1971). Percentage reinforcement of fixed-ratio andvariable-interval performance.Journal of theExperimental Analysis of Behavior, 15, 297-302.
Medelson. J.• & Chillag, D. (1970). Schedule-induced air licking in rats. Physiology andBehavior, 5, 535-537.
Morgan, M.J. (1972). FRperformance underconditions of delayed reinforcement. Journal oftheExperimental Analysis of Behavior. 17, 95-98.
Morse. W.H.• & Herrnstein, R.J. (1956). Effects of drugs oncharacteristics of behaviormaintained by complex schedules of intermittent positive reinforcement. Annals of theNewYork Academy of Sciences, 65. 303-317.
Mull9r, P.C., Crow. R.E.•& Cheney, C.D. (1979). Schedule-induced locomotor activity inhumans. Journal of the Experimental Analysis of Behavior, 31, 83·90.
Nevin, J.A., & Reynolds, G.S., (Eds.), (1973). The studyof behavior. Glenview. II:Scott.Forseman and Co.
Notterman, J.M.•& Mintz, D.E. (1965). Dynamics of response. New York: John Wiley.
oet, T.P.S., Singer. G.•Jeffreys, D.,Lang, W., & Latiff. A. (1978). SChedule induced selfinjection of nicotine. heroin andmethadone by naive animals. In EC. Colpeart & J.A.Rosencrans (Eds.), Stimulus properties of drugs: Ten years ofprogress. Amsterdam:Elsevier/North-Holland Biomedical Press.
Paer, J.J. (1975). Implications of the matching lawfor ratio respor.ding. Journal of theExperimental Analysis of Behavior. 23. 139-140.
Peden, B.F.•& Timberlake, W. (1984). Effects of reward magnitude onkeypecking andeatingby pigeons in a closed economy. The Psychological Record, 34,397-415.
154
Poppen, R. (1982). Human fixed-interval performance with concurrently programmedschedules: A parametric analysis. Journal of the Experimental Analysis of Behavior, 37,251-266.
Porter, J.H., Brown, RT., & Goldsmith, P.A. (1982). Adjunctive behavior in children on fixedinterval food reinforcement schedules. Physiology andBehavior, 28, 609-612.
Powell, RW. (1968). Theeffect of small sequential changes in FRsize upon the PRP. Journalof the Experimental Analysis of Behavior, 11, 589-593.
Powell, RW. (1969). The effect of reinforcement magnitude uponresponding underFRschedules. Journal of theExperimental Analysis of Behavior, 12, 605-608.
Priddle-Higson, P.J., Lowe, C.F., & Harzem, P. (1976). Aftereffects of reinforcement onvariable-ratio schedules. Journal of the Experimental Analysis of Behavior, 25, 347-354.
Rachlin, H. (1978). A molartheory of reinforcement schedules. Journal of the ExperimentalAnalysis of Behavior, 30, 345-360.
Rayfield, F.,Segal, M.,& Goldiamond, I. (1982). SChedule-induced defecation. Journal of theExperimental Analysis of Behavior, 38,19-34.
Reese, E.P. (1978). Human operant behavior: Ana/ysis andapplication. Debuque, Iowa: Wm. C.Brown Co.
Repp, A.C., & Deitz, S.M. (1975). A comparison of fixed-ratio andvariable-ratio token-productionschedules with human subjects. ThePsychological Record, 25, 13i-137.
Reynolds, G.S. (1975). A primerof operant conditioning. Glenview, IL:Scott, Forseman andCo.
Roper, T.J. (1978). Diversity and substitutibility of adjunctive activities underfixed-intervalschedules of food reinforcement. Journal of the Experimental Analysis of Behavior, 30,83-96.
Roper, T.J. (1981). What is meant by the term"schedule-induced," andhowgeneral is scheduleinduction? AnimalLearning andBehavior, 9, 433-440.
Sanders, A.M. (1969). Concurrent fixed-ratio fixed-interval performances in adulthumansubjects. Journal of the Experimental Analysis of Behavior, 12, 601-604.
Sidman, M. (1960). Tactics of scientific research. New York: Basic Books, Inc.
Sidman, M.,& Stebbins, W.C. (1954). Satiation effects underFRschedules of reinforcement.Journalof Comparative and Physiological Psychology, 47, 114-116.
Skinner, B.F. (1938). The behaviorof organisms. NewYork: Appleton-Centurey-Crofts.
Skinner, B.F. (1953). Science andhuman behavior. NewYork: McMillan.
