Festinger (1974) Inferences about the efferent system …wexler.free.fr/library/files/festinger (1974) inferences about the... · 46 LEON FESTINGER AND A. MONTAGUE EASTON had (a)

Psychohgkal Review1974, Vol. 81, No. 1. 44-58

INFERENCES ABOUT THE EFFERENT SYSTEMBASED ON A PERCEPTUAL ILLUSION

PRODUCED BY EYE MOVEMENTS1

LEON FESTINGER 2 AND A. MONTAGUE EASTON

New School for Social Research

Precise measurement of the position of the eye as it follows a moving targetmakes possible the exact computation of retinal information about the path ofmovement of that target. Comparing this retinal information with thereported visual perception of the path of movement enables inferences to bemade concerning what information about eye position was used by theperceptual system. On the assumption that information available to theperceptual system about eye position comes only from monitoring efferentcommands, these inferences are also about the content of those commands.Our data and analysis suggest that the efferent command for smooth pursuiteye movement, at the stage where it is monitored, contains good informationabout the direction of movement but only crude information about speed.

In the last decade neurophysiologists haveincreasingly emphasized the importance ofunderstanding the organization and func-tioning of the efferent system. The generalquestion is how, and in what form, motorcommands are formulated and issued andhow the nervous system controls motoractivity. Some progress has been made onsuch questions at a neurophysiological level.For example, studies have been reported onthe relationship between firing rates inmotor neurons and force exerted by limbmovements (Evarts, 1966, 1968, 1972);studies have identified cells in the centralnervous system that regularly fire in con-nection with specific motor movement(Bizzi, 1968; Bizzi & Schiller, 1970);advances have been made in the understand-ing of how muscle spindles function in fineregulation of movement (Granit, 1970;Matthews, 1964). It is not our intentionhere to review this material but only to pointout the nature of some of the questions that

1 The research on which this article is based wassupported by Grant MH-16327 from the NationalInstitute of Mental Health to L. Festinger. Theauthors wish to acknowledge the help, during thecourse of experimentation and interpretation, ofSaulo Sirigatti, Melvin K. Komoda, and HaroldA. Sedgwick.

2 Requests for reprints should be sent to LeonFestinger, Department of Psychology, New Schoolfor Social Research, 66 West 12th Street, NewYork, New York 10011.

have been addressed. The problems, issues,and current state of the work is wellpresented by Evarts, Bizzi, Burke, DeLong,and Thach (1971).

However, there has been little or no workthat addresses questions about the efferentsystem from a psychophysical point of view.There have been studies that have shown theimportance of the efferent system in enablingadaptation to distorted visual input (Fest-inger, Burnham, Ono, & Bamber, 1967;Held, 1961) but this work only implicatesthe efferent system—it does not explicate it.We think that it is possible to make progressin understanding the nature and functioningof the efferent system through psycho-physical methods. It is the purpose of thisarticle to propose a paradigm for doing this,to present the results of a study followingthis paradigm, and to present the beginningsof a theoretical model, necessarily narrow inscope, that suggests itself as a result ofthese findings.

AN APPROACH TO THE STUDY OF THEEFFERENT SYSTEM

Some aspects of visual perception are par-ticularly well suited for analyzing what hap-pens in the efferent system. This suitabilityarises from the fact that the perceptual sys-tem does not have access to informationabout eye movements or eye position basedon proprioceptive feedback from the extra-

EFFERENT SYSTEMS AND PERCEPTION 45

ocular muscles. Since there may be somewho would dispute the preceding statement,it is worthwhile to briefly summarize theevidence.

The extraocular muscles in the human docontain muscle spindle receptors and tendonreceptors. These are capable of transmittinginformation about length of the muscle orchange of length (spindle receptors) andabout tension or change of tension in themuscle (tendon receptors). Thus, it is pos-sible that such information is transmitted ina way that might be used by the perceptualsystem. The psychophysical evidence, how-ever, indicates that this is not the case.Brindley and Merton (1960), for example,anesthetized the surface of the eyeball andthe eyelids, put an opaque cap over thecornea to eliminate retinal information, andmoved the eyes of their subjects back andforth by seizing the tendon of the lateral ormedial muscle with forceps. They reportedthat under these circumstances the subjectcompletely lacks awareness that his eye ismoving—an awareness that might be ex-pected to exist if information from theextraocular muscles was available to theperceptual system. Brindley and Mertonfurther reported that if the eye is held com-pletely motionless, and the subject is toldto move his eyes, the subject reports thathis eye did move.

Recently Skavenski (1972) and Skaven-ski and Steinman (1970) have disputed thisresult. Among other things, they reportedthat if the eye is moved mechanically bylarge amounts (6° to 10°, the subject willcorrectly identify the direction of the eyemovement 75% to 80% of the time. Onecannot be certain in their study that someclues were not available from pressure onthe eyeball or on eyelids. In another study,however, Skavenski, Haddad, and Steinman(1972), using the same technique, showedconclusively that, even if any proprioceptiveinformation from the extraocular musclesexists, it is not used by the perceptual sys-tem for egocentric localization of visualdirection. They find that the only extra-retinal information that is used perceptuallyis "outflow" information, that is, informa-

tion about the efferent commands to themuscles.

We can, then, conclude that, in theabsence of retinal information, knowledgeabout eye movements or eye position that isavailable to the perceptual system is basedon information about what the eye wascommanded to do and not on informationabout what the eye actually did do. At somelevel of transmission of an efferent commandfrom the central nervous system, this com-mand is monitored and the information con-tained in that command at that level is avail-able to the perceptual system.

