Apparent contraction and disappearance of moving objects in the peripheral visual field

Vision Res. Vol. 13, pp. 959-975.

Pergamon Press1973.Printedin GreatBritain.

APPARENT CO~T~CTIO~ AND DISAPPEA~~CE OF MOVING OBJECTS IN THE PERIP~E~L VISUAL FIELD’ R. H. DAY Department of Psychology, Monash University, Clayton, Victoria 3168, Australia (Received 24 Jtdy I972)

INTRODUCTION

ANSBACHER (1938, 1944 rst drew attention to apparent contraction of a light arc rota& at between 0.5 and 2-OY.p.slfn a darkened peripheral field. The effect was later confirmed by MARSHALLand STAR (1964) and investigated in some detail by STANLEY(1964,1966, 1968). The experiments reported here were designed primarily to establish the basis of this reduction in apparent length. During the course of the enquiry it became clear that the disappearance of a light radial aperture travelling in the same peripheral path in a dark field as a leading aperture of greater luminance, noted much earlier by MCDOUGALL (1904), is a closely related phenomenon. Consideration of the stimulus conditions for both effects suggested that they may be attributable in part to visual masking by light following a fast rise in the light threshold as demonstrated by BAKER(1953), BOYNTON (196 I) and CRAWFORD (1947). The contribution of this form of masking was supported by data from experiments in which continuous and step-function luminance gradients of the moving stimuli were systematically varied. However, the occurrence of marked apparent contraction when the moving stimulus arc was located in the peripheral visual field and a negligible effect when it was more centrally located strongly suggests that “ Y-type” or “transient” retinal ganglion cells recently described by ENROTH-CU~ELLand ROBSON(1966) and CLELAND,DUBIN and LEVICX(1971) are primarily involved. The experiments reported by ANSBACHER(1944) showed that apparent contraction increases with angular velocity up to 2-O r.p.s. and is markedly less or absent when the trailing section of the stimulus does not follow in the same path as the leading section. In explanation Ansbacher assumed that neural functioning is intermittent involving “pulsations” with brief periods of inactivity between. He suggested that at a given angular velocity there would be some degree of re-stimulation of the same retinal region, the amount of restimulation varying with angular velocity. The region of maximum stimulation would determine apparent length in that this zone would be both clearer and lighter and would probably suppress adjacent dimmer regions by means of contrast, a form of “perceptual telescoping” to the maximally stimulated region. While the apparent contraction of a rotating arc has been frequently confirmed (MARSHALLand STANLEY,1964; STANLEY,1964, 1966, 1968; STANLEYand JACKSON,1969) the extent of contraction has generally been much less than originally found. Generally, there 1 The experimentsreported were carried out in the Department of Psychology, University of Exeter, U.K., during tenure of a VisitingProfessorshipawardedby the Associationof CommonwealthUniversities. 1 thank the Department of Psychology for providinglaboratory, technicaland clerical facilitiesduring the course of the experimentalprogramme. V.R.13/5-rl 959

960

R. H.

DAY

has been little support from these recent investigations for the re-stimulation explanation. However, it has been confirmed that the brightness of the perceptually shorter moving arc is greater than its stationary control (STANLEY, 1967). APPARATUS

AND

METHODS

Apparatus The standard stationary or moving stimulus objects were in the form of concentric arcs or radial slots cut out from black aluminum disks and illuminated from behind. The source of light was a 40 W incandescent and transluscent lamp the light from which was diffused through a sheet of opalescent plexiglass. The central region of this diffusing screen was of uniform luminance so that when the disks rotated in front of it, the luminance of the arcs or slots was constant throughout the circular path. Luminance was controlled by means of neutral density filters which were mounted over the apertures on the reverse side of the disks. In the semi-dark laboratory the observer viewed these moving or stationary apertures against the uniform luminance of the diffusing screen. Further details of the stimulus objects are described in appropriate places below. The disks were mounted on the shaft of a variable speed reversible motor and held in place by means of a dome-shaped nut 5 mm in diameter which served also as a fixation point. The standard viewing distance throughout all experiments was 40 cm and was maintained by means of individual dental bites. A black rectangular screen 54-S cm high and 80 cm wide with a circular aperture 26 cm in diameter was located 5 cm in front of the disk so that the moving or stationa~ stimulus objects were visible but the rim of the disk was occluded by the edge of the aperture. The variable stimulus with which the observer matched the apparent length of the standard arc consisted of a second arc 5 mm wide cut out of the screen to the left of the aperture, concentric with the centre of the disk and symmetrical about its horizontal diameter. The distance from the central fixation point to the outer edge of this arc was 13.7 cm so that its visual eccentric angle relative to the fixation point was 18” 54’. The variable was illuminated from behind by the blue emission (5-3 cd/m2) from a sheet of electroluminescent panel. The length of the arc could be continuously shortened by means of a slide which was moved from below upward. Throughout ail experiments the adjustment of this variable arc to match that of the moving or stationary standard arc was unde~aken by the experimenter. The observer indicated verbally when the point of subjective equality had been reached. The length of the variable was read in degrees subtended at the centre of the disk from a scale around the outer edge of the arc. This scale was too fine and too peripheral to be visible to the observer in the darkened laboratory. The angular velocity of the standards was adjusted by means of an’ electronic control (EC Motomatic). In all experiments except the second the velocity was 1 r.p.s. In the control trials the standards were stationary and positioned randomly in one of the four quadrants defined by the vertical and horizontal through the

centre of the disk. During experimental trials the laboratory was dimly illuminated by reflected light from the incandescent source behind the disk and screen. Procedure The observer while fixating the centre of the disk was required to indicate verbally when the variable arc moved by the experimenter was apparently equal in length to that of the moving or stationary standard. The observer was instructed to open his eyes, fixate the centre and indicate when the length of the variable achieved equality with the standard. Except for rest periods between groups of trials the observer remained with head positioned by the dental bite and with eyes closed, only opening them when told to do SO. Following the general instructions at the beginning of a sessionthe observerwas givenfour practicetrials, two with the standard stationary and two when it was rotating, in that order. The standard used during practice was always different from that used in subsequent experimental trials. Different groups of observers recruited from among undergraduates, graduate students and faculty were used in each experiment. However, these groups were not entirely independent in that some observers participated in more than one experiment.