155
Staats, A.W.,. _!'j, J.R., Osborne, J.G, Quinn,W.D., & Minke, K.A. (1963). The useof chainschedules in thestudyof reinforcement variables in a reading task. Unpublished manuscript(Technical Report No.25 (1963). Office of Naval Research. Contract Nonr-2794(02)).
Sulzer-Azaroff, B., & Mayer, G.R. (1977). Applying behavior-analysis procedures withchildrenandyouth. NewYork: Holt, Reinhart andWinston.
Thomas, J.R. (1974). Changes in progressive-ratio performance underincreased pressures ofair. Journal of the Experimental Analysis of Behavior, 21, 553-562.
Thompson, D.M. (1964). Escape fromdiscriminative stimulus associated with fixed-ratioreinforcement. Journalof theExperimental Analysis of Behavior, 7, 1-8.
Thompson, D.M. (1965). Timeout from FR reinforcement: A systematic replication.Psychonomic Science, 2, 109-110.
Thompson, D.M., & Moershbaecher, J.M. (1980). Effects of d-amphetamine andcocaine onstrained ratio behavior in a repeated-acquisition task. Journalof theExperimental Analysis ofBehavior, 33,141-148.
Villareal, J. (1967, April). Schedule-inducedpica. Paperpresented at meeting of the EasternPsychological Association, Boston, MA.
Wallace, M., &Singer, C. Adjunctive behavior and smoking induced by a maze solving schedulein humans. Physiology andBehavior, 17, 849-852.
Wasik, B.H. (1969). The effects of fixed ratio andcontingent time on human lever-pressingbehavior. The Psychological Record, 19, 95-104.
Weiner, H. (1964a). Response costeffects during extinction following fixed-intervalreinforcement in humans. Journal of theExperimental Analysis of Behavior, 7, 333-335.
Weiner, H. (1964b). Conditioning history andhuman fixed-interval performance. Journalof theExperimental Analysis of Behavior, 7, 383-385.
Weiner, H. (1966). Preference andswitching underratio contingencies with humans.Psychological Reports, 18,239-246.
Weiner, H. (1967). Costto payoff contributions to preference underfixed-ratio contingencieswith humans. Psychological Reports, 20,31-35.
Weiner, H. (1970). Instructional control of human operant responding during extinction followingfixed-ratio conditioning. Journal of the Experimental Analysis of Behavior, 13,391-394.
Weiner, H. (1972). Controlling human fixed·interval responding with fixed ratio responding ordifferential reinforcement of low rate responding in mixed schedules. Psychonomic Science,26,191-192.
156
Weiner, H. (19n). Anoperant analysis of human ahruistic responding. Journal of theExperimental Analysis of Benavioi, 27.515-528.
Weiner. H. (1979). Some observations on apathetic operant responding 01 chronicschizophrenics. Perceptual andMotorSkills, 48. 1083-1090.
Weiner, H. (1981a). Base-line muscle tensions andmotoric responding of schizophrenics.British Journalof Clinical Psychology, 20. 25-33.
Weiner, H. (1981b). Contributions of reinforcement schedule histories to our understanding ofdrugeffects in human subjects. InT. Thompson & C.E. Johanson (Eds.). Behavioralpharmacology of human drugdependence. National Institute on Drug Abuse ResearchMonograph 37. Washington. D.C.: U.S. Government Printing Office.
Weiner, H. (1982). Histories of response omission and human operant behavior underafixed-ratio schedule of reinforcement. The Psychological Record, 32, 409-434.
Weiner, H. (1983). Some thoughts on discrepant human-animal performances underschedulesof reinforcement. The Psychological Record, 33, 521-532.
Weissman. N.W.•& Crossman. E.K. (1966). A comparison of two types of extinction followingfixed-ratio training. Journal of the Experimental Analysis of Behavior, 9. 41-46.
Whaley. D.L.. & Malott, R.W. (1971). Elementary principles of behavior. Englewood Cliffs. NJ:Prentice-Hall.
Witters. D.R., & Bachrach. A.J. (1965). Theeffects of competing contingencies uponfixed ratiobaselines in a smallgroup situation. The Psychological Record, 15, 103-110.
Woods. K.A.. Happ, A.J.•& Adams. C.K. (1983). Ratio responding as a function of concurrentavoidance schedules, yoked shocks andratio value. Journal of the Experimental Analysis ofBehavior, 39. 449-456.
Yobum, B.C.•& Cohen, P.C. (1979). Schedule-induced attack on a pictorial target in feralpigeons (Columbia livia). Bulletin of thePsychonomic Society, 13. 7-8.
Zimmerman, J.•& Ferster. C.B. (1964). Some notes on timeout from reinforcement. JoumaloftheExperimental Analysis of Behavior. 7,13-19.
Related Documents