How does this help us to study the effer-ent system? Let us imagine a situation inwhich an observer watches a target movingon a completely contourless background, hiseyes more or less following the movementof the target. The observer's perception ofthe path of movement of the target mustdepend on a combination of two kinds ofinformation: information about movementof the target on the retina and informationabout eye movements. Only if both thesesets of information were accurate, and onlyif they were accurately combined, would theperception of the path of movement beveridical. If the eye is stationary, of course,perception could be based only on retinalinformation; but with a moving eye this isnot possible.

Let us tentatively assume that positionand movement on the retina are accuratelytransmitted to the perceptual system. Itseems reasonable to assume this since, if itwere not so, visual perception would rarelybe veridical and we know it usually is. Letus further tentatively assume that the "cal-culations" that combine retinal informationwith information about eye position areaccurate. We will examine this assumptionfurther in connection with specific data, butlet us proceed now accepting the assumption.

Under this set of assumptions .it is real-istically possible to collect data from whichone can infer the informational content ofthe efferent commands for eye movementsas they exist at the level of the output sys-tem at which they are monitored and areavailable to the perceptual system. If we

46 LEON FESTINGER AND A. MONTAGUE EASTON

had (a) precise measurement of the per-ception of the path of the target moving ona contourless field, (b) precise measure-ment of the position of the observer's eye atevery moment in time while observing thetarget, and (c) precise information of theactual position in space of the target at everymoment in time, we could then perform thefollowing computations. From the informa-tion about the actual position of the targetand the actual position of the eye, we couldcompute the actual path traveled by thetarget on the retina. By comparing thisretinal information with the perception ofthe path of movement of the target, we couldinfer what information the perceptual systemhad about the movement of the eye. Thisinference would, according to our reasoning,also be an inference about the informationalcontent of the monitored efferent commands.

If the observer's perception of the path ofmovement were veridical, the whole en-deavor would, of course, be in vain. Wewould simply conclude that the perceptualsystem had complete and accurate informa-tion about the eye movements—a conclusionthat would make this line of approach notvery productive. There is evidence in theliterature, however, that if the eye engagesin smooth pursuit movements the observer'sperception is, frequently, far from veridical.Johansson (1950), for example, reportedvarious interesting tnisperceptions of thepaths of movement when several luminouspoints move in a relatively contourless fieldat velocities that would be expected to elicitsmooth pursuit eye movements. Johanssoninterpreted his data differently, however,rejecting the possibility that eye movementsare important in the misperceptions. Asanother example, Stoper (1967) reportedthat during smooth pursuit movements ofthe eye, the perception of the location of abriefly flashed spot of light is determinedalmost entirely by its retinal location withoutmuch "compensation" for eye movement.

It is, then, possible that the monitoredefferent command for smooth pursuit eyemovements does not contain complete andaccurate information. The data collectionprocedure we outlined above might enable usto infer what information is actually avail-

able. We consequently chose one such"illusion" for intensive study.

The Fujii Illusion

Fujii (1943) reported an extensive seriesof experiments that dealt primarily with theperception of the diameter of the circularpath of a spot of light moving in an other-wise dark room. In the course of theseexperiments he also reported on the per-ceived path of motion of a spot that moved,for one cycle, in a triangular path or in asquare path. In his experiments each sideof the triangular path subtended a visualangle of about 17° and each side of thesquare path subtended a visual angle ofabout 12.5°. He employed velocities of 6°and 22° per second. He reported a sur-prising difference between the actual and theperceived path of movement under thesecircumstances. Figure 1 (adapted fromFigures 2 and 4 in Fujii, 1943) summarizeshis report of these perceived paths. Thesolid lines indicate the physical paths, andthe arrows indicate the direction of motion.The dashed lines describe the perceivedpaths. His report about this is very brief.

This misperception of the path of move-ment seemed to fit the conditions that wouldenable us to make inferences about the effer-ent output for smooth pursuit eye move-ments. We consequently engaged in a ser-ies of observations to see whether or not theFujii illusion did fit the conditions andassumptions that we described above. Theseobservations were all carried out with a spotrepetitively cycling in a square path on acathode ray tube display with a very fastphosphor so that there was no physical traceof the path of movement. A contrast screenin front of the cathode ray tube eliminatedany general glow from the surface and theroom was in total darkness. In short, forour observations the luminous spot movedon a contourless background.

The observations reported below arebased primarily on verbal reports from manyobservers. All observers were asked tofollow the spot with their eyes. In additionto the observations of the authors, colleagues,and graduate students in our laboratory, wealso obtained reports from naive observers.


Occasionally we encountered an observerwho simply did not perceive the illusion.We did not keep a record of how many sincethat was not our purpose at the time. Ourmemory is that this amounted to 2 or 3persons out of more than 50 observers whomwe used at one time or another. Theresults presented below were reported by allobservers who perceived the illusion. Wewill present these results organized aroundmajor questions.

Does the misperception of the path o\movement exist under conditions in whichsmooth pursuit movements are likely tooccur? We explored the perception of thepath of movement over a wide range ofsizes of the square path (3° to 10° of visualangle per side) and a wide range of fre-quencies (.1 to 2.0 cycles per second) andvelocities (2° to 60° per second). Forthese observations the spot always movedwith uniform speed, turning corners instan-taneously. The conclusions may be sum-marized as follows.