EXPERIMENT

l-CONTINUOUS

LUMINANCE

GRADIENTS

Typically a test stimulus is masked by another (the masking stimulus) of the same or when the latter precedes or follows the former. When the test stimulus follows the masking stimulus the effect is called forward masking and when it precedes it the term backward masking or the Crawford effect is used. The interval between the onsets of the two stimuli is referred to as stimulus onset asynchrony (SOA) and is conventionally positive for the sequence masking stimulus-test stimulus and negative for the opposite order. greater luminance

Contraction and Disappearanceof Moving Objects in the Visual Field

961

The degree of masking expressed in terms of test stimulus threshold varies directly with the luminance of the masking stimulus and inversely with SOA (BAKER,1953; BOYNTON, 1961). For the same SOA forward masking is greater than backward masking (BOYNTON, 1961; CRAWFORD,1947). Moreover, while forward masking occurs with SOAs up to about 700-800 msec the backward effect does so for SOAs up to about 100 msec beyond which there is a slight facilitation effect sometimes called backward sensitization (KAHNEMAN, 1968). With a rotating arc of uniform luminance as used by ANSBACHER (1944) it is reasonable to suppose that the leading section causes rapid light adaptation as described originally by SCHOUTEN and ORNSTEIN(1939) with a consequently elevated light threshold as found by BAKER (1953), BLANCHARD(1918), BOYNTON(1961) and CRAWFORD(1947). This fast threshold rise could in turn be expected to mask the trailing section of the arc so that in perceptual terms it would be apparently shorter than when stationary. If masking by light is a factor contributing to apparent contagion then it would be expected that the effect would be enhanced with a region of greater intensity leading, the condition for forward masking. A similar con~action, but smaller in magnitude, would be expected to occur with a region of lower luminance leading, the condition for backward masking. These predictions were tested in Experiment 1 using stimulus arcs with a continuous gradient of luminous intensity throughout the length of the arc and one arc of uniform luminance. At 1 r.p.s. the 36” arcs took 100 msec to travel their own length, a duration within the boundaries for both forward and backward masking. Stimulus arcs

The stimulus objects were arcs 1 cm wide with an outer radius of 13 cm cut from black disks as described earlier. The arcs were 8.2 cm on the outer edge and subtended 36” at the centre of the disk. An optical density gradient following approximately a logarithmic function was prepared by making a photographic print through a graded neutral density wedge (Barr and Stroud, Type WGN) and then making a film transparency from the print. Identical sections of this transparency were then cut and mounted over

Position

on stimulus

arc

1. Three continuous luminance gradients used in Experiment 1. I-I: region of higher intensity of arc leading, L: region of lower intensity leading, U: uniform intensity throughout.

FIG.

R. H. DAY

962

the arcs on two of the disks SO that for one the section of least density was in the lead for clockwise rotation and for the other the section of greatest density. A film transparency of uniform density made from the same film was cut and mounted over the arc in a third disk. The luminance of the stimulus arcs was measured at points separated by 5” around the arc with the disks in position. The luminance gradients of the three arcs are shown in Fig. 1. Experimental design and observers The 3 conditions of gradient, higher luminance leading, lower luminance leading, and uniform luminance, were presented in clockwise and counterclockwise rotation (1 r.p.s.) and each while stationary resulting in 9 conditions. There were 2 trials under each condition making 18 in all. Each of 12 observers, 6 men and 6 women, underwent the 9 pairs of trials in a different random order with a 2 min rest between each pair.

Results and discussion

A preliminary analysis of mean apparent lengths showed that differences between clockwise and counterwise conditions were small and statistically insignificant. Therefore the means for the two directions have been combined so that each is based on 4 trials. The mean apparent lengths of the three stimulus arcs while rotating and stationary are shown as histograms in Fig. 2 together with standard deviations. It can be seen that the apparent length of the stationary arcs corresponded closely with their physical extents. The data suggest that the rotating arcs exhibited apparent contraction, markedly so in the case of the arc with the area of higher luminance leading which contracted to about 49 per cent of its stationary length. The arc with the dimmer leading section contracted to about 79 per cent and the uniform arc to 84 per cent of its stationary apparent length. In general these n

Rotating

Higher intensity leading

f3Staticnory

tower intensity leading

Uniform intensity

Conditions

FIO. 2. Mean apparent length of 36” stimulus arca and standard deviations for three conditions of continuous luminance gradient (see Fig. 1) when rotating at 1 r.p.s. and when stationary.

