Various uses of minicomputers in psychology

JOSEPH B. SIDOWSKIUniversity of South Florida, Tampa, Florida 33620

Various uses of minicomputers in psychology

and using systems of differingmanufacture. As a first effort, such ahybrid composition would appearboth desirable and necessary. When wecomplete the conference, however, wemust ask some important questionsconceming possible future activities ofthis sort. Should future conferences beconsidered? Is a more formal orspecialized organizational structureneeded? If additional conferences areneeded, to whom should they bedirected, and at what level? I do notthink we should ignore thesemethodological questions, and I inviteyour suggestions.

Psychologists who use on-linecomputers in many ways have a specialobligation to their colleagues insideand outside of psychology, whichshould not be ignored. While the costof computer use has decreased inrecent years, the expenditures ofhigher education for computingservices have increased sixfold in thelast 6 years (Mosmann & Stefferud,1971). The cost of computer use willcontinue to decrease. Continuedreductions in the price of new systems,the increasing availability of usedcomputer systems, the expectedavailability of govemment surplusequipment, and software advancesshould insure this. Current universityand grant-agency economics will not,however, promote increased spendingto accommodate the more widespreaduse of on-line computer systems. Wedo have an obligation to ourcolleagues, naive and sophisticated, togenerate methods whereby efficientand appropriate applications of on-linecomputer systems are assured. In thepast, we have been mainly dependentu po n the manufacturers forapplications information and advice.As many of us will testify, there islittle to assure that these traditionalsources need be either satisfactory orreliable. We must consider thedevelopment of methods ofinformation exchange which areindependent of the manufacturers orof chance personal acquaintances.Equally important, perhaps, is theneed to develop software andhardware standards and methodstailored to on-line computerapplications in psychology butindependent of specific manufactureror experimenter idiosyncrasies.

Many of the problems andobligations mentioned above suggest aneed to improve communication andeducational opportunities for ourselvesand our colleagues. We must not forgetour obligations to other members ofthe scholarly world. In general,administrators and undergraduatesshare a common problem: little directcontact with computers and computerapplications (Scully, 1971). If on-line

computer systems are to enjoy evenwider popularity in the future, wemust consider the possibility thaton-line computer applications inpsychology are so specialized andspecialty-dependent that theappropriate education and trainingmay not be provided by specialistsfrom other disciplines. Just as wecannot depend upon manufacturersfor information and training, we alsocannot depend upon other disciplinesto assume these duties.

In summary, I hope that thisconference will be noted not only as afirst gathering of psychologists whouse on-line computers, but also as thestart of an attempt to organize effortstowards improving the overalldissemination of informationcon c e r ning on-line computerapplications in psychology. We mustconsider new and better ways of

I am not at all certain that the titleof this presentation adequatelydescribes what I have to say, for thereare various ways to look at the"Various uses of minicomputers inpsychology." A better title might havebeen "Who else uses these computers,so you'll know whom to contact whenthe lights go out" or "The variousmeans available to psychologists forobtaining information provided byother computer users from all sorts ofareas and installations to help themmake the damn things work in thelaboratory." Whatever, it is difficult toimprove on the computer-usedescriptions provided by Uttal'svolume on real-time computers (1968)and those of many others who foryears have attempted to educate usconcerning the ways in which smallcomputing machines might be usedboth on-line and in a real-timepro c e s s -con tr 0 I sen s e. Th eforthcoming volume edited by BemieWeiss (Digital computers in thebehavioral laboratory) will update theattempt (1972). So, in the timeavailable, I will attempt to skipthrough varied bits of informationconcerning minicomputer uses andhope, in the process, to communicatea little of something to each of you inthis diversified audience.

Before going on, however, I wouldlike to define minicomputer and willdo so in spite of some disagreement

assuring and expanding theappropriate use of computers in ourdiscipline.

REFERENCESARMINGTON. J. C.. TEPAS. D. I ..

KROPFL. w. J.. &. HENGST. W. H.Summated retinal potentials. In E. P.Home and M. A. Whitcomb (Bds.), Visionresearch reports. Washington. D.C:National Academy of Sciences-NationalResearch Council. 1960.

BOWDEN, L. The language of computers.American Scientist. 1970, 58, 43-53.

GREEN, B. F. Digital computers inresearch. New York: McGraw-Hill. 1963.

MICHAELS. P. Remote terminals replaceuniversity computer center. ElectronicSystems Engineering. May/June 1970.

MOSMANN. C.• &. STEFFERUD. E.Campus computing management.Datamation. March 1971, 20.23.

SCULLY, M. G. Computers. big in research.little-used by undergraduates. TheChronicle of Higher Education, October12. 1971. 4-5.

within the industry conceming anexact definition. Generally, themachines are considered smallerversions of the large scale,general-purpose computers, organizedin somewhat the same mannerin ternally, allowing for similarprogramming techniques and for thesame types of peripherals. A miniincludes a central processor withgeneral-purpose registers, core, and aset of instructions, and accepts andstores a program that the user canalter. The memory most often has aminimum of 4,096 words, with wordlengths varying from 8 to 24 bits,depending on the machine and theapplication. The user can inputcommands to the system, and hereceives results and instructions fromit. Finally, it is small in size and costsno more than $25,000-$30,000 for thebasic unit; the cost at the lower end ofthe scale can be less than $3,000(Sidowski, 1970). Beyond the aboverequirements, modem minis may haveall sorts of features which increasecapabilities and costs tremendously.For many process-control situationsfound in the laboratory, anoff-the-shelf basic minicomputer costslittle more than a specially designedlogic module controller. But there islittle justification for using it where acheap hard-wired controller can do thejob.

Obviously, in order to test

Behav. Res. Meth. &: Instru., 1972, Vol. 4 (2) 43

Fig. 1. The figure illustrates an incomplete operating system in need of ameans for communication between E and computer and between the computerand experimental apparatus.

COMPUTER

Boneau; a description of amessage-switching system for researchon human behavior in game situationsby a nonpsychologist (Andreae) fromthe Lawrence Radiation Laboratories;and Ragsdale's description of theUniversity of Pittsburgh'scomputer-assisted instructional andlearning research laboratory.

Skipping ahead to the recent 1971Spring DECUS Conference, we findpsychologists contributingsophisticated software information.Walker and Snapper, e.g., provide asoftware system to control 10different reinforcement schedules,including stimulus generation andresponse recording. Stout, of theMathematical Psychology Program atthe University of Michigan, describesIBIOS, which allows E-computerinteraction on-line while the S is beingrun in experiments involving learning,perception, cognition, and choicebehavior. The "Son of Zoro" is listedby Lewis and Osgood as asecond-generation multiprogrammingsystem for psychological research,following up an earlier publication onZoro. And Grabowy and Ellinwoodcover an on-line system for analyzingbursts of olfactory spindle activity inEEG.