If the speed of the moving spot exceeded40° to 45° per second, the misperception didnot exist. Sometimes, at such high veloc-ities the spot seemed to move erratically butthe corners were perceived as right angles.At lower velocities, frequency was a criticalvariable. If the frequency exceeded 1.6 to1.7 cycles per second, the illusion disap-peared and the path was seen as square.Within the range of frequencies and veloc-ities that produced the illusion, an additionalobservation is worth noting. At frequenciesbelow .3 to .4 cycles per second, the per-ceived path of motion was usually of theform illustrated in Figure 2. At frequenciesabove .5 to .6 cycles per second, the per-ception was of the form shown in Figure 3.

In summary, we can answer the questionpositively. The velocities and frequenciesthat produce the misperception are ones thatare likely to elicit some amount of smoothpursuit eye movement.

Is the retinal information accurate underthese conditions? Since there are manyvisual illusions that do not depend on eyemovements, it is important to know whetherthe Fujii illusion exists if the eye is station-ary and all of the information about the

FIGURE I. Perception of the path of a targetmoving in a square or triangular path. (Solid linesindicate the physical paths, and arrows indicate thedirection of motion. Dashed lines describe the per-ceived paths.) (Adapted from an article by E.Fujii from the 1943 Japanese Journal of Psychol-ogy. Copyrighted by the Japanese PsychologicalAssociation, 1943.)

path of movement is available from theretina. The answer to this question is clear.If a fixation point is provided, and the ob-


f

FIGURE 2. Perception of the path of a targetmoving in a square path at frequencies below .3to .4 cycles per second.

server is asked to fixate that point and notfollow the moving spot, the illusion dis-appears—the path is seen veridically as asquare.

In addition, it may be pointed out that if acathode ray tube display with a slow phos-phor is used so that there is an extendedphysical trace of the path of movement ofthe spot, the path is seen as a square.

We still, however, do not have a completeanswer to our question. One might suspect

FIGURE 3. Perception of the path of a targetmoving in a square path at frequencies above .5to .6 cycles per second.

that even if the retinal information wasaccurate when the eye was relatively station-ary, such information might be inaccuratewhile the eye was moving. Pursuing thehunch that the Fujii illusion might be due,at least in part, to the fact that the spotturned the corners instantaneously, a featthat the eye could not accomplish, we ob-served the path of movement with the eyefollowing the spot while the spot movedwith sinusoidally varying velocity along eachside of the square. Under these conditionsthe velocity of the spot is fastest in the mid-dle of each side, slowing up as it approacheseach corner. The results are again clear.With sinusoidally varying velocity of thespot, the path of movement is seen as square.We can, then, assert that even while the eyeis in motion the shape of the path of move-ment may be seen veridically. It may benoted that, under these conditions, the per-ceived square path is considerably smallerthan the physical path. We will discuss thisin detail later.

Is the computation that combines retinalinformation with information about eyeposition accurate? This question is, ofcourse, least amenable to a direct answer.Since we do not know the exact informationthat the perceptual system has about eyeposition (that is what we hope to be ableto infer), it is certainly difficult to show thatcomputations involving that information areaccurate. The best we can do is to showthat under some circumstances, even witheye movements, perception is veridical. Ifthe perception is veridical, it seems plausibleto assume that the computation was accurate.We have already mentioned that when thespot moves with sinusoidally varying veloc-ity, the path of movement is perceived cor-rectly as a square. This lends weight to theassumption that the computation is accurate.Another piece of evidence can also be pro-duced. If one provides two stationarypoints for fixation and asks the observer tomove his eye from one to the other whilethe moving spot is cycling with uniformspeed, one finds that again the perception isveridical. The path is seen as a square.Under these conditions, of course, the eyemovements are saccades and not smooth


pursuit movements; hence, we can at leastassert that the computation for combiningretinal information and information aboutsaccadic eye movements seems to beaccurate.

We may then say that, as far as we coulddetermine, the Fujii illusion does fit the con-ditions that we would like to have forattempting to infer the content of the mon-itored efferent command for smooth pursuiteye movements. We proceeded to collectthe relevant data.

Measurement of Eye PositionAll observations during which eye movements

were recorded were made with the observer's lefteye occluded. The position of the right eye wasrecorded continuously, using a noncontacting eyetracker developed by Cornsweet and Crane (1972).Infrared light was projected onto the eyeball andthe system measured the position of the reflectionfrom the rear surface of the eye lens (fourthPurkinje reflection) in relation to the position ofthe corneal reflection (first Purkinje reflection).Two direct-current voltage outputs were obtained,one corresponding to the horizontal, the other tothe vertical, component of eye position.

The output of the eye tracker has a noise levelof about 2' of arc and is linear (except for possiblelocal irregularities of the observer's eye) over agreater than 12° range both horizontally andvertically.

The data on eye movements were recorded intwo ways. A record of the horizontal and verticalcomponents was made, using an oscillograph, onchart paper moving at a rate of 100 millimetersper second. The two output voltages from theeye tracker were also put through two digitalvoltmeters used as analogue-digital converters,and the digital information was printed out onceevery 35 milliseconds by a high-speed printer.Event markers on both the chart paper and thedigital printout indicated the time at which themoving spot in the visual display turned eachcorner of its square path so that the tworecords could be collated in time, and also, ofcourse, to allow exact computation of the positionof the eye in relation to the position of the movingtarget at each moment in time. The digital volt-meters used as analogue-digital converters tookfrom 2 to 7 milliseconds from the time they re-ceived a command to sample to the time they com-pleted the sampling: hence, there is that much un-certainty in the temporal resolution of our data.The commands to the analogue-digital convertersto sample were synchronized with the movement ofthe spot in the visual display, so that a "samplecommand" always occurred at the moment thespot reached a corner of the square path.