963

Contraction and Disappearance of Moving Objects in the Visual Field

observations have been confirmed statistically. An analysis of variance (HAYS, 1968) has shown that there is a significant difference between the means for the stationary and rotating arcs, no such difference between the means for the three stationary arcs, and a significant

difference between those for the three rotating arcs. Further analysis using t-tests has confirmed that the apparent lengths of the three arcs while stationary do not differ significantly from their physical lengths and that the lengths of the arcs in motion are each significantly less than when stationary. That is, each moving arc underwent apparent contraction relative to its stationary judged length. Finally, while the apparent length of the moving arc with the area of higher luminance leading was significantly less than the other two in motion, those two were not significantly different from each other. The expectancies in terms of degree of masking for the three conditions have been generally realized. By far the greatest contraction occurred when the leading section of the stimulus arc was of greater intensity. When the area of higher luminance trailed significant contraction still occurred, although it was less than for the opposite gradient as would be expected for backward masking. Since forward masking would not be expected to occur with the area of lower luminance in the lead and backward masking would not do so with uniform luminance, it can be reasonably concluded that the similar degree of perceptual shrinkage for these two stimulus gradients is coincidental. It can be noted that with variations in angular velocity in Experiment 2 a difference between these two conditions emerged. EXPERIMENT

2-CONTINUOUS ANGULAR

LUMINANCE VELOCITY

GRADIENTS

AND

The test stimulus threshold rise associated with forward masking is greatest when SOA is between 0 and about 50 msec but is observable up to about 800 msec. However, backward masking is restricted to an SOA of about 100 msec beyond which sensitization (reduced threshold) has been reported. It follows that if apparent shrinkage is due to masking by light then forward masking will occur at much lower angular velocities than backward masking. Moreover, backward masking would not be apparent until a velocity at which the

arc takes about 100 msec to travel its own length. For an arc of 36” this velocity is 1 r.p.s. Stimulus arcs

The same stimulus arcs as described for Experiment 1 (Fig. 1) were used. However, since the data from the first experiment showed that direction of rotation was not a significant variable the arcs were presented only in counterclockwise rotation. The stimulus velocities were 0 (stationary control), 0.5, 1, 1.5 and 2 r.p.s. Experimental design and observers Each of 6 observers, 3 men and 3 women, none of whom had participated in the first experiment, matched the apparent length of the standard at each of the 5 angular velocities. With 3 stimulus arcs and 5 velocities there were 15 conditions. Each observer matched apparent length twice in succession under each condition, the 15 pairs of trials being undertaken in a different random order. There was a 2 min rest between each pair of trials.

Results and discussion Mean apparent length as a function of angular velocity between 0 and 2 r.p.s. is shown in Fig. 3 for the three conditions of luminance gradient, each point representing 12 matches, 2 by each observer. To avoid obscuring trends the standard deviations are not shown on the same graph. However, these were of about the same order for the rotating and stationary conditions as in Experiment 1 (see Fig. 2).

R. H.

964

t 0

I

O-5 Angular

DAY

/

10 velocity,

,

15

20

r.p.s.

FIG. 3. Apparent length of 36” stimulus arcs as a function of angular velocity for the three conditions of luminancegradient shown in Fig. 1. The functions have been fitted by the least squares method. Trend analyses (EDWARDS, 1968) have shown that all three trends are significant and that both the linear and quadratic components contribute to each. In addition, each trend is significantly different from the other two. Generally, the functions shown in Fig. 3 accord well with expectancy. For the forward masking condition with the more intense region of the arc in the lead there was noticeable apparent contraction at 0.5 r.p.s. whereas with the less intense section leading, the backward masking condition, contraction occurred only at higher velocities. It is also worth noting that between 0 and 0.5 r.p.s. the backward masking arrangement possibly gave rise to a slight degree of apparent expansion, an effect consonant with backward sensitization (KAHNEMAN,1968) for SOAs between about 100 and 150 msec. Although the degree of contraction for the arc of uniform luminance was less than that found earlier by ANSBACHER (1944) the general form of the function is essentially similar with only slight contraction at 0.5 r.p.s. but a steeply increasing effect for higher velocities. The difference in magnitude of effect between the data for the uniform arc reported here and Ansbacher’s results is possibly due to differences in luminance level and peripheral angle. EXPERIMENT 3-STEP-FUNCTION GRADIENTS Step-function gradients of luminance offer an alternative means of establishing whether masking due to threshold elevation is a factor in the apparent contraction effect. Furthermore, such gradients provide a means of substantiating the results and conclusions of the first two experiments. In addition, because of the more precise specification of areas of high and low luminance which step-functions allow it is possible to explore in more detail the

Contraction

and Disappearance

of

Moving Objects in the VisualField

965

effects of different proportions of each in the stimulus arc. However, as will be noted below, the regions of high and low luminance do not necessarily correspond with masking and test regions. The ~stribution of the two levels of luminance and the direction of rotation were varied systematically SO that the effects of both forward and backward masking could again be observed. Stimulus arcs The 36” arcs were arbitrarily divided into long (30”) and short (6”) sections with higher (1.5 cd/m’) and low (1.5 cd/mz) luminance assigned to each, thus r~ulting in 4 conditions, The step-functions were prepared by mounting gelatin neutral density filters of appropriate density over the arc on the reverse side of the disks. Care was taken in cutting the filters to ensure that the edges between the filters of different density fitted closely so that there was no visible band of light between them. Rotation was counterclockwise throughout. During rotation at 1 r.p.s. the time for which a given retinal element was stimulated by the short (6”) section of the stimulus was approximately 16 msec and by the long (30”) section approximately 84 msec. Experimental des&n and observers The 4 stimulus arcs were each observed and matched in length as described earlier during 2 trials when stationary and when rotating at 1 r.p.s. The 8 pairs of trials were undertaken in a different random order by each of 12 observers. Three observers had participated in one of the earlier experiments.

