- Email: [email protected]
Infants' sensitivity to uniform motion
Pergamon
0042-6989(95) 00216-2
Vision Res., Vol. 36, No. 11, pp. 1633-1640, 1996 Copyright © 1996 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0042-6989/96 $15.00 + 0.00
Infants' Sensitivity to Uniform Motion TOM BANTON,*t BENNETT I. BERTENTHAL* Received 20 January 1995; in revised form 26 July 1995
Uniform motion across the retina is a powerful cue to the perception of self-motion. In spite of its importance for adaptive functioning, little is known about the early development of uniform motion sensitivity. Six-, 12-, and 18-week-old infants viewed random-dot kinematograms depicting leftward or rightward uniform motion. The display induced optokinetic nystagmus (OKN), which a trained observer used to judge the direction of target motion. Both speed of motion and directional coherence were varied to obtain independent motion detection thresholds. Infants of all three ages could detect uniform motion, and their detection thresholds were constant during this period of development. This is in contrast to the clear improvements in relative motion sensitivity noted previously between 6 and 18 weeks of age with a preferential looking (PL) paradigm. The developmental differences between these studies may result from: (1) separate mechanisms for detecting uniform (absolute) and differential (relative) motion; or (2) separate mechanisms underlying OKN and PL response measures. Copyright © 1996 Elsevier Science Ltd.
Development
Infant vision Uniform motion
Relativemotion
INTRODUCTION Recent evidence from neurophysiological, neuropsychological, and psychophysical studies converge to suggest the existence of two visual motion mechanisms; one sensing uniform motion of the retinal image (as occurs with eye movement or when the entire visual field is moved) and one sensing motion discontinuities in the retinal image (as found with motion parallax, for example). Bridgeman (1972) referred to these as "absolute" and "relative" motion mechanisms respectively, and demonstrated their existence neurophysiologically in area V1 of the rhesus monkey. This dissociation is also supported through differential responses to uniform and relative motion in the pigeon tectum (Frost, 1978; Frost et al., 1990), the cat superior colliculus (Sterling & Wickelgren, 1969), lateral geniculate nucleus (Gulyas et al., 1987), lateral suprasylvian gyrus (von Grunau & Frost, 1983), area 17 (Orban et al., 1987b; Gulyas et al., 1987; Hammond & Smith, 1982, 1984; Hammond et al., 1986), area 18 (Orban et al., 1988), monkey superior colliculus (Davidson & Bender, 1991), and extrastriate cortex (Allman et al., 1985; Orban et al., 1987a). Regan et al. (1992) confirmed the separability of uniform and relative motion perception in persons with tempero-parietal lesions. These patients showed an inability to detect and discriminate motiondefined form, but had normal sensitivity to uniform motion. Psychophysical measures in visually normal
observers also point to a distinction between uniform and relative motion sensitivity (Smeets & Brenner, 1994; Snowden, 1992). It is believed that one function of the dissociation is to enable the visual system to distinguish eye movement from object movement (Bridgeman, 1972; Frost et al., 1990). If uniform and relative motion mechanisms are indeed distinct, then they will develop independently, perhaps following different timecourses. Many of the initial developmental investigations of motion sensitivity demonstrated that infants preferred moving to stationary targets (Volkmann & Dobson, 1976; Kaufmann et al., 1985; Freedland & Dannemiller, 1987). Furthermore, these studies suggested that motion detection improved with age. Although some of these stimuli left open the possibility that flicker-sensitive or positionsensitive mechanisms, rather than motion-sensitive mechanisms, were involved (Aslin & Shea, 1990; Braddick, 1993; Nakayama & Tyler, 1981), studies by Aslin and Shea (1990) and Dannemiller and Freedland (1989, 1991) confirmed the earlier findings and showed that