Procedure for Data CollectionThe moving spot was displayed on a Hewlett-

Packard model 1310A display scope with a PISphosphor. The movement of the spot was con-trolled by a Wavetek model 116 function generatorand a Tektronix model 4701 eight-channel multi-plexer. The visual display was located 100 centi-meters in front of the observer's eye.

Data on eye movements while viewing the tar-get that moved in a square path were collectedfrom three observers. One of these (M.) was ahighly experienced observer; another (T.) wasinexperienced but had prior knowledge about thephenomenon; the third observer (C.) was inex-perienced and totally ignorant about the phe-nomenon. After each observer was aligned in theapparatus, with head held steady using a biteboard,calibrations were made to determine the exact eyeposition measures that corresponded to positionsfixated in the visual field. There is undoubtedlysome error in these calibrations since the eye isnot steady in fixation. We estimate this error tobe no greater than 4' or 5' of arc.

The data were collected in total darkness. Eachobserver viewed four conditions of the visual dis-play as follows: With the spot moving with uni-form speed in a square path of 6° per side, thelength of time for the spot to traverse each sidewas either 350, 700, or 1,015 milliseconds. Inaddition each observer viewed the spot moving ina smaller square path, 3° per side, the length oftime to traverse each side being 350 milliseconds.The target velocities corresponding to these fourconditions are, in order, approximately 18°, 9°, 6°,and 9° per second. For each of these conditions,the observer fixated a stationary spot which thenstarted to move clockwise. After eight cycles ofthe square the spot stopped and the observerfixated the spot again. During this entire periodeye movements were recorded.

For the analysis of the data we relied primarilyon the digital printout at 35-millisecond intervals.The chart paper recordings were used for twopurposes: easy identification of cycles duringwhich eyeblinks occurred, which were excludedfrom analysis, and easy identification of the occur-rence of saccades. Sometimes the digital samplingoccurred during the course of a saccade and, whenthis happened, we used the chart paper record inan attempt to specifiy more exactly in time thebeginning and end of the saccade.

Patterns of Eye Movements

Although the plotting of the exact patternof eye movements that occur while the ob-server follows the moving spot is only thefirst step in the calculation we wish to make,it is nevertheless important to see what theylook like. We selected four examples ofsuch eye movements to illustrate each of the


four conditions of observation that weemployed and to represent each of the threeobservers. These examples are shown inFigures 4, 5, 6, and 7. They are quiterepresentative of the typical kinds of eyemovements.

Each figure shows the exact position ofthe eye in relation to the physical locationof the moving spot for one complete cycle ofthe square path. The small squares indicatethe physical location of the corners of thesquare path through which the spot travels.The filled circles in each figure indicate theposition of the eye at one moment in time;successive circles are separated by 35 milli-seconds. The position of the eye at themoment in time that the spot instantaneouslyturns each corner is indicated by an opencircle around that particular data point. Thecycle starts at the point labeled "S" and ends

•

/•

.

oba. C (-8° 101BMS.

FIGURE 4. Successive eye positions for ObserverC. for one cycle of target moving at .25 cycles persecond (6° path). (Small squares indicate thephysical location of the corners of the square path.Filled circles indicate position of the eye at onemoment in time; successive points are separatedby 35 milliseconds. Open circles around filledcircles indicate position of the eye at the momentthe spot instantaneously turns the corner. Thecycle starts at "S" and ends at "E". Unconnectedconsecutive circles indicate smooth pursuit move-ment. Circles connected by solid lines indicatesaccadic movements. Abbreviations : Obs. = ob-server; ms. = millsecond.)

\

•®

obs. M6° 700 MS

FIGURE S. Successive eye positions for ObserverM. for one cycle of target moving at .36 cycles persecond (6° path). (Small squares indicate thephysical location of the corners of the square path.Filled circles indicate position of the eye at onemoment in time ; successive points are separated by35 milliseconds. Open circles around filled circlesindicate position of the eye at the moment thespot instantaneously turns the corner. The cyclestarts as "S" and ends at "E." Unconnected con-secutive circles indicate smooth pursuit movement.Circles connected by solid lines indicate saccadicmovements. Abbreviations: Obs. — observer; ms.= milliseconds.)

at the point labeled "E." If consecutivecircles are not connected, that indicates thatthe eye was engaging in smooth pursuitmovement. If the eye engaged in saccadicmovements, the circles are connected by asolid line. The irregularities in the patternof eye movements, most noticeable at theslower speed shown in Figure 4, are prob-ably due to errors of measurement. Themagnitude of these irregularities is wellwithin the noise level and temporal resolu-tion capability of our equipment.

It may be seen that all the eye movementpatterns show a combination of smooth pur-suit movements with interspersed saccades.As the velocity of the moving spot increases,the velocity of the smooth pursuit move-ment, of course, also increases. In addi-tion, as the velocity increases, the inter-spersed saccades tend to become larger.

EFFERENT SYSTEMS AND PERCEPTION-

There are no periods in which the eye isrelatively stationary, as occurs in situationsin which only saccadic eye movements aremade. Here, the saccadic eye movement,when it occurs, seems to be superimposed on,and does not disturb or interrupt, thesmooth pursuit movement. This is evidentin all of the figures but can, perhaps, beseen most clearly in Figure 6. In this cycle,two saccades occurred while the eye wasfollowing the spot moving upward on theleft side of the square path. In each case, atthe end of the saccade, the eye immediatelyshows smooth pursuit movement in essen-tially the same direction as the movementprior to the saccade. It is as if the smoothmovement continued throughout with thesaccade overlaid on it. It may be worthnoting that Woodworth and Schlosberg(1956, p. 511) made this same point aboutsuch situations. It may also be worth

® .

obs. C380 MS.