A year earlier, the 1970 SpringDECUS meetings supplied us with atleast nine useful reports includingseveral in the physiological psychologyareas. One researcher (Cox), using aPDP-8/1 computer with 8K memory,provides a program which simulatesthe firing of 64 model neuronsconnected to form a functional nervenet with provisions for studying thepossible mechanisms of learning;another describes a PDP-12 system forthe on-line acquisition of heart-ratedata (Kerr et al); other reports coverthe study of EEG data reduction inthree states bordering on light sleep(Rodgers), an application of fastFourier transform in the automaticdetection of sleep spindles (Vo-Ngoc &Poussart), and the use of the PDP-8/1as a computer of average transients(Scott & Dzendolet).

So, from the mid-1960s through1971, psychologists have reported onthe use of the minicomputer in allsorts of applications. Neural, sensory,and muscle activity have beenmonitored, various models have beensimulated, data have been analyzed,students have been instructed in a CAl(computer-aided instruction) real-timesense, and human and animal Ss havebeen run in studies in which theminicomputer controlled stimuluspresentations, sensed and recordedresponses, and, in a closed-loopmanner, changed requirements for theSs as a function of the responses beingmade or the criterion reached. In some

INVESTIGATOR

in psychology and the involvement ofmany of our colleagues in thedevelopment of its technology.

A review of the DECUS Proceedingsbeginning in the early 19608 throughthe most recent, plus theDECUSCOPE publications and variousprofessional journals, provide abibliographical review which mirrorsthe many uses and users ofminicomputers in the various areas ofpsychology. A complete report of thisreview will be concluded shortly, but Iwould like to convey some initial datain my talk this morning.

Looking back a mere 6 years to theSpring 1965 DECUS ConferenceProceedings, I found a total of 16reports with one article by Brazael andBooth detailing the use of a high-speedCRT display system for man-computerinformation processing. A secondarticle by Savage introduced a PDP-4monitoring system (coincidentallynamed the Savage system), designedand written for the Center forCognitive Studies at Harvard with thehelp of Donald Norman, who modifiedDEC's package of two-Teletypeprogram routines and added someuseful routines for page line andspacing control. The other 14 articleshad little to offer the psychologist. Asan aside, it is interesting to note a fewof the questions asked at thatconference of only 6 years ago: "Arethere going to be any moreprogramming classes on PDP-1?" and"Can you tell me the selling price of aused PDP·1?" The answer to the latterquestion was that a reconditionedPDP·1 would cost in the neighborhoodof $50,000-$60,000 with 4K memory.Recall that in 1958, the machineoriginally sold for $120,000. Ten yearsafter it was introduced, its morecapable grandson, the PDP-8/L soldfor $8,500.

A review of the Fall DECUSProceedings of the same year (1965)resulted in three additional articles ofinterest: A general discussion coveringthe application of the LINC computerto operant conditioning research by

SUBJECTSTATION

experimental hypotheses in thelaboratory, the E must have a handleon his research problem and heordinarily would possess the necessarysupportive standard equipment, suchas the Skinner box, T-scope, or slideprojector. For the fortunate, ofcourse, there is a laboratory computer.Unfortunately, the fortunate are facedwith two problems which requireattention in order to make a researchsystem work, namely, the programs,which act as the means ofcommunication between the E and thecomputer, and the interface, which isthe means of connecting to andcommunicating between the computerand the required experimentalequipment. It is hard to improve on aportrayal of the connection illustratedat the 1967 APA meetings by LouisSiegel (see Figs. 1 and 2 for a revisedversion).

Now that I have provided some ofthe definitional bounds ofminicomputers, how are they used?When equipment monies were free,loose, and available, small computerswere purchased by some as technicallyadvanced laboratory decorations or,like the big pink Cadillac, as statussymbols indicating the great powerand strength of the researcher. Thenthere were those who purchased themachines in place of the less fancydesk calculators. But at the beginningof small digital computer use inpsychology, and since, there have beenothers. In 1963, as part of the LINCEvaluation Program (and LINC, by theway, is an acronym for LaboratoryInstrument Computer), severalscientists, including Bernard Weiss,Don Blough, and Alan Boneau, wereselected to determine whether or notpsychologists and biomedicalresearchers could learn to use andintegrate the small 2K memorycomputer into their laboratoryresearch endeavors. Since then, oneneed merely delve through the variousProceedings of the Digital EquipmentComputer Users Society to note thevarious applications of minicomputers

44 Behav. Res. Meth. & Instru., 1972, Vol. 4 (2)

Fig. 2. A complete system with program and interface.

lines, both 16-bit-word machines witheasy access for interfaclng.! TheNOVA is used in a number oflaboratories, and it is included in theLehigh Valley Electronics INTERACTpackage. At the Memorial Universityin Newfoundland, for example, theDepartment of Psychology utilizes an8K NOVA with a number ofperipherals, including a cassette tapeand specially built units with 80 bitsof output logic, 48 bits of input logic,and three interrupt nets. The interfaceis designed to produce permanent levelchanges and one-shot signals directlyinto BRS solid-state modular logicsystems. This system is used primarilyto run undergraduate animal andhuman experimental labs on-line;programming is accomplished inBASIC. At the Norwalk Hospital,Norwalk, Connecticut, two of thenewer NOVA 800 CPUs are employed,each with 8K core memory, a 256Kword disk, high-speed paper-tapereader, digital plotter, two Tektronixdigital displays, a Teletype, and anIBM Selectric typewriter. One machineis being used by Nair and Fischer toautomate the pulmonary functionlaboratory of the Department ofRespiratory Technology in order tostudy exercise effects.

Later today, P. Van Gelder, of thePsychology Department at theUniversity of Connecticut, willdescribe his application of an 8KNOVA system to information­processing-type perceptual work withreaction time as the most usefuldependent variable. In data analysis,this system has the capacity forautomatically displaying thereaction-time frequency distribution.

Two SUPERNOVAs, each with 12Kmemory core and 256K disks, areinstalled in the Computer Researchfacility of the Loma Linda Univeristyin California, along with a NOVA,800with 8K. The latter system costsapproximately $10,000, while each ofthe complete SUPERNOVA systemsrun $30,000 and $42,000,

r---. .---. .:.

