Nakayama (1981) Psychophysical Isolation of Movement Sensitivity by Removal of Familiar Position...

8/12/2019 Nakayama (1981) Psychophysical Isolation of Movement Sensitivity by Removal of Familiar Position Cues

1/7

,3342.6989~81.040427-07802.00/Opergamon Press Ltd

PSYCHOPHYSICAL ISOLATION OF MOVEMENTSENSITIVITY BY REMOVAL OF FAMILIAR

POSITION CUESKEN NAKAYAMAnd CHRISTOPHER . TYLER

Smith-Kettlewell Institute of Visual Sciences, San Francisco, CA 94115, U.S.A.(Received 5 June 1980; i n revised form 15 September 1980)

Abstract-To isolate differential motion sensitivity from the contaminating effects of position sensitivity,we used a field of random dots undergoing a differential shearing motion. Movement threshold ampli-tude vs temporal frequency showed L lope of - 1 on double-logarithmic coordinates, confirming thatthe detection was based on motion rather than on positional information. A comparison of motion andposition sensitivity showed both to be very high, each requiring a differential displacement of only 5 setarc. In comparison to position sensitivity, motion sensitivity showed a IO-fold superiority at the lowestspatial frequencies tested and a poorer sensitivity at higher spatial frequencies.We conclude that the random-dot stimulus can isolate motion- from position-sensitive mechanismsand that motion comprises a distinct form of sensitivity, not derivable from the measured forms ofposition sensitivity.

INTRODUCTIONA proven psychophysical strategy useful in the analy-sis of parallel visual mechanisms is to dissociate sys-tems by appropriate and selective variation of stimu-lus conditions. To isolate the separate contribution ofrods and cones, for example, appropriate selection ofwavelength, adapting background and flicker rate isneeded. Otherwise, attempts to measure one systemcan be contaminated by the responses of the other.

An even more serious form of contaminationappears in human motion sensitivity. In particular,the problem is caused by the fact that movement of avisual target always involves a change of shape orposition. Consider a familiar example-the barelyperceptible movement of the minute hand of a clock.After looking at it for some time, we know that it hasmoved. Bui did we really see it move or did we inferthat it moved because of its change in position?

This mixture of motion and position informationhas confounded previous attempts to measure theminimum motion threshold (Aubert, 1886; Graham eta/., 1948). Only in a few particular and relativelylimited cases has it been possible to deduce that thepsychophysical response is mediated by a motionrather than by a position system (Leibowitz, 1954;Tyler and Torres, 1972).

To provide a general method to overcome this con-tamination of position information and thereby to de-velop a widely applicable approach to the study ofmovement sensitivity, we have used a random-dot

* Supported by NIH Grants 5ROl-EY-01582,5P30-EY-01186, 2S07RR 05566, 5ROl-EY-02124 and the Smith-Kettlewell Eye Research Foundation. A preliminaryabstract of these results was published in Invest Ophrhal.ui sual Sci. Suppl ., 1978, p. 288.

pattern where adjacent sections move differentially.Our initial supposition was that because there are nofamiliar position cues in a random-dot pattern, theobserver would be unable to detect changes in thelocal shapes of the display as it moves.

METHODRandom-dot moving grating

In its static form, the stimulus consists of a 5 x 6rectangular field of random picture elements (pixels)on the face of a CRT (Hewlett-Packard 1332A). Each-picture element (pixel) subtended 2.8 arc min and hada 50% probability of being light or dark. Mean lumin-ance of the display was maintained at lOcd/m* asmeasured by a Pritchard photometer and the contrastwas set to 800/,. A high contrast photograph of a por-tion of the display is shown in Fig. 1B.The electronic circuitry required to generate thisTV raster and noise is schematized in Fig. IA. Inbrief, the highly dense field of random-dots was gener-ated using a pseudorandom noise source based on 14bit shift register circuitry, clocked at a very high rate(6 MHz). The repeating noise sequence was reset oneach frame. As a consequence, the same pattern ofvisual noise was refreshed 400 times/set. To generatea differential displacement in the horizontal direction,an amplitude modulated function generator (FG-2 inFig. IA), synchronized to the vertical sweep, was con-nected to the horizontal oscilloscope axis. Thisdeformed the whole raster, such that the relative hori-zontai position of each row was a sinusoidal functionof its vertical position. It should be clear that thespatial frequency of this deformation is determined bythe carrier frequency of FG-2. To generate differentialhorizontal movement in the display, the AM gener-