flicker- and position-sensitive mechanisms were unlikely to account for infants' motion preferences. Thus, improvements with age were attributed to the development of motion mechanisms. Recently, the development of motion sensitivity has been scrutinized more closely by looking at particular types of motion. In a study of relative motion, Bertenthal and Bradbury (1992) measured infants' sensitivity to the shearing motion of a random-dot sinusoid of 0.16 c/deg. Motion detection improved from 3.5 deg/sec at 13 weeks to 1.2 deg/sec at 20 weeks of age. Wattam-Bell (1992) reported significant improvement in motion-defined square wave detection between 11 and 16 weeks of age. Banton and Bertenthal (1995) found that size thresholds
*Room 171, Gilmer Hall, Department of Psychology, University of Virginia, Charlottesville VA 22903, U.S.A. tTo whom all correspondence should be addressed [Email [email protected]]. 1633
1634
T. BANTON and B. I. BERTENTHAL
for detecting a motion-defined rectangle could not be measured at 8 weeks of age, but showed a significant improvement between 12 and 16 weeks of age. Finally, Giaschi and Regan (1985) showed that mechanisms for motion-defined letter recognition (a high spatial frequency relative motion task) did not reach adult levels until 7-9 yr of age. Taken together, these studies show that relative motion sensitivity improves steadily from at least 11 weeks of age through 7-9 yr of age. It is unclear if this same pattern of development is evident in the few studies investigating infant sensitivity to uniform motion. Hamer and Norcia (1994) used visual evoked potentials (VEPs) to measure infants' thresholds to oscillatory uniform motion of a grating. Motion thresholds were unchanged between 7 weeks and 1 yr of age, suggesting a different developmental pattern than previously found with preferential looking to relative motion. Mikami and Fujita (1992) used preferential looking to investigate the development of uniform motion detection in rhesus and Japanese monkeys. In this case, linearly plotted thresholds for detecting a moving grating improved rapidly between 1 day and 2 weeks of age, after which thresholds began to plateau. After applying a four-to-one correction for life span differences between monkey and man (Boothe et al., 1985), the data predict that human uniform motion detection may improve between 4 days and 8 weeks of age, and might begin to plateau by 8 weeks of age. However, Wattam-Bell (1991) found an insensitivity to oscillatory uniform motion of a random-dot field in human infants prior to 8 weeks of age, consistent with the view that infants are poor at sensing motion prior to this age (Braddick, 1993). Recently, Jouen et al., (1996) found that newborns can make motor responses scaled to the velocity of peripherally presented gratings that move in depth. Although these results suggest that uniform motion sensitivity develops early and then plateaus, it is difficult to establish the age at which this phase occurs, given the differences in methods, species, and stimuli. The purpose of the present study was to conduct a behavioral test of the development of human infants' sensitivity to uniform motion between 6 and 18 weeks of age, a time period when the onset of the developmental plateau may occur. It was anticipated that the results of this study would clarify how human sensitivity to uniform motion changes with age, and would facilitate comparison between the development of uniform and relative motion sensitivity.
METHODS
Subjects A total of 177 infants at 6, 12, and 18 weeks of age (_+1 week) participated in the study. Infants were recruited from birth announcements in the local newspaper. All infants who completed at least 10 trials per condition were included in the final analysis (n = 121). Fifty-six of the 177 infants completed an inadequate number of trials due to sleeping, fussing, or inattentive-
ness, and were not included in the analysis. Six adults from the Laboratory for Infant Studies also participated. One adult observer (author TB) was aware of the purpose of the study.