I I I IC II I

-./'N- I II n I ••••• I P II t I ••••• I r II e ••••• 0 II r I ••••• (IJ Ir l f I ••••• I r I

-01 ~ I ~ I a II e I I m II I I I1.___1 I I----

SUBJECT COMPUTER INVESTIGATOR

OPERATING SYSTEM

use of a PDP-9 in a psychometriclaboratory (University of NorthCarolina) with the research falling intofour categories: individual decisionbehavior, group decision behavior,problem solving, and multidimensionalscaling. Aiken will discuss a 4KPDP-8L with disk storage and magtape. The latter system is located inMillward's Brown University facilityand is used to study humaninformation processing and problemsolving, in particular to helpinvestigate strategies for the solutionof concept identification problems.Whitmore's PDP-12 system (Universityof Vermont) is used to analyze EEGdata in real time from recorded dataproviding a graphical output and awritten summary of activity andfrequency changes. In G. A.Van Gelder's operation at Iowa State,a 4K LINC-8 with disk and mag tape isused to analyze EEG in real time, tolog data from two FI schedulessimultaneously, and to log data from aLab-K controller. J. L. Young, atStony Brook (SUNY), employs a DECPDP-8L computer in a system built byInfotec; this system is used also byDoll. Verbal learning experiments arerun in which temporal and proceduralvariables are varied within the same S'strials.

Obviously, computer systems of theDigital Equipment Corporation are themost widely used in the field and arethe best known. There are, however, anumber of other machines installed inpsychological laboratories whichwarrant attention also. Since several ofthese will be covered in talks thisafternoon, and in this evening'sworkshop meetings, I would like tospend a few moments discussing theseveral minicomputers noted on yourprogram, with the exception of theDEC and IBM machines. Specs forthese systems are printed in thehandouts (refer to Notes 1, 2, 3, and4).

First, note the low-cost DataGeneral NOVA and SUPERNOVA

cases, updated interval histograms havebeen displayed. In summary, thelaboratory minicomputer can be usedfor data acquisition and data logging,and it can perform process monitoringand control. It can display data afteracquisition, and it can store andretrieve it. Data reduction can becarried out by preprocessing and thentransmitted to a larger computer forfurther processing, or it can be fullyprocessed at home. And it can be usedin other ways.

A number of laboratories anddepartments have turned to thepurchase and use of time-sharingsystem; for the incorporation ofmultiple users with variousexperimental problems. The pros andcons of using dedicated computers vstime-sharing systems, of course, havebeen discussed by others and need notbe drawn out here. It is important tonote, however, that there areadvantages and disadvantagesassociated with either approach. Keepin mind throughout this discussionthat the more required of thecomputer (time shared or a dedicatedstand-alone), the more expensive thesystem becomes because of the needfor added memory, peripherals, etc.

Now there are over 100minicomputer systems available(Sidowski, 1970). Yet an inspection ofyour conference program indicatesthat the overwhelming majority ofmachines listed are marketed by theDigital Equipment Corporation (DEC).For example, W. Siegel's departmentat Western Ontario uses a basic 4KPDP-8S for operant studies; a secondmachine, the PDP-12 with 12K of coreand peripherals, is largely devoted toperceptual experiments, although ithas been used to study attitudechange, personality characteristics, andhuman memory. Doll reports that theDepartment at Stony Brook (SUNY)uses a 4K PDP-8L for on-lineautomated verbal associate learning,recognition memory, andreaction-time studies; a second PDP-8Lsystem with 8K core has a Carouselrandom-access slide-projector systemconnected to the output lines and sixresponse boxes inputting. Watson, atthe Centrallnstitute for the Deaf, usesthe mini DEC mini, the PDP-8S, in apsychoacoustic facility to control thedelivery of sounds, varying infrequency level and duration, tolisteners in four booths, eachcontaining six buffered response lines.A meter is interfaced to measurestimulus frequencies and durations,and a D/A converter is sometimes usedto produce acoustic waveforms.Problem; investigated include masking,pattern recognition, time intensitytrading, and so on. AttQnight'sworkshop, Wallsten will descri()e the

Behav. Res. Meth. & Instru., 1972, Vol. 4 (2) 45

respectively. The $42,000 network isbeing used for monitoringphysiological signals from human Ss.The $30,000 SUPERNOVA system isused as a time-sharing unit, handlingup to 17 simultaneous users with theBASIC language. The NOVA 800 isemployed both as a laboratory and asa teaching device. Plans are to use theinexpensive 800 on a time-sharingbasis and to ultimately connect it tothe IBM 360 to handle themultiplexing of up to 32 graphicsterminals. These terminals will be usedas instructional units for students,providing both textual and pictorialrepresentation of information. Thisrelatively small system will have a diskand possibly 4K of core added shortly.

A 16K NOVA unit with a diskoperating system is being used in theDartmouth Medical School forprocessing EKG signals; at thePsychiatric Clinic of BEL-AIR inSwitzerland, a 4K NOVA system andseven-track analog tape recorder areutilized to diagnose and analyze sleeppatterns. Computer analyses arecarried out on data from eyemovements, heart rate, andrespiration. In the Department ofBiochemistry and Biophysics at TexasA & M, Professor Meyer is using aNOVA to drive a display system thatcould be described asthree-dimensional.

At the University of Notre Dame,H. Each, of the Biology Department, isemploying the newer NOVA 1200(priced with 4K 16-bit words and1,200-nsec memory cycle time at$5,450) with 8K memory in an on-lineanalysis of electrophysiological data(control of insect flight). A secondresearcher uses the system off-line toevaluate and plot the daily activity ofmosquitoes from data stored on papertape.

A little-known machine in thebehavioral sciences is the IMLACPDS-1, a stand-alone, programmable,display computer system.2 The PDS-1comprises a 14·in. CRT display screen,a processor, and a solid-state keyboardwith software-programmed functions.The CRT is refreshed from localmemory at 40 frames/sec normallyand can display about 1,200 charactersor 500/800 in. of graphics, orcombinations of characters andgraphics, depending on the efficiencyof the character and graphicsdefinitions in the display list. Theoperator is able to make any numberof deletions, insertions, or changes ofcharacters or graphics. The graphicscapability includes the generation ofeight 1/8-in.-Iong directional vectors inkeyboard graphics mode. It is possiblealso to generate largeeight-character-length vectors alongfour major axes. These vectors, and

smaller ones used to draw smoothcurves, can be transmitted ineight-level code. The two-processorsystem has a central processor fordeveloping display lists which guidethe CRT display processor. Thememory (4K expandable to 32K16-bit words) can be accessed by bothprocessors.