V.R. 21.4~-* 427


2/7

428 KEN N K Y M nd CHRISTOPHER W. f\iirH

51 2 KHzRAMP

1OGPOT

Fig. 1A. Block diagram of apparatus used to generate sinusoidal movement gratings in a random-dotpattern. A raster was formed by suitably triggering two electrically generated ramps at 51.2 KHz and4OOHz and feeding them into the X and Y axes respectively. Video noise was produced by a triggeredshift-register circuit. To generate differential motion according to equation 1) a triggered AM-modu-lated carrier signal from FG-2) was fed into the X axis which, in turn. was modulated by a lowfrequency sinusoid from FG-it B. A portion of the random-dot display as it appears in its static form

on the CRT screen.

ator was amplitude modulated by another functiongenerator (FG-1 in Fig. 1A).

It should be apparent that the movement of thewhole screen of random dots is identical to a standingwave, with characteristic nodes and antinodes. Theinstantaneous horizontal velocity can be described bythe expression

J? = sin (2nf;r) sin (2n/, Y) (1)where the temporal frequency of oscillation (f,), thespatial frequency (f,), and the amplitude (A) are underelectronic control. Time is represerted by t and thevertical position of each pixel is represented by Y.Because it is a transverse standing wave, with pointsin any horizontal row moving at the same horizontafvelocity, there is no expansion or contraction in anygiven region. Dot density always remains constant.

A vector representation of the instantaneous vel-ocity field for two different spatial frequencies ofmotion is depicted in Fig. 2. Note the sinusoidalchange in velocity as a function of the vertical pos-ition of each row and the fact that the spatial fre-quency of this velocity profile can be varied (cf. A vsB). It should be noted that a similar type of differen-tial movement and its relation to depth perceptionhas been described by Rogers and Graham (1979).

M rn~di~~a~edinusoidal lineTo provide a comparison with the results obtained

from a stimulus which contains no familiar positioncues, we also made measurements with a stimuluswhich would be least expected to isolate motion pro-cessing. This consisted of a single vertical line whichwas sinusoidai~y deformed in exactty the same man-

Fig. 2. Vector lkld representation of the instantaneous velocity fiefd of the movement of random dotsfor two different spatial frequencies, Moment spatial frequency in A is twice that of B. Length of eacharrow is proportional to velocity. It should be stressed that spatial frequency refers to the spatialdistribution of the differential movement rather than to the luminance distribution.


3/7

Isolation of movement sensitivity 429ner as the screen full of random dots. It was generatedby disconnecting the noise source from the Z inputand the horizontal 51.2 Hz sawtooth ramp from the Xinput. Although its changing appearance was radic-ally different from that of the random-dot stimulus(see inset of Fig. 4), its motion remains a standingwave conforming to equation (1).Peri odic v erni er acuit y target

As a final point of comparison, we tested positionsensitivity using a single static sinusoidal line having avariable spatial frequency (sinusoidal periods/degreeof visual angle). In this case the observer was notrequired to detect motion, but to detect a deviationfrom collinearity or straightness. Such a stimulus hasbeen previously investigated by Tyler (1973) andappears to be related to vernier acuity. Two suchstimuli having different spatial frequencies are shownin the inset of Fig. 6.General procedures

Two practiced observers, the authors, were usedthroughout the study. Thresholds were obtained bythe method of adjustment, varying the amplitude ofthe stimulus by a logarithmic potentiometer placedbetween the AM function generator and the X axisinput (see Fig. 1A). At least 4 separate settings wereobtained for each plotted point.

Viewing distance was kept at 114 cm, the roomillumination was maintained at a low photopic level,and the observer was instructed to restrict fixation tothe center of the screen.