The mean apparent lengths of the moving and stationary arcs are shown as histograms with the standard deviations. An analysis of variance showed that the differences in apparent length between the four stationary arcs do not achieve statistical significance but those between the rotating arcs do so. Furthermore, each of the differences between the 0 and 1 r.p.s. conditions are sibilant, indicating that all rotating arcs underwent apparent contraction. Further analysis (HAYS,1968) showed that apparent contraction with the forward masking arrangement (higher luminance leading) was greater than with the backward masking sequence. Finally, in contrasting the magnitude of apparent contraction with the long and short sections of either higher or lower luminance leading it was found that in each case the difference achieved significance, i.e. the degree of shrinkage with a short section leading was different from that with a Iong section leading. The differences between the extent of shrinkage under both forward and backward masking conditions with long and short sections of higher luminance deserve further consideration from the viewpoint of the SOAs involved. With the forward masking arrangement (high luminance leading) the extent of shrinkage relative to the stationary control with the 6” section leading was reliably greater (53.2 per cent) than with the longer section in the lead (46.6 per cent). This difference is expected in terms of the different SOAs between areas of high and low luminance. At 1 r.p.s. the 6” section of the arc would stimulate a retinal point for approximately 16 msec and the 30” section for about 84 msec before the area of lower luminan~ impinged. Threshold elevation is known to be markedly greater for shorter SOAs (CRAWFORD,1947). However, the issue is complicated by the fact that for a reduction of 46.6 per cent in apparent length with the 30” section leading, the earlier part of this section must have masked not only the 6” dimmer section, but also part of its own length. Thus the difference between the two forward masking situations is attributable to differences in both SOAs and the relative luminance levels involved. The difference in apparent contraction between the two backward masking conditions is also in accord with expectancy based on SOAs. With the 30” region of lower luminance leading there was 7.2 per cent contraction relative to stationary length and with the 6” section leading 252 per cent. The in Fig. 4 together

966

R. H.

/

i

Rotating

6” Higher intensity leading

DAY

Stotionory

El

30° Lower intensity leading

j* Lower intensity leading

SOOHigher intensity leading

T

I i

_. : Conditisns

i

FIG. 4. Mean apparent length of 36” stimulus arcs and standard deviations for four conditions of step-function luminance gradient when rotating at 1 r.p.s. and when stationary. For each arc the region of higher intensity was 15 cd/m2 and that of lower intensity 15 cd/m’.

SOA of about 84 msec for the first approaches the limit for effective backward masking while that of about 16 msec for the second is therefore in good agreement with expectancy. EXPERIMENT

4-DISAPPEARANCE

OF A TRAILING

STIMULUS

From the earlier discussion it would be expected that a small area of low intensity following in the path of another of greater intensity under the general conditians described would either appear less bright or disappear altogether. That is, if instead of a complete arc, only the terminal sections in the form of radial slots in the disks were visible, then, depending on angular velocity and relative intensities, only one would be visible due to masking by the other. The use of two separate apertures in the disks and the simple task of reporting whether one or two were visible provides an alternative method of testing the role of masking in apparent contraction. Therefore, an experiment was carried out to test the prediction that at appropriate angular velocities a trailing aperture of lower intensity than that leading it is rendered invisible. During the course of the experiment a review of the hterature revealed that MCDOUGALL (1904) had carried out an essentially similar investigation as that proposed. This early experiment is discussed below. SG~~ulus patterns Pairs of apertures 3” wide and 1 cm in radial length were cut in two black metal disks with the outer contours separated by 36” and the inner contours by 30” so that they were coincident with 3” terminal

Contraction

and Disappearance

of Moving Objects in the Visual Field

967

sections of the arcs used in the previous experiments. Neutral density filters were mounted over each aperture so that when the disk was in position the lnminance of one aperture was 15 cd/m2 and the other l-5 cd/m’. The filters were arranged so that the aperture of greater intensity preceded that of lower intensity for one disk and vice versa for the other. In rotation the angular velocity of both disks was 1 r.p.s. so that a retinal point was stimulated for approximately 85 msec by the moving aperture and the SOA was approximately 91.5 msec. Experimental design and observers There were 3 conditions: rotation (1 r.p.s.) with the higher intensity aperture leading, rotation with the lower intensity aperture leading, and stationary. Each of 10 observers, none of whom had observed previously, served once under each of the 3 conditions which were presented in random order. The observer’s task was merely to report whether one or two apertures were visible.

Results and discussion The mean frequencies with which one or two apertures were reported by the 10 observers are shown in Table 1. These results are clear in showing that with the aperture of greater

TABLE

1. FREQUENCIES

WERE

REPORTED

AND

LOWER

DURING

INTENSITY

WITH

WHICH

ROTATION

ONE AT

APERTURES

1

OR TWO I-.p.S.

APERTURES

WITH

LEADING

AND

HIGHER WHEN

STATIONARY

Condition Rotation with higher intensity leading: Rotation with lower intensity leading: Stationary: Total :

One

Two

10 2 1

0 8 9

10

10

intensity in the lead only a single aperture was visible to ah 10 observers, whereas with the lower intensity slot leading 8 of the 10 reported two, compared with 9 reporting two when the disk was stationary. It can be confidently concluded therefore that the leading aperture of greater intensity masked the trailing stimulus rendering it invisible. It is worth noting that al1 subjects when carefully questioned and when occasionally asked to make further observations confirmed their reports of a single visible aperture under this condition. Since most observers reported two apertures when the lower intensity preceded the higher it can be concluded that the relationship between the very brief flash of about 8.5 msec followed by the 915 msec “dark period” was not appropriate for producing backward masking of the leading aperture. This result is not surprising in view of the smaller threshold change associated with the backward effect. It is of course conceivable that while the weaker stimulus was not rendered invisible when it preceded the stronger its brightness was reduced, This possible effect was not however investigated. An essentially similar experiment involving 2 rotating apertures but only 1 observer was reported by MCDOUGALL (1904) as one of a series concerned with the effects of intermittent stimulation. McDougall conduded : That the initial images or reactions provoked by the travelhng object-light really exert a strong inhibiting influence upon the retino-cerebral elements over which the image has passed in the preceding moments, and that the gap in the series of images is due to this influence, is

968

R. H.

DAY

proved by the following experiments: two radial slits 3” wide and 7 cm Iong are made in the disc at an angular interval of 20” . _. so that on rotation the slit b follows in the path of slit a. Then on rotation at 1 revi3” before the brightly lit surfa=, b appears much less bright than a; and if then a sheet of white paper is pasted over b so as to reduce the brightness of that slit by rather more than 50 per cent, and the rotation repeated, b is quite invisible so long as the eye remains at rest (pp. 102-103).