At the University of Washington,Seattle, Drs. Nancy Frost and E. Huntuse an IMLAC PDS-1 for measuringstimulus-reaction times and forstudying verbal memory abilities, digitspan, and two-choice reaction-timeperformance, all using text displays.The minicomputer system has alsobeen programmed to run visualtracking types of decision-makingexperiments with pictorial displays.

Turning to the Hewle't-Packardminicomputer line,3 we find that mosthave been applied in the biomedicalareas. There are, however, a number inpsychology departments, with one ofthe exceptionally large systems beingat UCLA. A departmental time-sharedsystem, the HP-2116 computer utilizes32K 16-bit words of memory andincludes teleprinters, paper-tapepunches, card reader, andmagnetic-tape unit, along with otherperipherals. The entire system is underthe control of a Hewlett-Packardreal-time executive software operatingsystem which allows members of thePsychology Department to conductexperiments in a real-timepriority-oriented multiprogrammingenvironment. At the same time, thesystem allows low-priority activitiessuch as programming, compilation,debugging, data conversion tomagnetic tape, card reading, and otheractivities to take place. On-line labsconnected to the computer areequipped with 12-eharacter keyboardsto provide communication with themachine; up to 12 stepping motors, 1in each lab if necessary. can beoperated simultaneously. Arandom-access slide projector can beprogrammed to display anyone of 80slides upon request, and the user canprogram display orders and directionsdependent upon the S's response. Thesystem card reader, of course, can beused for grading exams and forrecording and analyzing the results ofquestionnaires or tests. The UCLAHP-2116 system can process simple~eometric pictures for visual displayeither on an oscilloscope or an x-yplotter, and the computer provides forcontrol of certain waveforms and pitchscales in audition, as well as for variousparameters in vision experiments.

In addition, at the UCLA PerceptualSystems Laboratory, Dr. Carteretteand his colleagues have a dedicatedHp·2116B, with 16K of memory, tiedto a moving-head disk drive. This

particular Hewlett-Packard packagewill not be discussed further becausethe variou! uses of the system in thestudy· of psychological andphysiological acoustic experiments, aswell as some visual studies, will bedetailed in the workshop tonight. Forinterested researchers, the HP-2116B isdescribed in the handout, along withits successors, the more inexpensiveHP-2116C and 2100A.3 (TheGA 18/30 system, the application ofwhich will be discussed later today, isdescribed also. 4 )

Note that the above systems withperipherals are very expensive and, aswith some of the DEC and NOVAsystems noted, exceed the.$26,000-$30,000 limit placed on theminicomputer definition. The originalminimal basic package for theHP-2116B, however, was priced at$24,000; for the 2116C with 8K, it is$14,000. So we are still in theballpark.3

To overcome the financial costproblem, some researchers have turnedto government surplus equipment.Several years ago, Dr. Beck, anengineer at Tulane University,proposed that obsolete Minuteman Igeneral-purpose airborne computers,costing $234,000 each with relatedtest equipment, be put to work forresearch and education in the UnitedStates. Today more than a hundredlabs, universities, and industries havethe machines. Rather than scrappingthe obsolete D17B guidance computer,the government has virtually given thesystems to any institution workingunder a federal contract or grant. It isinteresting to note that approximately800 Minuteman D17B computers willbe declared excess by the USAFthrough 1974.

Of course, the required interfacingof these systems is a problem. Beckreports an implementation of thesystem for $525, including the price ofa Flexowriter. The reports of otherusers indicate the hookup price to below also. But technical help must beavailable to do the job. So, beforerushing off for a D17B, be sure thatsomeone in the laboratory is welltrained in the fundamentals ofelectronics and is willing to do thenecessary work.

Note the description of thecomputer in your handout and themailing address for acquisitioninformation.5 The D17B has a2,700-word 24-bit/word disk memoryand a word time of 78 microsec--notexactly the speediest in the business.The delay-type memory is aferrous-oxide-coated magnetic diskwhich rotates at 6,000 rpm. Eachmemory channel contains 128 of the24-bit words. Program and datacharacters can be read into the

46 Behav. Res. Meth. & Instru.,1972, Vol. 4 (2)

Fig. 3. Minuteman D17B computer (cylindrical with fiberglass dome) storedunder a desk-top-type encasement, along with power supply and cooling system.

machine at 800 5-bit char/sec or takenfrom a keyboard entry. Forty-threediscrete input lines are under programcontrol. Program-controlled outputsinclude a 4-bit character out, withparity, a 6-bit register controlling 28discrete output lines, 12 analog voltageoutputs, and pulse-type output lines.Execution time for basic instructionsis one word time for either ll-bitsingle-precision or 24-bitdouble-precision data. Parallelprocessing (execution of two identicalsingle-precision add, subtract, shift, ormultiple instructions) is possible, andhardware divide is available_ There areeight rapid-acceu memory loops.Instruction repertoire contains 39separate codes. And discrete outputlines can be used to sequenceexperiments inYOlving very accuratetiming.

In response to the interfacingproblem, the Autonetics Division ofNorth American recently announcedthe marketing of a D17B input/outputdevice in two configurations:6 aninterface device including an ASR-33Teletype for $5,200, and the samepackage without the ASR-33 for$3,500. Computer mode controlswitches are installed on the dial panelof the Teletype. The system addsmultiplexed ADC capability to thecomputer and provides capability forbulk data storage expansion with anoptional mega-bit tape cassette. Asuggested layout, with the computer,power supply, and venting systemlocated under a desk-type enclosure, isshown in Fig. 3. Of course, the cost ofsuch a commercial interface deviceremoves the D17B from theinexpensive surplus"equipment moneymarket. It adds a cost that mightbetter be applied to the purchase priceof one of the many inexpensiveminicomputers listed elsewhere(Sidowski,1970).

A list of Minuteman computeroWners of interest to psychologists isincluded in your handout.6 The users'group totals over 100, and it is veryactive in providing the necessaryknow-how for setting up the D17B,operating it, and servicing. The groupshares programming and spare parts atlittle or no cost.