FIGURE 6. Successive eye positions for ObserverC for one cycle of target moving at .71 cyclesper second (6° path). (Small squares indicatethe physical location of the corners of the squarepath. Filled circles indicate position of the eye atone moment in time; successive points are sep-arated by 35 milliseconds. Open circles aroundfilled circles indicate position of the eye at themoment the spot instantaneously turns the corner.The cycle starts at "S" and ends at "E." Un-connected consecutive circles indicate smooth pur-suit movement. Circles connected by solid linesindicate saccadic movements. Abbreviations: Obs.= observer; ms. = milliseconds.)

®'

® '•

®

••®

obs. T3° 35OMS

FIGURE 7. Successive eye positions for ObserverT. for one cycle of target moving at .71 cycles persecond (3° path). (Small squares indicate thephysical location of the corners of the square path.Filled circles indicate position of the eye at onemoment in time; successive points are separatedby 35 milliseconds. Open circles around filledcircles indicate position of the eye at the momentthe spot instantaneously turns the corner. Thecycle starts at "S" and ends at "E." Unconnectedconsecutive circles indicate smooth pursuit move-ment. Circles connected by solid lines indicatesaccadic movements. Abbreviations: Obs. = ob-server; ms. = milliseconds.)

noting, in passing, that there tend to be morevertically than horizontally oriented sac-cades. We do not find it mentioned in exist-ing literature but, in our data, the strongsuggestion exists that smooth pursuit move-ment in the horizontal direction is moreadequate than in the vertical direction.

It is particularly relevant to our presentpurpose to examine, in Figures 4, 5, 6, and7, the pattern of eye movements around thetime that the moving spot actually turns acorner of the square, that is, near the en-circled data points. It is clear, in all theeye movement records, that the eye does notturn corners instantaneously as does thespot. When the spot turns a corner the eyemakes a rather gradual turn with a com-ponent of motion in the previous directionpersisting for some time. At higher veloc-ities particularly, the eye sometimes, prob-ably in anticipation, begins its gradual turn


even before the spot has reached the corner.While the eye is making these gradual turns,of course, the movement of the spot on theretina would show an indentation from thecorner similar to what is reported percep-tually. This gives us our first clue as towhy the misperception of the path of move-ment takes its particular form, namely, theindentation from the corners.

Calculation of "Corrected RetinalInformation"

From the data on eye movements such asthose shown in the Figures 4, 5, 6, and 7,we could, of course, compute and plot on agraph the exact movement of the spot on theretina of the observer. This would tell uswhat retinal information existed concerningthe path of movement of the spot. Thiswould be an adequate procedure for ourpurposes if the eye movement patterns con-sisted entirely of smooth pursuit movements.We would then be able to make inferencesabout what information concerning thesesmooth pursuit movements is available tothe perceptual system.

The eye movement patterns, however, donot contain only smooth pursuit movementsbut also show interspersed saccades. It isnecessary then, for our purposes, to makesome assumptions about what informationthe perceptual system has about saccadiceye movements and to correct the calculationof retinal information to take this intoaccount. In other words, we want to cal-culate a combination of what is known aboutthe path of movement of the spot from retinalinformation and from information aboutsaccadic eye movements. Then the onlymissing information would be about thesmooth pursuit eye movements, and we couldproceed with our inferences about thatinformation. It is clear, of course, that thevalidity of these inferences will depend, inpart, on the validity of our assumptionsconcerning information about saccades.

It seems most plausible to make theassumption that the perceptual system doeshave good information about the extent anddirection of a saccadic eye movement. Sucheye movements are very rapid and are bal-

listic in nature. The efferent commands forsuch eye movements must, hence, be entirelypreprogrammed. That is, the complete setof instructions for the movement has to beissued before the movement starts since,once the ballistic eye movement is underway, it is no longer controllable. It seemssensible to assume, then, that the monitoredefferent command for such a movement con-tains all the information. In addition, ourown informal observations indicate that iffixation points are provided, and only sac-cadic eye movements presumably occur, theperception is veridical. So in our calcula-tions we will assume that, whenever a sac-cadic eye movement occurs, the perceptualsystem knows accurately and exactly thedirection and magnitude of that movement.

Before we can proceed with these calcu-lations one other decision has to be made.Do we assume that when a saccade occurs ithas interrupted and replaced the smoothpursuit movement, or do we assume thatthe saccade is overlaid on, or added to, thesmooth pursuit movement which persists?As mentioned above, particularly in refer-ence to Figure 6 when we were discussingthe patterns of eye movements, the recordslook as though the most plausible assump-tion is the latter—that the smooth pursuitcomponent of the motion of the eye persistsand the saccadic movement is added on tothat. In our calculations we have attributeda speed and direction of smooth pursuitmovement during the interval occupied bya saccade by interpolation from the smoothpursuit movements immediately precedingand immediately following the saccade.