EXPERIMENT

5-COMPARISON UNIFORM

OF ARCS

AND

APERTURE

PAIRS

OF

LUMINANCE

Informal observations during the course of Experiment 4 indicated that complete disappearance of the trailing aperture did not occur when the two were of equal luminance, although the second was often reported to be less bright than the first. Presumably conditions of a brief (8.5 msec) stimulation period followed by the 91.5 msec interval combined with equal luminance were insufficient fully to mask the trailing aperture of equal luminance. This finding suggested a further and critical test of the masking explanation. Contraction of an arc and disappearance of a trailing aperture are held to be due to the same process, contraction being conceived of as “disappearance” of the end of the arc. Thus, if during rotation of the two apertures of the same luminance both are visible their apparent distance apart should be the same as when they are stationary, i.e. if there is no disappearance there can be no contraction due to this factor. In the experiment now to be reported an arc of uniform luminance and two apertures, all of the same luminance, were rotated. It was expected that there would be some apparent contraction of the first, as in Experiment 1, but little, if any, of the second, providing the two elements remained visible. An arc 36” long and two radial slots with their outer ends 36” apart and each 3” long were prepared in the same manner as for the eariier experiments. All apertures were 1 cm wide in the radiaf direction. When the disks were positioned the luminance of both arc and radial slots was 12 cd/m’. Experimental design and observers Each of 10 observers, two of whom had served in an earlier experiment, matched the length of the arc and the distance between the extremities of the radial slots during two successivetrials when the patterns were in rotation and when they were stationary. Each observer was presented with these four conditions in

a different random order. The score for each observer was the mean apparent length (or separation) based on the pair of trials.

It can be noted first that ail observers matched the distance between the ends of the two rotating slots without difficulty indicating that both apertures were visible. The mean apparent length of the arcs and mean apparent separation between the slot extremities, together with standard deviations, are shown in Fig. 5 for the stationary and moving conditions. Inspection of this figure shows that during rotation t;he arc underwent apparent contraction of about 12 per cent relative to its stationary apparent length. However, the apparent distance between the slots showed a slight increase during movement to about 6 per cent of that of the stationary control. An analysis of variance has shown that the difference in apparent length between the complete arc and the slots representing its extremities is not significant for the stationary condition but achieved significance for the rotating condition. Further analysis has shown that whereas the difference in apparent length between the rotating and stationary arc is signi~cant that between the radial slots for the same two conditions is not. Thus, the 12 per

969

Contraction and Disappearance of Moving Objects in the Visual Field

q

Ratating

Single

arc

HStationary

Two radio1 apertures

Conditions

FIG. 5. Mean apparent length of stimulus arcs and mean apparent distance between the outer ends of two radial slots when rotating at 1 r.p.s. and when stationary. The actual length and distance were 36” and the luminance of all stimuli 12 cd/m2.

cent reduction in apparent length of the rotating arc is significant but the 6 per cent increase in separation between the ends of the slots does not achieve significance. It can be concluded from these results that apparent contraction does in fact represent in large part the disappearance of the end-sections of the arc. When these extremities remain visible, as with the two 3” apertures coincident with the ends of the arc, then no significant change in length occurs during rotary movement. EXPERIMENT

6-APPARENT

CONTRACTION

AND ECCENTRIC

ANGLE

In earlier reports of the Ansbacher contraction effect there is an implicit assumption that it is characteristic of stimulus objects moving in the peripheral visual field. However, so far there has been no report of a systematic investigation of the effect as a function of eccentric angle, i.e. the angle between the fixation axis and that of the stimulus object. In the final experiment of this series the eccentric angle between the axis of tiation and the circular path of movement was varied within the range 6-18” in approximately 4” steps. In varying eccentric angle the manner of controlling stimulus velocity and the visual angle subtended by the arc must be considered. If angular velocity is constant then linear

velocity varies as the circumference of the path of movement changes. In addition, if the length of the arc is constant with change in eccentricity, then its length relative to that of its path must also vary. In an investigation of the role of masking following initial threshold

R. H. DAY

970

elevation it can be argued that maintaining stimulus duration is the first consideration. For this reason angular velocity was constant at 1 r.p.s. and arc length constant at 36” of circumference.

Four arcs 36” of circ~ference long and 1 cm wide were cut in black metal disks so that their outer edges were 82, 70, 50 and 32 mm long and 141, 111, 82 and 51 mm from the centre respectively. When in position and viewed from the standard viewing distance of 40 cm the eccentric angle of the arcs at points midway between outer and inner edges were 18” 36, 14” 42, 10” 36’ and 6” 24’. The luminance of all arcs was 8.2 cd/m2. Angular velocity during rotation was constant at 1 r.p.s. Experimental

design and obseraers

The 4 arcs were observed and matched for length as described earlier while rotating and stationary so that there were 8 conditions. Each of 12 observers, none of whom had participated previously, underwent these conditions in a different random order. There were 2 successive trials for each condition. Results

and discussion

Although all arcs subtended 36” at the centre of the disks their linear extents varied between 82 and 141 mm. In order to render apparent contraction comparable over the 4 conditions it has been expressed as a percentage of the apparent length of the stationary arc. Mean per cent contraction as a function of eccentric angle is shown in Fig. 6. On inspection these data indicate that while apparent shrinkage at eccentric angles of 14” 42’ (12-5 per cent) and 18” 36’ (156 per cent) is of about the same order found earlier for uniform arcs (Experiment 1, Fig. 3) that at angles of 6” 24’ (3.8 per cent} and 10” 36’ (3.2 per cent) is considerably less. An analysis of variance has shown that overall differences in apparent shrinkage during rotation between different degrees of eccentricity are significant. However,

0

I

/

1

5

IO Angle

of

I_.