L. Happel, Department ofNeurology at the Louisiana StateUniversity Medical Center in NewOrleans, is working with two D17Bs inbiomedical research. Total cost for thedevelopment of the two machines hasbeen between $800 and $1,000. Thecomputers are interfaced to atypewriter and also to a small Fabritekcomputer of average transients; thetotal system appears to be veryeffective. In another installation, atthe Center for Computer AssistedInstruction, Florida State University,

the design of a cooling system, manualcontrol panel, and Flexowriterinterface have been completed, and itappears that the conversion will beaccomplished for less than $500. Theunit will be used for training andpsychological testing. The PsychologyDepartment Computer Lab at theUniversity of Colorado, together withthe Department of ElectricalEngineering, is developing a graphicsprogram around two D17B computerswhich will be used to drive interactiveterminals with video man-machineinterfaces. On the other hand, becauseof the technical requirementsnecessary for making modifications,the machine in the PsychologyDepartment at Bowling GreenUniversity is as yet unused.

If we direct our attention to moreall-inclusive "turnkey" packages, twosoftware/interface computer systemsare offered by psychologicalequipment firms. These are the LehighValley Electronics (LVE) INTERACTand the Grason-Stadler SCAT systems.The INTERACT system has been in­stalled or contracted for by six univer­sities. These include the systems of J. R.Millenson, University of Oxford, whodesigned and uses the ACT languagewith his own hardware and computer,and Harry Coppock at the Universityof Wisconsin-Kenosha, who purchasedan eight-station INTERACT with aPDP-8L 8K CPU and a special audiopackage to be shared with the PhysicsDepartment. Keith Rodewald atCentral Michigan University has afour-station INTERACT system usinga PDP-8/I 8K unit for running pigeons

in student lab-type boxes. AlanKamil's INTERACT system at theUniversity of Massachusetts is used torun rats in avoidance programs; thesystem is shared with others. The mostrecent INTERACT systems have beenpurchased by Northern MichiganUniversity and the University ofWinnipeg. The former installation is afour-station system with a 4K NOVAcomputer unit for running animals oncomplex avoidance schedules. TheWinnipeg installation is aneight-station system with a 16KNOVA. It will serve as a multiusersystem for several investigatorsrequiring complex matching programsand a great deal of output data.

The Grason-Stadler SCAT packageis in six permanent installations. Aseventh SCAT software/interfacesegment was used temporarily byBernie Weiss (University of Rochester,Department of Radiation Biology andBiophysics) on a loan basis forpurposes of system evaluation.

The SCAT system at the Universityof Wisconsin Regional PrimateResearch Center will be discussed atthis meeting by John Davenport, sonothing further will be said concerningits function. An additional unit isinstalled in Professor Amsel'slaboratory at the University of Texas,Austin. The PDP-8II system is used inthe study of performance in therunway maze and the activity wheel,with provisions provided for typicalparameters such as shock andreinforcement. A SCAT installation atthe Aberdeen Proving Grounds has a16K DEC computer, and it is used to

Behav. Res. Meth. & Instru., 1972, Vol. 4 (2) 47

study the short-term memory ofhuman Ss. At the University of TexasMedical School in Galveston (REACTCenter), R. Benton is using SCAT inh is work on the vocationalrehabilitation of the hard-coreunemployed. Basically, the machine isused as a data logger, while providingsome experimental control withfeedback. A work table is provided onwhich Ss are required to dismantlevacuum cleaners, auto carburators, andother devices and to carry out typicalstockroom logging work. The SCATsystem is used for timing, sequencing,and for providing knowledge ofresults. The Department of Psychologyat Florida State University (J. Gilbert)hasa 12K core SCAT system and timeshares 24 h a day. The systemservices a number of people in severalareas in the department. Generally,compiling is done off-line on a PDP-8or PDP-12. Finally, a five-station 8Ksystem is used to monitor animalactivity in 10 cages at the RudolphMagnus Institute, University ofUtrecht, The Netherlands.

Now, if the DEC computers are themost widely used minima chines in theworld, the Digital EquipmentComputer Users Society (DECUS) isits most valuable adjunct. Because ofthe Society's orientation towardsoftware and hardware informationexchange and because of its programlibrary, DECUS often serves as ahighly valuable and respected"security blanket" which helps drawnew computer purchasers to DECproducts. In 1961, DECUS had sixmembers. In 1964 the membershipgrew to 109, followed by 850 in 1966,2,008 in 1967, and 9,110 in 1970.Indeed, during the past year, themembership grew by 50%. The mostrecent figures show the DEC User'sSociety membership exceeding10,000. DEGUS meetings are held inthe spring and fall of each year, andpaper presentations covering variousaspects of computer use andtechnology are published in theProceedings; annual Proceedings reportthe European and Canadianconferences, respectively. ProgramLibrary Catalogs are available to users,as is DECUSCOPE, a publicationoriginally released as a newsletter, butwhich over time has published anincreasing number of technical articlescovering useful bits of information onDEC software and hardware. Now atechnical journal is planned which willpublish submitted articles coveringDEC products.

Even JUG (the Joint Users Group)and LUGS (local Users Groups) havebeen organized. JUG plans to publish acombined Program Library Directory,the first issue of which will contain1,000 programs.

Turning to other companies, theData General Corporation ComputerUsers Group (NOVA andSUPERNOVA), although relativelynew, is growing rapidly, with over2,000 machines installed. Users reportgeneral satisfaction with companysupport and are pleased with thecomputer products and prices. Butmany voice concern over the presentlevel of software support and userinformation exchange, as might beexpected of a relatively new company(the first NOVA computer wasexhibited publicly in 1968). Of course,the group's program library isexpanding and users' meetings arescheduled regularly for purposes ofinformation exchange.

'Ibe Varian Data Machines UsersGroup lists over 300 members invarious disciplines and industries. Mostsystems of interest to psychologists arein the physiological areas. The recentannouncement of several new medicaland physiological system packages willprobably increase the number of usersin these categories. (The MR-20medical research system, for example,includes a 12K 620/L 16-bit wordcomputer, a 61K fixed-head disk, ateleprinter, an analog input/outputsubsystem, computer option package,all necessary cabling and cabinetry,standard Varian software package, andExtended Basic for $27.740.)