RESULTSExperi ment l -D etecti on of relat iv e movement in therandom-dot di splay bypasses posit ion sensiti vi ty

To test the supposition that movement in therandom-dot display is mediated by velocity sensitiveelements rather than by position sensitive mech-ansims, we relied on the fact that the peak velocityof any sinusoidally oscillating point in the displayincreases in proportion to temporal frequency(d sin wt/dr = w cos wt). The peak velocity of a 1 Hzoscillating motion for example is 10 times that of a0.1 Hz oscillation, given a constant amplitude. Thus, ifthe detection were mediated by a motion system,requiring a minimum threshold velocity rather than aminimum displacement, proportionately less amph-tude would be necessary as the temporal frequencywas increased, at least over some significant range.On the other hand, if the detection were based solelyon position, there would be no dependence ofthreshold on temporal frequency.The dependence of threshold amplitude on tem-poral frequency was determined for a constant spatialfrequency of movement (3 c/deg). Figure 3 shows thatover the range between 0.1 and 1.0 Hz, thresholds fallsteadily and conform well to the slope of - 1 on thelog-log representation. Because points along this part

I I \ I I.l .2 .5 1 2 5Temporal Frequency (Hz)

Fig. 3. Motion threshold amplitude vs temporal frequencyfor random-dot moving grating (3 c/deg). Note thatbetween 0.1 and 1 Hz, the relation is reciprocal, indicatingmovement rather than position sensitivity. Observer KN(open symbols), observer CWT (filled symbols) for this andsubsequent figures. Diagonal line represents slope of - 1.of the function all represent movements having thesame peak velocity, the result indicates that motiondetection in the random-dot grating is determined ex-clusively by the peak velocity in the stimulus. Thus itconfirmed our original supposition that the observerwould be unable to detect local deformations in therandom-dot pattern even though they were fixated infovea1 vision. As such, the measurement of motionsensitivity using random dots appears to bypass thecontaminating effects of position sensitivity.Experi ment 2-Contami nati on by posit ion nformati on

Before describing a spatial characteristic of motionsensitivity, we first provide a contrasting examplewhere an attempted measure of motion sensitivity ishighly contaminated by recognizable position infor-mation. For this we used the moving sinusoidal line(described in Methods) modulated with the identicalstanding wave motion. Unlike the random-dot pat-tern, it shows a recognizable change in appearance asit goes through successive stages of movement (seeinset of Fig. 4) and the experiments indicate that therelation between threshold amplitude and temporalfrequency is very different. The function is essentiallyflat over the 0.1-I Hz range just where the thresholdsfor the moving gratings fell by a factor of 10 (seeFig. 4).Di scussi on: Experi ments I and 2

Although Experiments 1 and 2 both use movingstimuli, it is clear that only Experiment 1 measuresmotion sensitivity. In Experiment 2 there is anobvious and recognizable change in the appearance ofthe line as it moves and this occurs irrespective ofhow slowly the process proceeds. In other words,recognizable shape or position rather than movementdetermine the threshold, so the sinusoidal line stimu-


4/7


5/7

isolation of movement sensitivity 431

Spatial Frequency (C/deQ)Fig. 6. Position threshold ampiitude as a function of spa-tial frequency using a periodic vernier acuity task. Insetabove the plotted function shows the appearance of thestatic line showing a low and a high frequency sinusoidaidisplacement profile. The observer is to set the amplitudeof this sinusoidally modulated line so that it is just notice-ably different from a straight line. Note that in contrast tomotion seen in the random-dot grating (as in Fig. 5), thelowest thresholds are at a relatively high spatial frequency.At the lowest spatial frequency tested, position thresholdsare 4-10 times higher than motion thresholds (compareFigs 5 and 6).smallest sinusoidal amplitude that would produce aperceptible deviation from collinearity or straightnessin the static line. These threshold settings are plottedagainst spatial frequency in Fig. 6 and replicate theresults obtained by Tyler (1973). Several featuresdeserve special mention, particularly as they differ sig-nificantly from the spatial characteristics of motionsensitivity depicted in Fig. 5. First is the fact that theshapes of the curves are completely different. Positionsensitivity peaks at a relatively high spatial frequency,about 2c/deg, and there is an inverse relationship

between spatial frequency and threshold over therange of 0.2-2c/deg. Because points along this por-tion of the function share the same peak angular devi-ation from the vertical, it supports the emerging viewthat vernier and other static hyperacuity tasks aredetermined by orientation sensitive mechanisms(Andrews, 1973; Tyler, 1973). The second feature ofinterest is that sensitivity for position at these verylow spatial frequencies is far worse than that formotion sensitivity, with a difference up to tenfold. Acomparison of the lowest frequency portions of Figs 5and 6 shows that position sensitivity is very poor at0.2c/deg where motion sensitivity is at its best. Thethird result is the extent to which position sensitivityis clearly superior to motion sensitivity at the higherspatial frequencies. Position sensitivity is still increas-ing as spatial frequency is increased while motion sen-sitivity is clearly decreasing.Experi ment S-Low spafid frequency fine stimuli canalso isolat e differenti l movement mech nisms