15

20

eccentricity

Ftc. 6. Per cent apparent contraction of rotating stimulus arcs as a function of visual eccentric angle relative to the visual fixation axis. The arcs were of constant length (36” of disk circumference) and rotated at constant angular velocity (1 r.p.s.). Luminance was 8*2 cd/m3.

further analysis showed that while contraction at the two greater eccentric angles is significant that at the two smaller angles is not. It is worth noting that for the eccentric angles of 6” 24’ and 10” 36’ apparent contraction occurred with only about half the observers, the remainder showing either no effect or apparent lengthening. However, at 14” 42’ and

Contraction and Disappearance of Moving Objects in the Visual Field

971

18” 36’ contraction occurred for 10 and 11 observers out of the 12 respectively. It must be concluded, therefore, that for rotating arcs of uniform luminance apparent contraction occurs only at relatively large eccentric angle, at least in excess of about 11-12”. The implications of these findings are discussed below. CONCLUSIONS It will be useful first to summarize the results from the six experiments. Experiment

1 showed that while a rotating 36” stimulus arc of uniform luminance exhibited about 16 per cent apparent contraction relative to its stationary apparent length, one with a continuous near-logarithmic gradient of intensity from end to end showed about 51 per cent contraction with the greater intensity in the lead and about 22 per cent with the lower intensity in the lead. When the angular velocity varied between 0 and 2 r.p.s. in 0.5 r.p.s. steps (Experiment 2) apparent contraction in length occurred at 05 r.p.s. with the greater intensity leading but did not do so until a velocity of 1 r.p.s. with the area of lower intensity leading and when the arc was of uniform luminance. In Experiment 3 ste~function luminance gradients were varied and, in general, the data of Experiment 1 were confirmed. With a 6 or 30” section of greater intensity in the lead apparent contraction (53 and 47 per cent) was substantially greater than that (25 and 7 per cent) when similar sections of lower intensity led. Greatest contraction (53 per cent) occurred when the shorter section of higher intensity led and least (7 per cent) when the larger section of lower intensity did so. Experiment 4 showed that when only the end sections of the 36” arc in the form of two 5” radial slots rotated with one of higher luminance than the other that of lower intensity was invisible when it trailed that of higher intensity. Generally, both apertures were visibfe when they were stationary and when that of higher intensity trailed that of lower intensity. The results from Experiment 5 confirmed an informal observation that when two radial slots of equal luminance, coincident with the end sections of a 36” arc, were both visibIe during rotation at 1 r.p.s. no apparent contraction occurs. However, a complete arc of the same uniform luminance exhibited apparent contraction of about 12 per cent during rotation. In Experiment 6 the eccentric angle between the visual fixation axis and the circular motion path of the rotating arc was varied between about 6 and 18” with the angular length and velocity of the arcs constant at 36” and 1 r.p.s. respectively. The results showed that while at eccentric angles of 6” 24’ and 10” 36’ apparent contraction relative to stationary controls was negligible (3.8 and 3.2 per cent respectively) that at 14” 42’ and 18” 36’ (12.5 and 15.6 per cent) was significant. The case for the involvement of a masking process in the apparent contraction of rotating stimulus arcs and the disappearance of one of a pair of rotating apertures is generally well supported by the data. Those conditions which are known to give rise to the greatest degree of masking as measured by the test stimulus threshold also produce the greatest degree of apparent contraction. It was pointed out in the introduction that a rotating stimulus arc produces similar stimulus conditions to two discrete stimulus flashes; one part of the arc stimulates a retinal region slightly ahead in time of another. Therefore, depending on the relative luminance intensity of the arc from the leading to the trailing end, so some forward or backward masking of a terminal section would be expected with consequent apparent contraction. However, as will be noted below, the involvement of a moving stimulus, as opposed to discrete stationary flashes, may be of considerable significance in terms of the retinal locus of masking. Three points which will be discussed in greater detail are the nature of the processes

972

R. H. DAY

involved in the visual suppression of part of the rotating stimulus, the significance of the variation in apparent contraction as a function of eccentric angle, and the contribution of processes other than masking to apparent contraction, This latter point would seem to be central to anv consideration of the mechanisms involved in visual masking by light. A fast transient rise in the light threshold following a brief flash, originally called alpha adaptation by SCHOUTEN and ORNSTEIN(1939), with a subsequently elevated “instantaneous threshold” at offset have been observed and systematically investigated by BAKER (1953), BLANCHARD (X918), BOYNTON(19611, CRAWFORD (1947) and others. The occurrence of these rapid changes within 100 msec after stimulation has been taken as evidence for a mainly neural rather than a photochemical basis for the effect by BAKER (1953) and RIGGS (1971). Of immediate interest is whether the masking of a stimulus of the same or less intensity is due to sheer elevation of the threshold by the masking flash or to more complex and subtle interactions between the two. KAHNEMAN (1965, 1968) has drawn attention to the effect of masking by light on the formation of bounding contours. It is possible that since in the experiments reported here the bounding contours of the “masking” and “test” sections of the rotating arcs and of the radial apertures were perfectly coincident retardation of contour function as well as invisibility of the area due to threshold elevation may have contributed to the final outcome in contraction. However, apart from drawing attention to effects other than threshold elevation which might have contributed to the apparent shrinkage of the arc and the invisibility of one member of a rotating stimulus pair, it is not possible on the evidence reported to state which of these particular processes are involved. Clearly, however, the roles of contour formation and reduced contrast due to threshold elevation in apparent contraction demand further investigation. Using essentially the same procedures as reported it would not be unduly difficult to arrange a stimulus so that the contours of the leading and more intense section are non-coincident with those of the trailing section and thus tease out the role of contour processes. The occurrence of the apparent contraction of the rotating stimulus arcs only at relatively large eccentric angles, at least in excess of about 1I”, immediately raises the question of whether the structures and mechanisms are the same as those associated with the typical two-flash stimulus arrangement. Under the usual conditions for the masking of one flash by another the two stimuli fall on the central retinai region with the subject fixating a point at the centre of the stimulus pattern. An obvious and potentially important difference between the stimulus conditions in a typical masking experiment and those reported here is the involvement of moving stimuli in the present series. In a typical masking experiment the stimuli are stationary, discrete and separated in time. In the experiments reported here the stimuli were moving and, in the case of the arcs, continuously visible. Thus the moving stimuli were neither discrete nor temporally discontinuous. It is conceivable that the masking associated with apparent contraction and disappearance of moving stimuli involves units which, in addition to responding to movement, have their greatest concentration in the periphery of the retina. That is, while the masking process itself may be essentially the same as for more centrally located structures when stimulated appropriately by two successive flashes the structures themselves are peripheral. There is support for this view from the recent classi~cation of retinal gangliOn cek!s into"X-units" and “ Y-units” by ENROTH-CUGELL and ROBSON(1966), renamed “sustained” and “transient” cells respectively by CLELAND,DUBIN and LEWK (1971). The Y-unit or transient cells are of principal interest here. CLELAND,DUBIN and LEVICK(1971) have shown that the transient cell is distinguishable