Information covering a sampling ofusers groups may be obtained from theaddresses listed in yourhandout.9.10.11.12

If we shift our attention away fromexamining computer uses throughmachine -type or user-groupinformation exchanges. it is interestingto note the citations of minicomputersin professional psychological journals.I have reviewed some of the 1971issues of various journals and foundthe following: In four issues of theJournal of Verbal Learning and VerbalBehavior, containing a total of 63articles, no on-line computer wascited, although a few authors notedthe use of random stimulus listsgenerated by large machines. Threeissues of the Journal of theExperimental Analysis of Behaviorshowed 52 articles with two citingminicomputers-in one case, a PDP-8was used in a monkey study ofgeneralization gradients of responselatency, and in the second report, aPDP-8/I with 4K was used to programall parameters, delay times, etc., in amultiple-choice test of monkeyshort-term spatial memory. Four issuesof Psychophysiology showed fivecitations in 50 articles, with the LINebeing used to compute average evokedresponses and to generate and controlstimuli in two studies, and the samebasic type of computer employed to

read intervals between successiveR-waves in an investigation of heartrate during infancy. Reviewing fiveissues of the Journal of ExperimentalPsychology resulted in a total of 141publications with only five or sixreferencing minicomputer use. Onestudy dealt with verbal memory andreaction time and another, with theeffect of masking tone on auditoryimages. In a short-term memoryexperiment, a minicomputer was usedto generate and display stimuli on aCRT and to monitor Ss' responses andperform initial data tabulations. Hillinterfaced a LINC-8 to an array of 24airjet tactual stimulators, or to a visualdisplay box with a 24-lamp array, toinvestigate the processing of tactualand visual point stimuli sequentiallypresented at high rates. Threerandomly selected 1971 issues of theJournal of Comparative andPhysiological Psychology showed twocitations; five copies of Perception andPsychophysics indicated that 9 of the57 articles cited the employment ofon-line minimachines, with the oldstandbys, Uttal, Pollack, and Mayzner,contributing to the group. I say oldstandbys because in my recent reviewwith Ken Carter of a number ofpsychological journals covering theyears 1965-1970, 11 authors provided52% of the minicomputer citations.Among these were Blough, Pribram,Bliss, Mayzner, Uttal, Weiss, andPollack. The same survey indicatedthat 77% of the on-line computersused in the research were DECmachines, with the PDP-8 and LINCcitations dominating. In classifyingusage, 43% of the minicomputers wereemployed to control stimuluspresentations and reinforcements, tocollect data, and to displayinformation in a real-timeexperimental control sense; 29% of themachines were used strictly for datahandling and reduction. The remainderof the applications includedsimulations, dedicated stimuluscontrol, etc.

As promised, this has been askip-through of the uses ofminicomputers. More detailedcoverages will be presented by others.Before signing off, however, I wouldlike to leave you with one detailedexample from the laboratories of J. R.Millenson of Oxford, England.

Millenson and Jacobs are currentlyusing the ACT (AutomaticContingency Translator) language toautomate a series of visualpsychophysical experiments withdoves. The birds are trained -todiscriminate visual flashes, theirnumerosity, hue, angular separation,or form. The dove's performance isused to measure visual fields, acuity,i nterocular relationships, and

48 Behav. Res. Meth. & Instru., 1972, Vol. 4 (2)

1. Data General CorporationRoute 9Southboro, Mass. 01772

49

SUPERNOVASUPERNOVA has the same basic

configuration as the NOVA: 16·bit word.four accumulators, interchangeable core andread-only memory. and the same packagingdesign. This computer, however, isconsiderably faster than the NOVA. A fullmemory cycle using core takes 800 nsec forthe SUPERNOVA (2.6 microsec for theNOV A). Using read·only memory, cycletime is 300 nsec for SUPERNOVA and2.4 microsec for the NOVA. SUPERNOVAoverlaps the fetch and execute portions ofarithmetic and logical instructiona fromread-only memory, so two numbers can beadded in one 300-nsec memory cycle. Themachine operates in fully parallel fashion onits 16-bit word_all 16 bits at once in asingle 300-nsec machine cycle. With theNOV A, the word is divided into four 4-bitnibbles that require four 400-nsec cycles toprocess.

NOTES

NOVAThe original NOVA is a 16·bit machine

with four accumulators, two of which maybe used as index registers. It offers a choiceof core or read-only memory of lK, 2K, 4K,8K, and up to 32K (using an optionalexpansion chassis) 16-bit words. Directaddressing is required for the balance ofmemory. Number of instructions: 200.Total memory cycle time depends on thetype of module accessed; 2.6 microsec isrequired for a 4K module, 3,9 microsec fora 2K core, 6.5 microsec for a lK core. and2,4 microsec for a lK read-only module.With a 4K core module. typical addinstruction time is less than 5.9 microsec,Up to 62 devices may be connected to the110 bus and may be individually selected bythe central processor.

NOVA comes in desk-top console or in a51,4-in.-tall standard rack-mount package.Both basic versions can hold UP to 20K16·bit words of memory or interfaceswithout expansion chassis. The standard 4Kconfiguration with Teletype interface waspriced at $7.950.

DECUS Proceedings. Spring 1965.Maynard, Mass: Digital EquipmentComputer Users Society, 1965.Pp.93·127.

SCOTT, D. E.• &. DZENDOLET, E. ThePDP-811 as a CAT. In: DECUSProceedings. Spring 1970. Maynard.Mass: Digital Equipment Computer UsersSociety, 1970. Pp. 301-303.

SIDOWSKI, J. B. Minicomputers. BehaviorResearch Methods &. Instrumentation.1970,2.267-288.

STOUT, R. L. IBIOS-an interactivebuffered input-output system for theLINC-8. In: DECUS Proceedings, Spring1971. Maynard. Mass: Digital EquigmentComputer Users Society, 1971.Pp.39.-42.

UTTAL. W. R. Real time computers:Technique and applications in thepsychological sciences. New Yolk: Harper&. Row, 1968.

VO-NGOC. B.. &. POUSSART, D. Anapplication of the FFT in automaticdetection of sleep spindles. 10: DECUSProceedings. Spring 1970. Maynard.Mass: Digital Equipment Computer UsersSociety, 1970. PD. 297-300.

WALKER. A., &. SNAPPER, A. G.Improvements to the. SKED processorcontrol software system. In: DEC USProceedings, Spring 1971. Maynard,Mass: Digital Equipment Computer UsersSociety. 1971. Pp. 7·12.

WEISS. a. (Ed.) Digital computers in thebehavioral laboratory. New Yolk:Appleton-Century-Crofts, 1972, in press.

forthcoming book Digital computersin the behavioral laboratory (1972).

Refer to Alleman and Platt (1972)for an interesting application of thePDP-8II to the study of differentialreinforcement of interresponse times,with controlled probability ofreinforcement per response of pigeons.

Now, if only the doves, rats, andpigeons could be taught to appreciateand understand the English language.Then, as some human Ss are, theycould be ushered into theexperimental chamber and shaped tostart the study by striking a key,following which the machine wouldtake over and say "Hello: I am aminicomputer, and I am working as aproperly programmed psychologist."And it would automatically go onfrom there.