"Corrected Retinal Paths" of theMoving Spot

Given these assumptions we can calculatewhat the perception of the path of motion ofthe spot would be if it were based solely oninformation about movement on the retinaand correct information about saccades thatwere superimposed over an ongoing smoothpursuit motion. Representative examples ofthe results of these calculations are pre-sented in Figures 8, 9, 10, and 11. They arechosen, again, to show one example from

EFFERENT SYSTEMS AND PERCEPTION S3

position of the spot at the moment that itturns a corner. For the sake of visualclarity, the retinal path for each side of thesquare is separated from the others. It maybe remembered that the direction of targetmovement is always clockwise. On eachfigure is indicated the scale equivalent of 1°of visual angle. At the lowest frequency,illustrated in Figure 8, many of the datapoints cluster so closely that the path on theretina is not readily apparent. Where thisis the case we have indicated, with a dashedline, the general path on the retina.

Let us first point to the most salient fea-tures of Figures 8, 9, 10, and 11. It isclear that in many instances they conformto the general features of the reported per-ceptions of the path of movement of thespot. At the lower frequencies, illustratedin Figures 8 and 9, the typical pattern alongone side of the path is a rapid inward inden-

ob«.l

FIGURE 8. Corrected retinal path for ObserverM. for one cycle of target moving at .25 cycles persecond (6° path). (Each filled circle representssuccessive relative positions at 35-millisecond inter-vals of the moving spot on the retina, plotted interms of visual field rather than the reversedretinal field, corrected for saccadic eye movements.Encircled circles indicate retinal position of thespot at the moment it turns a corner. For visualclarity, the retinal path for each side of the squareis separated from the others. Dashed line indicatesgeneral path on retina where data points wereclustered very closely. Abbreviation: Obs. =observer.)

each of our four observation conditions andat least one example from each of our threeobservers. These figures are quite typicalof all of the data.

Each of the figures shows the "correctedretinal path" for a single cycle of the eyefollowing the spot. Each filled circle againrepresents successive relative positions (35-millisecond intervals) of the moving spot onthe retina (plotted in terms of visual fieldrather than the reversed retinal field) cor-rected for saccadic eye movements. Theencircled data points again indicate retinal

FIGURE 9. Corrected retinal path for Observer C.for one cycle of target moving at .36 cycles persecond (6° path). (Each filled circle representssuccessive relative positions at 35-millisecond inter-vals of the moving spot on the retina, plotted interms of visual field rather than the reversedretinal field, corrected for saccadic eye movements.Encircled circles indicate retinal position of thespot at the moment it turns a corner. For visualclarity, the retinal path for each side of the squareis separated from the others. Abbreviation: Obs.= observer.)


®

T., whose data are shown in Figure 10, oftenreported the perception of asymmetricalsides even at higher frequencies.

It is clear from an examination of Figures8, 9, 10, and 11 that the information con-tained in these corrected retinal paths pro-vides a major basis for the Fujii illusion.One can readily come to the conclusion thatthe information available to the perceptualsystem about the position of the eye duringsmooth pursuit movement is not very great.Our purpose, however, is to try to infer asmuch as we can about exactly what informa-tion is available about smooth pursuit move-ments; hence, our next step should be toexamine the major ways in which these pat-terns of corrected retinal path differ fromthe reported perceptions. There are two

oba.T

FIGURE 10. Corrected retinal path for ObserverT. for one cycle of target moving at .71 cycles persecond (6° path). (Each filled circle representssuccessive relative positions at 35-millisecondintervals of the moving spot on the retina,plotted in terms of visual field rather than thereversed retinal field, corrected for saccadic eyemovements. Encircled circles indicate retinal posi-tion of the spot at the moment it turns a corner.For visual clarity, the retinal path for each sideof the square is separated from the others. Abbre-viation: Obs. = observer.)

tation from the corner point, followed eitherby a slow stretching out of the path untilthe next corner, as in Figure 9, or by a closeclustering of points, as in Figure 8. Thebest illustrations of this are, perhaps, thepath on the left-hand side of Figure 8 andon the lower side of Figure 9.

At the higher frequencies, illustrated inFigures 10 and 11, the corrected retinalpaths look much more symmetrically in-dented, conforming to the general form ofthe perception shown in Figure 3. The bestexamples of this are the right-hand side ofFigure 10 and several sides in Figure 11.It may be of interest to note that Observer

oba.M1°

FIGURE 11. Corrected retinal path for ObserverM. for one cycle of target moving at .71 cycles persecond (3° path). (Each filled circle representssuccessive relative positions at 35-millisecond inter-vals of the moving spot on the retina, plotted interms of visual field rather than the reversedretinal field, corrected for saccadic eye movements.Encircled circles indicate retinal position of thespot at the moment it turns a corner. For visualclarity, the retinal path for each side of the squareis separated from the others, Abbrevation: Obs.= observer.)


major differences that are important to pointout.

Let us consider the corrected retinal pathsshown in Figure 8. It is clear that they alldiffer considerably from the perception ofthe path of movement of the spot. Neverdid this observer, or any other observer,report seeing the moving spot circle back orreverse direction along any one side. Thesedata points must, obviously, be stretched outin space in a temporal sequence if they areto conform more closely to the actual percep-tion. In other words, the perceptual sys-tem does add something about velocity ofsmooth pursuit eye movement into theperception.

The same point can be made again, on aslightly different basis, considering any ofthe figures showing the corrected retinalpaths. (Figures 8, 9, 10, and 11). Forexample, these paths in Figure 9 would indi-cate that the perceived path would have anapparent extent of about 2° of visual angleper side on the average. This, again, doesnot correspond to the perception. Althoughwe tried, we were not successful in obtainingprecise estimates of the perceived extentfrom our observers. It was a very difficultthing for them to report on.