Contraction and Disappearanceof Moving Objectsin the Visual Field

973

from the sustained unit in terms of four criteria: its response to standing contrast, to the size of the stimulus, to grating patterns of varying frequency and its response to stimuli in the periphery. This last characteristic is especially important here in that it refers to the excitation of mainly transient cells by visual stimuli applied well away from the conventionaIIy defined receptive field. The “peripheral response” is elicited by continued movement of any large pattern falling on a retinal region 15” or more from the centre of the receptive field. CLELAND,DUBIN and LEVICK(1971) suggest that while the sustained (X-type) cells are capable of signalling steady local differences of illumination the transient (Y-type) cells constitute an initial stage in the development of specific sensitivity to movement. Such a function accords well with the data reported here. It is conceivable that by means of a masking process the Y cells serve to enhance the visibility of objects moving in the peripheral field by increasing brightness (STANLEY,1967) at the expense of size. The total light is effectively restricted to a smaller area during peripheral movement. While the association between Y-type transient cells and the mainly peripheral masking phenomenon response here is speculative the fact that the cells are distinguishable on alternative criteria offers a basis for further psychophysical and neurophysiological enquiry. Finally, while the results from these experiments strongly suggest that masking by light following a rapid threshold rise is involved in the apparent contraction of the rotating arcs, it is not maintained that this form of masking is the onIy process contributing to the effect. In addition to other classes of masking effect it is conceivable that other neural interactions are involved.

REFERENCES investigation of the Harold C. Brown shrinkage phenomenon: A new approach to the study of the perception of movement. Psychol. Bull. 35,701. ANSBACHER,H. L. (1944). Distortion in the perception of real movement. J. exp. Psychoi. 34,1-23. BAKER,H. D. (1953).The ins~~t~~~ threshold and early dark adaptation. f. opr. Sot. Am. 43,798-803. BLANCHARD,J. (1918).The brightnesssensibilityof the retina. Phys. Rev. 11, 81-99. BOYNTON,R. M. (1961). Some temporal factors in vision. In Sensory Communication (edited by W. A. ROSENBLITH),Wiley, New York. CLELAND,B. G., DUBIN, M. W. and LEVICK, W. R. (1971). Sustained and transient neurons in the cat’s retina and lateral geniculate nucleus. J. Physiol., Land. 217,473-496. CRAWFORD,B. H. (1947). Visual adaptation in relation to brief conditioning stimuli. Proc. R. Sot. B. 134, 283-302. EDWARDS,A. L. (1968). Experimental Design in PsychoIogicaI Research, Holt, Rinehart and Winston, New York. ENROTH-CUGELL,C. and ROBSON,J. G. (1966). The contrast sensitivity of retinal cells of the cat. J. Physiol., LoFzd.187,517-552. HAYS, W. L. (1963). St~t~tics~~ Psychoiog~ts, Holt, Rinehart and Winston, New York. KAHNEMAN,D. (1965). Exposure duration and effective Iigure ground contrast. Q. J. exp. Psychol. 17, 308-314. KAHNEMAN,D. (1968). Method, findings, and theory in studies of visual masking. Psychol. Bull. 70,404-425. MCDOUGALL,W. (1904). The sensations excited by a single momentary stimulation of the eye. B. J. Psychol. ANSBACHER,H. L. (1938). Further

1,78-113.

MARSHXL, A. J. and STANLEY,G. (1964). The apparent length of light and dark arcs seen peripherally in rotary motion. Austrui. J. Psychol. 16, 120-128. RIGGS, L. A. (1971). Vision. In: Woodworth and Schlosberg’s Experimental Psychology (edited by J. W. KLING and L. A. RIGGS), Methuen, London. SCHOUTEN,J. F. and ORNSTEIN,L. S. (1939). Measurements on direct and indirect adaptation by means of a binocular method. J. opt. Sot. Am. 19,168-180. STANLEY,G. (1964). A study of some variables influencing the Ansbacher shrinkage effect. Acta. Psychol. 22, 109-118. STANLEY,G. (1966). Apparent

shrinkage of a rotating arc as a function of Iuminance relations between figure and ground. Acta. Psychol. 25, 357-364.