REFERENCESALLEMAN, H. D•• &. PLATT, J. R.

Differential reinforcement ofinterresponse times with controlledprobability of reinforcement perresponse. Learning &. Motivation. 1972.in press.

ANDREAE, S. Message switching systemusing the PDP-5. In: DECUS Proceedings,Fall 1965. Maynard. Mass: DigitalEquipment Computer Users Society,1966. Pp. 55-64.

BONEAU. C. A. Application of the LINCcomputer to operant conditioning. In:DECUS Proceedings. Fall 1965. Maynard.Mass: Digital Equipment Computer UsersSociety. 1966. Pp, 41-44.

BRAZEAL. E. H•• JR•• &. BOOTH. T. L. Ahigh-speed man-computer communcationsystem. In: DECUS Proceedings. Spring1965. Maynard, Mass: Digital EquipmentComputer Users Society. 1965.Pp.33-46.

COX. B. Simulation of neural nets. In:DEC US Proceedings. Spring 1970.Maynard. Mass: Digital EquipmentComputer Users Society, 1970.PP. 203-207.

GRABOWY, R. S.• &. ELLINWOOD, E. H.,JR. On-line detection of EEG spindleactivity. In: DECUS Proceedings. Spring1971. Maynard. Mass: Digital EquipmentComputer Users Society, 1971.Pp.85-87.

KER R, J.. WILLIAMS, T. A.,KHACHATURIAN. J. S.• MILKMAN. N.,&. TOBIN, M. A PDP-12 system foron-line acquisition of heart rate data. In:DECUS Proceedings, Spring 1970.Maynard, Mass: Digital Equipment Com­puter Users Society, 1970. PP. 273-275.

LEWIS, J. L., &. OSGOOD, G. W. Son ofZORO: A second generationmultiprogramming system forpsychological research. In: DECUSProceedings, Spring 1971. Maynard.Mass: Digital Equipment Computer UsersSociety, 1971. Pp. 73-74.

MILLENSON, J. R. On-line sequentialcontrol of experiments by an automatedcontingency translator. In B. Weiss (Ed.),Digital computers in the behaviorallaboratory. New York:Appleton-GenturY-Grofts, 1972, in press.

RAGSDALE, R. G. The learning researchand development center's computerassisted laboratory. In: DECUSProceedings, Fall 1965. Maynard. Mass:Digital Equipment Computer UsersSociety, 1966. Pp. 65-68.

RODGERS, G. F. A study in EEG datareduction. In: DEC US Proceedings,Spring 1970. Maynard. Mass: DigitalEquipment Computer Users Society,1970. PP. 291-296.

SAVAGE, B. I. A monitor system. In:

Behav. Res. Meth. & Instru., 1972, Vol. 4 (2)

monocular bias. Visual stimuli arepresented by a slide projector locatedabove the plane of the apparatus, thebeam of which passes through asplitting system. The split beam thenpasses via two electromechanicalshutters onto tilting mirrors. Whenthese individual shutters open, imagesprojected from the slide are reflecteddown into the animal S apparatus,where they are focused onto twotangential vertical screens and act asdiscriminative stimuli. The dove standsin the center of this array in aconfining transparent box, whichallows the bird access to a food hopperand three vertically arrayed keys. Atrial starts with the lighting of anobserving keylight. When the key ispecked, a shutter opens for a shorttime and exposes one of two stimuliprojected from the slide.Concurrently, the two response keysare illuminated and the dove isrequired to respond to one of the two.If the response is correct, the bird isallowed 3 sec of access to grain on aproportion of trials. Incorrectresponses, on the other hand, result intime-out in darkness with the sameprobability. After the trial, events arerecorded by counters or by theprogram, the next stimulus slide .isselected, and another trial begins.

The houselights, three keylights,food magazine, counters, shutters,projector slide-tray, and optical wedgemotor are all controlled by thecomputerized system. However, thenumber of controlled devices exceedsthe maximum of 11 stimulus bitsavailable, so 5 bits of the stimulusbank have been gated via conventionaltransistor logic; any combination ofthese specifies a unique output line.Thus, 38 possible stimulus lines areavailable, 6 independent and 25 gated.

Millenson is using the same PDP-BISin ACT mode in a series of studies ofconditioned emotional-responseeffects on a variety of behavioralbaselines. In collaboration withDeVilliers, one experiment involves atwo-lever concurrent variable milkreinforcement schedule, with the twolevers associated with differentreinforcement durations. A "danger"stimulus is presented at variable timeintervals. When present, brief gridshocks occur randomly.

The computer/ACT system controlsreinforcement contingencies andadjusts reinforcement durations,change-over delays, and shockproperties as a function of the rat'sresponses. The program is responsiblealso for controlling ink recorders andfor printing out comparative responsedata in safe and "danger" periods. Thisstudy, and the psychophysical doveresearch, are managed jointly. Moredetails of the methodology may befound in Millenson's chapter in the

With .K of core memory, Teletypeinterface, and automatic program load,SUPERNOVA had been priced at $11,700.As of August I, 1970, a new price of $9,600was announced for the same configuration.The SUPERNOVA central processor hasbeen reduced from $6,860 to $6,600, and.K 16-bit memory subsystems have beenreduced from ",600 to $3,660. TheASR-33 is priced at about $1,.00.Multiply-divide and memory protection areoptions.

NOVA was announced in 1968 and firstshipped about 2 years ago. TheSUPERNOVA was introduced in 1969, andthe first machines were delivered in 1970.Both models sold exceptionally well.

The above machines have beensucceeded by three new computers:NOVA 1200 ($5,.50), NOVA 800($6,950), and SUPERNOVA SC ($11,900).NOVA 1200 is a 16-bit multi-accumulatormachine with l,200-nsec memory cycletime. NOV A 800 has a basic memory cycletime of 800 nsec. SUPERNOVA SC has amonolithic memory which, with overlappinainstruction retrieval and execution, give itthe ability to execute arithmetic and logicalinstructions in a single 3000nsec memorycycle.

2. IMLAC Corporation296 Newton StreetWaltham, Mass. 0215.

PDS-1The PDS-1 is a programmable

stand-alone display computer system whichhas been priced at about $10,000 for thebasic unit. The system comprises a CRTdisplay screen, a central processor, and asolid-state keyboard withsoftware-programmed functions. The CRTscreen is capable of displaying about 1,200flicker-free characters, and it may beoriented horizontally or vertically. The14-in. CRT is refreshed from local memoryat, normally, .0 frames/second. CPU wordsize is 16 bits, and memory size is.K words,expandable to 32K. Cycle time is2 microsec; direct addressing, 2K words,with indirect addressing at 32K single level.The CPU has 16 index registers (auto). I/Omaximum word rate is 500 kHz.