The most successful procedure was tohave the observer follow the moving spotfor a number of cycles and then to suddenlypaint the entire square on the cathode raytube display. One could then obtain grosscomparisons of the perceived size and theactual size. What can be said from suchreports is that at lower frequencies the extentof the perceived path usually looked smallerthan at higher frequencies; that the per-ceived extent of motion was always less thanthe actual physical extent; and that rarelydid the perceived extent seem less thanabout half of the physical extent. The cor-rected retinal paths shown in Figures 8, 9,10, and 11 are almost all too short. Againwe must come to the conclusion that theperception has had something added in aboutthe position of the eye during smooth pur-suit movements.

The second major difference between thecorrected retinal paths and the perceptionsof the observers concerns the directional

orientation of the paths. This can be illu-strated by examining the paths in Figure 11.For this observer, as well as for all others,the perceptions are oriented correctly asthey are shown in Figures 2 and 3. But thecorrected retinal paths are frequently tilted.This orientation difference does not alwaysappear, but as can be seen in Figures 8, 9,10, and 11, it frequently does exist. We areconsequently led to the conclusion that theperceptual system knows something aboutthe direction in which the eye was com-manded to move during smooth pursuit.

Content of the Efferent Command forSmooth Pursuit Eye Movement

We previously stated, with support fromthe research literature, that whatever infor-mation the perceptual system has about thesmooth pursuit movement of the eye mustrepresent the information contained in theefferent commands for those movements atthe level at which those commands are mon-itored. Our task now is to infer, as preciselyas we can, what this exact information is.We have already said that the informationcannot be very exact and specific—the cor-rected retinal paths are too similar to theperceptions for this to be the case. We havealso said that some information about move-ment of the eye must be present—the cor-rected retinal paths need to be stretched outin temporal order; and that informationabout the commanded direction of the eyemovement must be present—the perceivedorientation of the paths was not tilted as aremany of the corrected retinal paths.

If we had been successful in obtainingvery precise measurement of the perceptionof the path of the moving spot, we wouldhave an easier task at this point. However,we were not successful at this. We werenot able to get adequate measures of themagnitude of perceived indentation from thecorners, or of the perceived extent of move-ment of the spot. On the basis of the gen-eral reports of perception that we obtainedit seems reasonable, however, to make thefollowing inferences:

1. The efferent command for a smoothpursuit eye movement contains specification


of the direction for the eye to move. Anyinstance in which the corrected retinal pathdoes not correspond to the physical path, butthe perception is veridical, must indicate thatthe efferent command contained correctinformation about the relevant aspect of theeye movement. The correctly perceivedorientation of the path indicates that theremust be some information about direction inwhich the eye is commanded to move. Thisdoes not, of course, mean that the perceptualsystem knows the direction in which the eyeis actually moving. It only knows the direc-tion in which the eye was commanded tomove. In the particular instance of theFujii illusion, since the eye cannot turncorners suddenly, the actual directions ofeye movements probably differ from whatthe eye was commanded to do in the neigh-borhood of the corners of the path. Thisaccounts for the perceived indentations fromthe corners that are reflected in the correctedretinal paths. The perceptual system onlyknows the direction contained in the efferentcommand.

2. The efferent command for a smoothpursuit eye movement contains very inade-quate information about the speed of theeye movement. Since the extent of the per-ceived path was always considerably lessthan the physical extent of movement, weknow that the efferent command does notcontain accurate information about speed ofeye movement. On the other hand, as wediscussed in the previous section, someinformation about this is present. It mustbe that the actual speed of the eye movementis controlled by some more peripheral loopin the efferent output system so that thespeed of the eye can be approximatelymatched to the speed of the target eventhough the command, at the monitored level,does not contain this information.

We cannot say more about this with greatconfidence at this point but, in speculation,we would propose the following. The effer-ent command for smooth pursuit movement,in addition to specifying the direction forthat movement, contains only the commandto move. Hence, the perceptual systemknows that the eye is moving and willattribute at least some minimal velocity to

that movement. When the target changesdirection suddenly a new command will beissued which, in addition to ordering adirection change, may contain instructionsto move faster or slower. Thus, the per-ceptual system could know that, at differentparts of the path, the speed of the eye dif-fered as instructed. This is our guess.Further explorations are needed in situationsin which the perception can be measuredmore exactly.

Possibility of an Alternative Interpretation

The validity of our conclusions about theefferent output for smooth pursuit eyemovements rests, obviously, on a chain ofreasoning that includes a number of assump-tions. It is important, consequently, toconsider other possible interpretations of theFujii illusion and the data we have collectedconcerning it. A successful alternativeinterpretation would cast serious doubt onour analysis.

There is a major possible alternative thathas been suggested to us many times. Thisalternative would hold that the perceptualsystem does have accurate information abouteye position during smooth pursuit move-ments, but that there is some time mismatchbetween retinal information and informationabout eye position. Perhaps the retinalinformation arrives at some central process-ing stage sometime later (or earlier) thanthe eye position information. Then theretinal information might be matched to thelater (or earlier) eye position, and thistemporal mismatch might produce the mis-perception of path of movement rather thaninadequate information about eye position,as we have argued.