Stimuli Random-dot kinematograms were used as stimuli. The kinematograms were 22 deg wide and 19 deg high at the 62cm viewing distance. Each kinematogram was composed of individual "dots" having a luminance of 70 cd/m 2 and subtending 4.1 by 5.3 min arc. The dots were randomly distributed with a density of 10% on a 0.013 cd/m 2 background. The average screen luminance was 6 cd/m 2. The direction of dot motion defined the display: a proportion of the dots drifted in one direction (left or right) while the remaining dots drifted in random directions. Adults perceived these displays as surfaces drifting to the right or to the left. The stimulus strength was varied by changing the dot velocity (Experiment 1) or by changing the correlation (proportion of dots that drifted in a single direction--Experiment 2). Kinematograms with two-frame dot lifetimes were used to insure that individual elements were not tracked and motion sensitive mechanisms were being studied. The one exception involved displays with 100% correlation, because individual dot lifetimes were necessarily unlimited. A discussion relevant to the spectral properties of random dot displays is found in Britten et al. (1993). Apparatus and procedure Stimuli were presented on a CRT (Zeos CMS-1461/ LR) under the control of a Zeos 486 computer. The CRT was located at the back of a matte-black viewing chamber designed to minimize visual distractions (Fig. 1). Diffuse light was shown from below the CRT to produce enough light to monitor eye movements with a video camera (Panasonic WV-BD400) mounted between the light source and the CRT. Eye movements were displayed on a video monitor (Panasonic WV-5400) located above the viewing chamber. During testing, the infant was seated in the looking chamber, approx. 62 cm from the stimulus CRT and the room lights were turned off. The roof of the chamber shielded the adult holding the infant from the stimuli. Motion detection thresholds were measured using a direction discrimination task. On each trial, a kinematogram containing continuous leftward or rightward motion was presented. When the infant attended to the kinematogram, optokinetic nystagmus (OKN) was induced, in which there was a slow eye movement corresponding to the direction of dot motion followed by a rapid refixation in the opposite direction. Eye movements were captured by the video camera and presented in real-time on the video monitor. A trained observer, who was shielded from the stimulus, viewed the video monitor and made a forcedchoice response as to whether stimulus motion was to the left or right. The observer was given an unlimited viewing duration, but judgments were typically made
INFANTS' SENSITIVITYTO UNIFORM MOTION
1635
Infant Dr )r
~pening
FIGURE 1. A schematic illustration of the infant testing chamber.
after the infant looked for 5-10 sec. Once the observer responded, the correct direction of stimulus movement was provided as feedback. The goal was to present 20 trials per condition. The data from infants completing at least 10 trials in any condition were included in the final analysis. Both the dot velocity and the percent correlation were varied in separate experiments to reach independent motion detection thresholds (defined as 75% correct discrimination). Adult coherence thresholds were obtained by the same procedures as with the infants. As a control, adult coherence and minimum velocity thresholds were also obtained by having subjects report the perceived direction of motion, eliminating variability attributable to the trained observer. Adult performance improved by an average of only 3.8% without an observer, indicating that observer sensitivity was reasonably good.
E X P E R I M E N T 1: SENSITIVITY TO SPEED
In Experiment 1, the sensitivity of 6-, 12-, and 18week-old infants to the speed of uniform motion was measured. The dots moved at one of three speeds (1.2, 3.0, or 7.4 deg/sec), created by displacing each dot 12.3 rain arc per frame at frame durations of 167, 69, and 28 msec, respectively. The experiment was conducted at dot correlations of 50 and 100%. The design and the number of infants per condition is presented in Table 1. Each infant was randomly assigned to the 50 or the 100% correlation condition, and was tested at all three speeds. Speed was randomly assigned on each trial. A total of 46 infants completed 10 trials at all three speeds. Some infants completed 10 trials in only one (n = 10) or two (n = 15) of the three speed conditions, and were included in the appropriate cells. Minimum speed thresholds at 50 and 100% correlation for each age group were determined. Figure 2(A) shows infants' performance on the TABLE 1. Design of Experiment 1 minimum speed task (100% correlation condition). Note that performance did not differ as a function of age Target speed (F[2,92] = 0.50, NS). This was also true when the 1.2 deg/sec 3.0 deg/sec 7.4 deg/sec velocity conditions were analyzed individually (F].2 d/s [2,31] = 0.31, NS; F3.o,1/s[2,38] = 0.45, NS; F7.4d/s (a) 100% correlation [2,23] = 0.47, NS). These data are replotted in Fig. 2(B) 6 weeks n = 11 n = 13 n=9 12 w e e k s n - 15 n = 19 n = 9 to show that performance improved at each age as dot 18 weeks n = 8 n = 10 n = 10 speed increased from 1.2 to 7.4 deg/sec (F[2,92] = 28.97, Adult n = 6 n = 6 P < 0.001). Minimum speed thresholds were obtained (b) 50% correlation from these functions by linear interpolation to the 75% 6weeks n = 8 n = 8 n = 8 correct level. Thresholds were 2.7 deg/sec at 6 weeks, 12 weeks n = 8 n = 12 n = 8 3.7 deg/sec at 12 weeks, and 3.3 deg/sec at 18 weeks of 18 weeks n = 8 n = 8 n = 8 age, and are marked by arrows in Fig. 2(B). Adult n = 6 n = 6 The results were fairly similar with targets displayed at Infants received three speeds of dot motion that were (a) 100% 50% correlation and a two-frame dot lifetime. Figure correlated, or (b) 50% correlated. Adults were tested on two of 3(A) shows that sensitivity to motion did not change as a these dot speeds. Each cell contains the number of subjects tested in each condition. function of age (F[2,67] = 2.02, NS). Further analysis
1636
T. BANTON and B. I. BERTENTHAL A
A
1 lOO% correlat ion ~
1 7 50%
0.8
ILl W W 00 Z
J correlat ion
7.4 deg/sec
___---+
3.0 deg/sec 1.2 deg/sec
0.6-
o
,,,b
0 w
o_
~
10 ,
,
-
,
,
,
15 r
.