As a standard package, the IMLACPDS-l comes with a text andgraphics-editing program. A proprietaryinteractive graphics pattern permitsline-drawing graPhics editing from thekeyboard, as well as limited text.

3. Hewlett-PackardCupertino Division11000 Wolf~ RoadCupertino, Calif. 95014

HP-2116BThe HP-2116B has an 8K memory,

expandable to 32K, with 16-bit wordlengtb. Memory cycle time is 1.6 microsec,add time is 3.2 microsec, and multiply is19.2 microsec. The cOmPuter has 16priority interrupts in the iRandard packaae,expandable to 48. Maximum I/O is600 kbytes/second. The machine has 72basic instructions and seven hardwalere.ste!S. Software includes FORTRAN IV,ALGOL, Basic, and leal-time executive.Orilinal price of the basic packaae wasabout $2.,000. An additional 8K memorywas $10,000.

HP-2116CThe newer HP-2116C basic paekaae also

includes an 8K memory (16-bit processor),self-contained power supplies, and 16input/output channels. Specs are similar tothe HP-2116B, but the price oftbe basic 8Kunit is $1.,000; with a 32K-word memory,the 2116C is $2.,000.

HP-2100AThe more recent HP-2100A computer is

priced at $3,750 without memory modules.The basic cost includes operator's panel,extended arithmetic instructions, memoryparity check without interrupt, power-failinterrupt with auto restart, memory protect,and 1. input/output locations. A .Kmemory module is $3,500; 8K is $7,000;16K is $21,000. An input/output extender($3,500) provides power supplies andprewired slots for 31 additional I/Ochannels. The HP-2100A has a folded planarcore memory, 16-bit words, and a cycletime of 980 nsec. Multiply time is10.7 microsec.

•. General Automation, Inc.Automation Products Div.706 West KatellaOrange, Calif. 92667

GA 18/30The Model GA 18/30 is

software-compatible with IBM's 1130 and1800, hence the designation 18/30. Tbesystem bas a 16-bit word lenath, single- anddouble-precision arithmetic bardwaremultiply-divide plus parity and storaaeprotection bits. The processor has a960-nsec memory with the basic 4Kexpandable to 32K and a 16-word 35-nsecscratch pad. There are six classes ofinstructions with over 400 modifications.Add time (single precision) is 2.• microsec,multiply time is 12.0 microsec, and datachannel transfers, 1.2 microsec. Maximumword transfer rate is 960 kHz. Software isprogram-compatible with the two IBMmachines and includes FORTRAN,assemblers, and real-time executives. Priceof the basic system was about $18,000; thecost was $19,500 with ASR-33. The unittakes 19 x 15% x 2211I in. of rack space andweighs 85 to 105 lb, depending on options.

5. Minuteman D17B

For Minuteman D17B computerpurchasing information write: DefenseSupply Agency, Reutilization Office,Headquarters, Cameron Station, Alexandria,Va. 2231•. Refer to: DSAH-LSR.

For information covering a D17Btechnical data package write: NorthAmerican Rockwell Corp., AutoneticsDivision, 3770 Miraloma, Anaheim, Calif.92803.

6. Several Minuteman computer owners arelisted below.

Beaver Colleae, Psychology, Glenside,Pa. (Dr. B. Mausner)

Bowlina Green State University,PsychololY, Bowlina Gleen, Ohio(Dr. R. Tweney)

University of Colorado, Psychology,Boulder, Colo. (Director CLIPR)

LoulsiaDa State University, Neuroloay,New Odeans, La. (Dr. Lee Happel)

Medical University of South Carolina,Neurosuraery, Charleston, S.C.(Chairman)

M.I.T. Educational Research Center,Cambridae, M-. (Director)

Princeton University, PsychololY,Princeton, N s.

Florida State University, Center forComputer Assisted Instruction,Tallahassee, Fla. (Dr. Paul Merrill)

University of Virginia, Psychology,Charlottesville, Va.

For information on the D17B Users'Group, write to Charles E. Beck, SystemsLab of Electrical Engineering, TulaneUniversity, New Orleans, La. 70118.

7. INTERACT Users (Lehigh Valley Elec­tronics)

J. R. Milienson, Institute of Experimen­tal Psycholoay, University of Ox­ford. Oxford, England (U88S ACT)

H. Coppock, Psychology, University ofWisconsin-PaJkside, Kenosha, Wis.

H. Keith Rodewald, Psychology, CentralMichiaan University, Mt. Pleasant,Mich.

Harlyn Hamm, Psychology, NorthernMichigan University, Marquette,Mich.

Harold Strub, University of Winnipeg,Winnipeg, Man., Canada

8. SCAT System Users (Grason-Stadler)

J. Gilbert, Psycholoay, Florida StateUniversity, Tallahassee, Fla.

A. Amsel, Psychology, University ofTexas, Austin, Tex.

R. Benton, University of Texas MedicalSchool, R.E.A.C.T., Galveston, Tex.

J. Davenport, University of WisconsinRegional Primate Research Center:Madison, Wis.

J. L. Slanger, Rudolph Magnus Institute,University of Utrecht, Utrecht, TheNetherlands

R. Monty, Aberdeen Proving Grounds,Human Research Center, Aberdeen,Md.

B. Weiss, Radiation Biology and Bio­physics, University of Rochester,School of Medicine, Rochester, N.Y.(Used SCAT software and hardwarewith his own PDP-12)

9. DEC Users Group

Mrs. Angela Cossette, ExecutiveDirector, Digital EquipmentComputer Users Society, Maynard,Mass. 01764(DECUS Program Libraryinformation at same address.)

10. NOVA and SUPERNOVA Users Group

Robert C. Baskin, Executive Secletary,Users' Group, Data General Corp.,

Southboro, Mass. 01772For Proaram Abstracts: Data General

Corp. Program Library, Southboro,Mass. 01772

11. VARIAN Users Group

Executive Secretary, Users' Group,Varian Data Machines, 2722 Michelson

Dr., Irvine, Calif. 9266.

12. Hewlett-l'ackard User (Information)

John J. Post, Muketina Support,Hewlett-Packard, MedicalElectronics Div., 175 Wyman se.,Waltham, Mass. 02154

50 Behav. Res. Meth. & Instru., 1972, Vol. 4 (2)