We can, of course, take our records of eyemovements, from which we know the exactposition of the eye and the position of thetarget on the retina at each point in time,and compute what the predicted perceptionwould be for various assumed differences ofarrival times of the two pieces of infor-mation. We have done this and in noinstance does the perception predicted inthis manner at all resemble the Fujii illusion.Figure 12 shows examples of the results ofthese computations to illustrate the out-


come. In this figure the upper set of circles,labeled A, shows the result of our compu-tation of corrected retinal path for ObserverC. along the upper side of the 6° squarepath, the spot taking 350 milliseconds totraverse each side. The next two sets ofcircles, B and C, show the results of calcu-lations that assume eye position is accuratelyknown, and assume that the retinal infor-mation arrives 70 milliseconds earlier (B)and 70 milliseconds later (C) than the infor-mation about eye position. The fourth setof circles (D) represents the eye and targetpositions from which the above calculationswere made. These examples are quitetypical of these computations. They do notresemble the perceived Fujii illusion. Thisparticular alternative interpretation does notseem viable.

SUMMARY

When an observer's eyes follow a targetmoving on a contourless background, hisperception of the path through which thetarget moved has to be based on a combi-nation of retinal information and informa-tion about eye position. Precise measure-ment of the eye position as it follows thetarget can provide exact computation ofwhat information was available on the retina.A comparison of this retinal informationwith the reported visual perception of thepath of movement enables inferences to bemade concerning what information abouteye position was used by the perceptualsystem.

Such observations and computations weremade using a situation in which the path ofthe moving target is known to be misper-ceived. On the assumption that the per-ceptual system does have accurate informa-tion about saccadic eye movements, someconclusions could be reached about what theperceptual system knows about smooth pur-suit eye movements.

Since existing data point to the fact thatthe perceptual system does not have infor-mation about eye position based on feedbackfrom the extraocular muscles, it seems thatwhatever information about eye position isknown is based on monitoring the efferentcommands issued to those muscles. Thus,

®

®

o o o o o o o o o

®

FIGURE 12. Comparison with computations basedon differential latencies of retinal and eye-positioninformation. (Section A shows corrected retinalpath; Section B, eye-position, information delayed70 milliseconds; Section C, retinal informationdelayed 70 milliseconds; Section D, eye and targetpositions used in computations. Small squaresindicate physical location of the corners of thepath. Open circles indicate target positions, filledcircles eye positions when the spot turns the corner.Solid line indicates saccadic eye movements.)

inferences and conclusions about what theperceptual system knows about smooth pur-suit eye movements may also be regardedas inferences about the content of the effer-ent commands for such movements at thatpoint in the efferent system at which thosecommands are monitored.

Our data and analysis suggest that theefferent command for a smooth pursuit eyemovement, where it is monitored, contains


information about the direction of movementbut does not contain precise informationabout the velocity of the movement.

REFERENCES

Brzzi, E. Discharge of frontal eye field neuronsduring saccadic and following eye movements inunanesthetized monkeys. Experimental BrainResearch, 1968, 6, 69-80.

BIZZI, E., & SCHILLER, P. H. Single unit activityin the frontal eye fields of unanesthetized mon-keys during eye and head movement. Experi-mental Brain Research, 1970, 10, 151-158.

BKINDLEY, G. S., & MEETON, P. A. The absenceof position sense in the human eye. Journal ofPhysiology, 1960, 153, 127-130.

CORNSWEET, T. N., & CRANE, H. D. An accurateht'O dimensional eye tracker using first and fourthPurkinje images. Menlo Park, Calif.: StanfordResearch Institute, 1972.

EVARTS, E. V. Pyramidal tract activity associatedwith a conditioned hand movement in the mon-key. Journal of Neurophysiology, 1966, 29,1011-1027.

EVARTS, E. V. Relation of pyramidal tract activityto force exerted during voluntary movement.Journal of Neurophysiology, 1968, 31, 14-27.

EVARTS, E. V. Activity of motor cortex neuronsin association with learned movement. Interna-tional Journal of Neuroscience, 1972, 3, 113-124.

EVARTS, E. V., BIZZI, E., BURKE, R. E., BELONG,M., & THACH, W. T., JK. Central control ofmovement. Neurosciences Research ProgramBulletin, 1971, No. 9.

FESTINGER, L., BURNHAM, C. A., ONO, H., &BAMBER, D. Efference and the conscious experi-ence of perception. Journal of ExperimentalPsychology Monograph, 1967, 74(4, Ft. 2).

FUJII, E. Forming a figure by movement of aluminous point. Japanese Journal of Psychology,1943, 18, 196-232.

GRANIT, R. The basis of motor control. NewYork: Academic Press, 1970.

HELD, R. Exposure-history as a factor in main-taining stability of perception and coordination.Journal of Nervous and Mental Disease, 1961,132, 26-32.

JOHANSSON, G. Configurations in event perception.Uppsala, Sweden: Almquist & Wicksell, 1950.

MATTHEWS, P. B. C. Muscle spindles and theirmotor control. Physiological Reviews, 1964, 44,219-288.

SKAVENSKI, A. A. Inflow as a source of extra-retinal eye position information. Vision Re-search, 1972, 12, 221-229.

SKAVENSKI, A. A., HADDAD, G., & STEINMAN, R.M. The extraretinal signal for the visual percep-tion of direction. Perception & Psychophysics,1972, 11, 287-290.

SKAVENSKI, A. A., & STEINMAN, R. M. Controlof eye position in the dark. Vision Research,1970, 10, 193-203.

STOPER, A. E. Vision during pursuit movement:The role of oculomotor information. Unpub-lished doctoral dissertation, Brandeis University,1967.

WOODWORTII, R. S., & SCHLOSBERG, H. Experi-mental psychology. New York: Holt, 1956.

(Received May 29, 1973)

Festinger (1974) Inferences about the efferent system …wexler.free.fr/library/files/festinger (1974) inferences about the... · 46 LEON FESTINGER AND A. MONTAGUE EASTON had (a)

Documents