974

R. H. DAY

STANLEY,G. (1967). Apparent brightness of a rotating arc-line as a function of speed of rotation. Acta Psych&. 26, 1J-21. STANLEY, G. (1968). Apparent length of a rotating arc-line as a function of speed and rotation. Acta Psychol. 28.398403. STAN&Y, G., FINLAY,D. C. and BARTLETT,W. K. (1971). Regions of brightness and darkness in the sequential presentation of partially overlapping straight lines. 3. exp. Rrychol. 88, 314-318. STANLEY,G. and JACKSON,R. (1969). The effect of leading contour on the relative lengths of moving light and dark arcs. J. Psychol. 71, 83-88.

Abstract-A light arc rotating at between 03 and 2 r.p.s. in the darker peripheral field undergoes apparent contraction and a radial aperture following in the path of another of higher intensity is not visible. Six experiments were designed primarily to establish whether the apparent contraction and disappearance are due to visual masking resulting in the suppression of terminal sections of the arc or one aperture. By using arcs with continuous and step-function gradients of luminance it has been shown that forward masking of the trailing section and backward masking of the leading section of the arc occurs and that both forms of masking are dependent on angular velocity. Disappearance of one of a pair of apertures can also be attributed to masking. The contraction effect was shown to be largely restricted to the periphery of the retina at eccentric angles greater than about 11’. It is suggested that the masking process associated with apparent contraction and disappearance of moving stimuli in the periphery is attributable to the recently reported “Y-type” or “transient” ganglion cells.

R&urn&-Un arc lumineux tournant entre 0,s et 2 tours par set dans le champ peripherique sombre subit une contraction apparente et une ouverture radiale continuant une partie d’intensite plus forte n’est pas visible. On imagine six experiences pour determiner si la contraction apparente et la disparition sont un masquage visuel dQ a la suppression des sections terminales de l’arc ou dune ouverture. En employant des arcs a gradients de luminance continus ou en escaliers, on demontre le masquage vers l’avant de la section terminale et le masquage vets l’arriere de la section anterieure, ces deux masquages d&pendant de la vitesse angulaire. La disparition dune ouverture dans une paire est aussi attribuable au masquage. On montre que l’effet de contraction est surtout restreint a la p&ipherie retinienne, pour des angles d’excentricite d&passant 11”. On suggere que les processus de masquage associb a la contraction apparente et a la disparition de stimuli mobiles dans la peripherie sont attribuables aux cellules ganglionnaires du “type Y” ou “transitoi~” d&crites r~emment.

~~~~-Ein routierendes Lichtse~ent mit 0,5 bis 2,0 Umdrehungen pro Sekunde zeigt im dunkleren, periphereren Feld eine offensichtliche Verkleinerung, und wenn ihm ein zweites Segment mit hiiherer Intensitlt folgt, so ist es nicht sichtbar. Sechs Versuche wurden vorgenommen, in erster Linie urn festzustellen, ob die offensichtliche Verkleinerung und das Verschwinden des Lichtsegmentes auf visuelle Masking-Ergebnisse bei der Suppression der Siusseren Segmentteile oder der des Zwischenraumes zuriickgefiihrt werden kann. Wenn Segmente mit kontinuierlichen und stufenf~~igen Lichtgradienten benutzt wurden, zeigt es sich, dass das Vorw~~s-Masking der nachfolgenden Abschnitte und das R~ckw~rts-Masking der ftihrenden Abschnitte des Segmentes vorkommt und dass beide Formen des Maskings abhlngig sind von der Winkelgeschwindigkeit. Das Verschwinden eines Paares von Zwischenr&umen kann ebenso auf Masking zurtickgeftihrt werden. Der Verkleinerungseffekt war weitgehend auf die Peripherie der Netzhaut verlagert und bei exzentrischem Winkel grosser als 11 Grad. Es wird vermutet, dass der Masking-Props, verbunden mit der offensichtlichen Verkleinerung, und das Verschwinden von bewegten Reizen in der Peripherie, dem ktirzlich beschriebenen “Y-Typ” oder den “Transient”-Ganglienzellen zugeschrieben werden kann.

Contraction

and Disappearance

of Moving

Objects

in the Visual Field

PesowCee~qcx

ma B 6onee mit~enb~o~ nepwJepwe4xcob4 none c nepHoxoh4 apaugem OT 0,5 no 2 06lcerr nepqemiario yxopawraaema, npmer4 pamumaa anepqpa, Crmamipoaaao 6 cnenyrorsaa~Apymg;mcprypbg,~npirag,~e~~.

mcnepmiemoa m Toro, YTO~M ~wmmb, B nepaym osepe~acwmcn m coxpa~~eme wfa H mie3Hoaeme anepTypbz *EIoM hmmipo~m, 303mnuuome& Bcmwmzfe nommemm KOHIlOB JJyrE mm OJuiOii anepTypbx. EiclI~SOearm EyrA c pa3HOB IIpKOCTbIO, a3hiemomek~ m60 HenpepblBAo, J&O maram. lTo~a3asi0, PTO rrpo~cxo~~ xax ameMaCXEpOBKa CJleJlyiOIQe@YaCTIi JIJTZi,TZlK Ii ~TpO-M~CKE~BK~ Bt?$lyl@ VWTH jIJ’l-E, IIpaYeM 06t3 BB.JWi MaCKSipOBKH3aBXCIIT Of ~JIOOBOfi CKOpoCrE. MaCKKpOBKe XX MOXHO I-lpKIIKCaTb mzYe3noaeme omozt x3 n14yx anepTyp. Ilorasawo, YT~ coxpameme ,nym orpamiueao, 06mo nepa&pae@ ceTsam (ymbz 6onee 9ew, npmepH0, 1lo). llperpionaraeTcx, 910 BblmmeHHal MaCKIipOBICaMOXeT 6bmb OTHeceHa 38 CYeT HeAaBHo orrpbmlx rammio3Hblx Tm0~ “transient” kimi “nepexomib~x”.

V.R.

13/5-F

975