.
.
.
-- . . . .
0.2
1.2 deg/sec
2'0
i 5
10
(weeks)
AGE
*3.0deo,see
04
on 0 n
0
7.4 deg/sec
I
,
E
0.2
±
0° 0.6-] Q
0.4-
15
20
AGE (weeks)
B
B
1 50%
1- lOO%
c
bILl
0.8
O O Z O
0.6
o
r
r
e
l
rr W 0 0
•
•. . . . . . . .
correlation
~
rr n-
j_~
0.6
Z _o
/ 0.4
0.4 O e_ O Cc ct
0.0 t ~
2
I • adult l -E:b-- 18 wk
0.2
-~ a-~u,t i
m
--O - 12 wk 6 wk
--~ -
0
0
;~
4
-
o~
i
1
6
8
0
10
VELOCITY (deg/sec)
FIGURE 2. Sensitivity to dot velocity at a correlation of 101)%. (A) Direction discrimination plotted as a function of age. Solid lines are the best linear fits for each speed of motion tested. In all conditions, infants' performance does not improve significantly with age. (B) Direction discrimination replotted as a function of dot velocity. Minimum speed thresholds interpolatedfrom these data are markedby arrows. Each datum is the average of at least eight infants or six adults.
showed no change with age across individual velocity conditions (F1.2 d/s[2,21] = 0.99, NS; F3.o ,Vs[2,25] = 0.15, NS; F7.4 d/s[2,21] = 1.39, NS). The replotted data in Fig. 3(B) show that at each age performance improved as dot speed increased (F[2,67] = 20.44, P < 0.001), and minimum speed thresholds were 4.8 deg/sec at 6 w e e k s , 4.1 deg/sec at 12 weeks, and 6.9 deg/sec at 18 weeks of age. Although performance appears to be poorer at 50% than at 100% correlation, these thresholds show again that sensitivity to the speed of uniform motion remains constant between 6 and 18 weeks of age. This plateau in uniform motion sensitivity is consistent with the early development of uniform motion sensitivity found in infant monkeys (Mikami & Fujita, 1992) and humans (Hamer & Norcia, 1994). The plateau in the infants' data does not suggest that adult levels of performance were reached. Adults performed at the 100% level on the two speeds at which they were tested [see Figs 2(B) and 3(B)]. It was not practical to measure adult minimum speed thresholds with this method, because large increases in the frame duration to produce very slow speeds gave the display the
.... , ~ ........ 2
i
'
~-,-
4 VELOCITY
,
_~
---A-- 6 wk
. . . . ,
6
8
7 10
(deg/sec)
3. Sensitivity to dot velocity at a c o r r e l a t i o n o f 5()9~, Direction discrimination plotted a s a function of age. Solid lines are the best linear fits for each speed of motion tested. In all conditions, infants" performance does not improve significantly with a g e . ( B ) Direction discrimination replotted as a function of dot velocity, Minimum speed thresholds interpolated from these data are marked by arrows. Each datum is the average of at least eight infants or six adults. FIGURE
(A)
appearance of unsmooth motion. A comparison of infant and adult minimum speed thresholds was therefore not made, but it is clear that infants' sensitivity to uniform motion is less acute than adult sensitivity. This is likely to reflect retinal and cortical limitations (Wilson, 1988), as well as attentional constraints. The timecourse of the development of motion sensitivity may be rather long, as adult speed thresholds for recognizing motion-defined targets are not reached until 7 - 9 yr of age (Giaschi & Regan, 1995). EXPERIMENT 2: SENSITIVITY TO COHERENCE To provide a second measure of sensitivity to uniform motion, coherence thresholds were determined for adults and 6-, 12-, and 18-week-old infants. The percent correlation of uniform motion necessary for discriminating left vs right motion, i.e., the minimum correlation that will drive the OKN response, was measured. Four correlation levels (12, 25, 50, 100%) were investigated. The dot speed was fixed at 7.4deg/sec, closely approximating the dot speed used by Wattam-Bell (1994) to measure coherence thresholds in 12-week-old
INFANTS' SENSITIVITYTO UNIFORM MOTION TABLE 2. Design of Experiment 2
A 1
Target correlation 3%
6%
12%
25%
50%
100%
7.4 deg/sec 6weeks 12 weeks 18 weeks Adult
n=6
n=6
n=9 n=8 n=9 n=6
n=8 n=8 n=8 n=6
1637
n=8 n=8 n=8 n=6
n=9 n=9 n=10 n=6
Subjects received different dot correlations at a dot speed of 7.4 deg/sec. Each cell contains the number of subjects tested in each condition.