The preceding comparisons between motion andposition sensitivity provide further insight as to howmotion and position mechanisms will determinethresholds. If a target has recognizable position cuesand the distance over which these recognizable differ-ences in position are small, then position sensitivitywill dominate as its threshold will be lower than themotion threshold. This occurred when a relativelyhigh spatial frequency of 3 cJdeg was used (Fig. 4).There is no indication of motion sensitivity at thisspatial frequency as this is close to the optimal spatialfrequency to stimulate position sensitivity as shownby the function depicted in Fig. 6.

On the other hand, if we choose a moving linestimulus requiring the comparison of position over alarge distance, the position system would be poorlystimulated and the results would again be determined

I I 1 , I.l .2 .5 1 2 5 IOTemporal Frequency (Hz)

Fig. 7. Threshold amplitude vs temporal frequency for a moving sinusoidal line (moving according toequation 1) and having a spatial frequency of only 0.2c/deg. Note that in comparison to the linestimulus having the higher spatial frequency (as shown in Fig 4) the relation between threshold ampli-tude and frequency is reciprocal. indicating that the threshold depends on motion sensitivity.


6/7

432 KEN N R Y M and CHRISTOPHER . TYLC)Iby sensitivity to velocity. To test this idea, wemeasured motion thresholds of a moving sinusoidalline exactly as in experiment 2, except that a muchlower spatial frequency was employed (0.2 c/deg,equivalent to one sinusoidal cycfe/screen). In com-parison to Experiment II (described in Fig. 4), therelation between threshold displacement and tem-poral frequency was reciprocal, indicating motionrather than position sensitivity (see Fig. 7). Eventhough it is a line stimulus, with recognizable positionfeatures, the spatial configuration of the stimulus isnot optimal for position sensitivity and thereby per-mits motion sensitivity to dominate. An analogousresult has been described by Westheimer (1979) wherehe showed that the threshold displacement of a hop-ping line is somewhat lower than the threshold differ-ence required to note a static difference in distance incases when the two reference lines were relatively farapart.

DISCUSSIONSeparate organization of motion and posi t ion sensi t iv i ty

In physical terms, motion is always correlated witha change in position. As a consequence, one mightinitially suppose that relative motion coding is a pto-cess derived after the coding of relative position. Thisis a serial view and it suggests that motion sensitivitymight share at least some characteristics with relativeposition sensitivity.

It should be clear that the results of the presentstudy contradict this expectation. Of primary impor-tance is the fact that motion and position sensitivityhave completely different spatial tuning functions.Static position sensitivity appears best when the dis-tance between differential displacements is relativelysmall, showing a peak in the spatial frequency curveat 2 c/deg. Thus the best comparisons are made over apeak-to-peak range of 15 arc min. Motion sensitivity.on the other hand, is at its best even at the low valueof 0.2 c/deg.* As such, it is remarkable in its ability tomake a differential comparison of a tiny displacement(5 arc set) over a very large retinal distance, exceeding4000 arc sec. Because differential motion sensitivity isso much better than position sensitivity when thecomparison is made over these very large distances

* A question arises as to whether the motion of the dotsrelative to the edge of the screen at 3 eccentricity is suf-ficient to account for the very low movement thresholdsobtained at the lowest spatial frequencies. We reject thisinterpretation for two reasons. First, we compared differen-tial motion 0.2c/de& with whole field motion in a separ-ate experiment using fotced-choice feedback procedures.The whole-field motion threshold for KN was 12 arc sec.more than twice the 5 arc set figure obtained for differen-tial movement (or higher by 4 standard errors of the varia-bility). Thus thresholds determined by the peripheral edgecomparison are far too h&h to explain the lowestthresholds seen in Fig. 5.Furthermore, even in the sinusoidal line stimulus whichis 3 away from the edge of the screen, relative motionthresholds have the same low value of 5 arc set at 0.2 c,/deg(see Fig. 7).

(compare the low spatial frequency portions o 1 gs iand 61, it is unlikely to be derived from the psycho-physically measured position sensitivity WC sugpeht itrepresents a distinct form of hyperacuity ;md c;tn bebest seen as a system wired in parallel to positionsensitivity. Preliminary investigations show that dli-ferentia motion sensitivity also differs in its lack oftolerance to retinal image motion (Nakayama. 1979).a result which stands in contrast to the findingsobtained for acuity and hyperacuity targets (west-heimer and McKee, 1975).Receptivefield interpretation

The spatial characteristics of motion sensitivity canbe interpreted physiologically if we assume that thereare movement-specific neurons having very largereceptive fields and that these neurons are driven byinput subunits having high sensitivity and smallreceptive fields. Such units have been analyzed indetail in single unit studies (Barlow and Levick, 1964:Michael, 1968). It must be assumed that subunits hav-ing the opposite motion preference will generate inhi-bition rather than excitation at the level at which inte-gration occurs. Such a neuron could be sensitive tovery small displacements as it integrates over a largearea but would have reduced outputs if there werealso movements in the opposing direction within thereceptive field. It would respond best to spatial fre-quencies of movement which had half periods largerthan the receptive field diameter. If we consider thefact that movement sensitivity falls aff above 0.7cycie/deg (see Fig. 5), we can conclude that these intebgrating receptive fields for motion are as large as 2 inextent, even in the human fovea1 region. A similarconclusion was reached by Richards (1971) on thebasis of motion aftereffect data.

REFERENCESAndrews D. P., Butcher A. K and Buckley B. (1973) Acui-ties for spatial arrangements in line figures: human andideal observers compared. I/ision Res. 13,599-620.Aubert H. (1886) Die Bewegungsemfindung. Pfliigers Archsges. Physi ol . 39, 347-370.Barlow H. B. and Levick W. R. (1964) The mechanism ofdirectionally selective units in rabbits retina. J. Physiok

I f 477-504.Brown J. F. (1931) The thresholds for visual ~overnefl~~Psychol. Forsch. 14* 249-232.Graham C. H., Baker K. E., Hecht M.. Lloyd V. V. (1948)Factors influencing thresholds for monocular movementparallax. J. exp. Psychol. 38, 205-223.Leibowitz H. (1955) The relation between the ratethreshold for the perception of movement and luminancefor various durations of exposure. J. exp. Psy oi. 49.209-214.Michael C. R. (1968) Receptive fields of single optic nervefibers in a mammal with an all cone retina. II: Directio-nally selective units. J. Neurophy siol. 31, 257-267.Nakayama K. (1979) Common image motion severely dis-rupts motion parallax sensitivity. J. opt. Sue. Am. 69,1487.Nakayama K. and Qler C. W. (1978a) Spatial character-istics of fovea1 movement sensitivity in random-dot grat-ings. Jnuest. Opht hal. v isual Sci. Suppi. p. 288.


7/7

Isolation of movement sensitivity 433Nakayama K. and Tyler C. W. (1978b) Relative motioninduced between stationary tines. Vision Res. 18

1663-1668.Richards W. (1971) Motion detection in man and animals.Brain Behav. Evol. 4 162-181.Rogers B. and Graham M. (1979) Motion parallax as aninde~ndent cue for depth perception. perception 8.125-134.Tyler C. W. and Torres J. (1972) Frequency responsecharacteristics for sinusoidal movement in the fovea andperiphery. Percept. Psychophys. 12.232-236.

Tyler C. W. (1973) Periodic vernier acuity. J. Physiol. 228.637-641.Westheimer G. (1977) Spatial frequency and light spreaddescriptions of visual acuity and hyperacuity. J. opt. Sot.Am. 67.207-212.Westheimer G. and McKee S. P. (1975) Visual acuity in thepresence of retinal image motion. J. opt. Sot. Am. 65847-850.Westheimer G. (1979) The spatial sense of the eye. Invest.Ophthal. visual Sci. 18,893-912.

Nakayama (1981) Psychophysical Isolation of Movement Sensitivity by Removal of Familiar Position...

Documents

Transcript of Nakayama (1981) Psychophysical Isolation of Movement Sensitivity by Removal of Familiar Position...