7.4 deg/sec
0.8 n.. nO
, 50% 25%
I
,I,
0
4~
0.6-
4~ 12%
Z 0
0.4O n~
0.2-
0
. . . .
b
5
. . . . AGE
infants. Table 2 shows the design and number of infants used to obtain these motion detection thresholds. Since data for the 50 and 100% conditions were collected in Experiment 1, only the 12 and 25% correlation conditions were measured in Experiment 2. Each infant was tested in only one of these two conditions, because task difficulty and testing duration increased at lower correlations. Figure 4(A) shows infants' direction discrimination at four levels of directional coherence. As in Experiment 1, there was no systematic improvement in uniform motion sensitivity as a function of age (F[2,65] = 2.75, NS)*. Further analysis showed the lack of a developmental effect at each coherence level (F12%[2,23] = 3.04, NS; F25%[2,21] = 0.38, NS; F5o%[2,21] = 1.39, NS; Ftoo% [2,23] = 0.47, NS). Figure 4(B) replots the data to show that performance improved as the percent correlation increased (F[2,65] = 12.67, P < 0.001). Coherence thresholds (the percentage of uniformly moving dots required to correctly discriminate the direction of motion) were calculated from these functions by linear interpolation to the 75% correct level. Coherence thresholds were 36% at 6 weeks, 29% at 12 weeks, 37% at 18 weeks of age, and 11% by adulthood, and are marked by arrows in Fig. 4(B). The adult coherence threshold is higher than previously reported (5-7%; Wattam-Bell, 1994). It is important to remember that the adults were unpracticed, like the infants. The one adult observer with previous psychophysical experience (TB) had a coherence threshold of 7.7%. Interestingly, infant thresholds were lower than those reported by Wattam-Bell (1994). Lower thresholds may be due to either the large integration region in the uniform motion display resulting from the increased display size, threshold variability resulting from the shallow psychometric functions in Fig. 4(B), or
• Note that all omnibus analyses of these data do not include the 100% correlation condition, because the data in this condition were drawn from the same sample as the data in the 50% condition. Thus, inclusion of these data would have violated the between-subjects design used in the analyses of variance. The results from the 100% condition are nevertheless informative because they represent an upper limit on performance, and are therefore included whenever possible.
i 15
1
. . . .
i 20
(weeks)
B 1 7 7.4 deg/sec
~
~
o
•
LU 112 n"
8 Z O
n
oil O.
g
-~
-~ ~
°0w .
0
20
40
2 wk
• adult P-, 18 wk
.
.
6'0
.
i
80
100
% CORRELATION
FIGURE 4. Sensitivity to dot coherence at a speed of 7.4 deg/sec. (A) Direction discrimination plotted as a function of age. Solid lines are the best linear fits for each coherencelevel tested. In all conditions, infants' performance does not improve significantly with age. (B) Directiondiscrimination replotted as a function of dot coherence. Minimum coherence thresholds interpolated from these data are marked by arrows. Each datum is the averageof at least eight infants or six adults, unless otherwise noted.
a true processing difference between uniform and relative motion.
GENERAL DISCUSSION The present research shows that infants' sensitivity to uniform motion undergoes little change between 6 and 18 weeks of age. Since developmental functions generally show a period of rapid improvement that tapers into a relatively fiat slope as maximum performance levels are reached, the present data may reflect a developmental process that is near completion. A second possibility is that the fiat slope is reflecting a period of slow development prior to a stage of further improvement. In either case, the flat slope is a surprising result because it is in striking contrast to studies of acuity and contrast sensitivity (Banks & Salapatek, 1978), vernier localization (Manny & Klein, 1985; Shimojo et al., 1984), temporal resolution (Apkarian, 1993), binocularity (Shimojo et al., 1986), and other visual functions, all
1638
T. BANTON and B. I. BERTENTHAL
1 . 2 -
1
. -
_..-
U.I
rr" 0
0.8
5
°°l,-
Frr
0 rr
•
adult Macaque MT neurons (Britten et al, 1993) >,: adult human •--,~-.-18 week human (shifted) --4>- 12 week human (shifted) -~6 week human (shifted)
0.4 0.2
,
0
,
2~0
,
j
40
,
,
6'0
,
8'0
10 0
% CORRELATION
FIGURE 5. Comparison of uniform motion sensitivity in monkey and man with the coherence paradigm. Direction selectivity of six macaque MT neurons replotted from Britten et al. (1993), and direction discrimination by infants and adults in Experiment 2. The infant data have been shifted upward by a factor of 0.22. Power functions are fit to each data set. Note the similarity in the fits of monkey cortical motion mechanisms and the present data.
of which show improvement between 6 and 18 weeks of age. It is especially noteworthy that the developmental trajectory for uniform motion sensitivity differs from that of relative motion sensitivity, because it suggests that motion sensitivity is not a singular process: there may be separate mechanisms for sensing uniform and relative motion. However, important methodological differences between studies still remain and must be considered before drawing this conclusion. For example, Atkinson and Braddick (1981) argued that OKN is derived from subcortical structures until 2-3 months of age. Therefore, one might suspect that the OKN response measure invoked a subcortical motion mechanism in the present study of uniform motion. This contrasts with the cortical motion mechanisms presumably probed by VEP and preferential looking (PL) procedures (see Wattam-Bell, 1991) in studies investigating relative motion. Thus, the different developmental timecourses between this and previous studies may be unrelated to the development of uniform and relative motion, reflecting instead the emergence of subcortical mechanisms prior to cortical mechanisms. Although this hypothesis is plausible, we reject it for two reasons. First, the present response pattern using an OKN measure is similar to the cortical response to uniform motion obtained by Hamer and Norcia (1994) using VEPs. Second, there are striking similarities between the present coherence data and single unit recordings from macaque cortical area MT for the same task (Britten et al., 1993). Figure 5 highlights the similar shapes of the infant, adult, and neuronal functions. All the infant data were shifted upward by a constant (0.22), showing that only sensitivity changes by adulthood. Mechanism shifts, such as the transition from rod to cone activity, are classically accompanied by changes in function shape. The present data do not indicate that a shift (presumably from subcortical to
cortical processing) has occurred. In sum, it seems unlikely that our stimuli induced a motion response reflecting primarily subcortical activity. The different methodologies also impose different requirements on infants' attention, which might explain the current findings. The present uniform motion study used an eye movement measure requiring sufficient time to observe a slow following eye movement. This attentional constraint was not present in previous relative motion studies because a forced-choice PL procedure was typically used. The importance of a constraint on visual attention is that it could conceivably mask a change in the detectability of motion with age. If infants' attention spans decrease with age, their eye movements might be interrupted by attentional shifts, and the observer would have less information from which to extract the direction of eye movement. Thus, infants' performance might appear unchanged with age, but would actually reflect suboptimal performance by the older infants due to their decreased attention span. We raise this possibility because we noticed informally that 18-week-old infants often viewed the stimulus with several short looks in succession, while 6-week-old infants were usually content to stare at the screen for prolonged periods of time. Counter to this argument, our trained observer did not find it difficult to extract complete eye movements from 18-week-old infants, and videotape review of several sessions of 18-weekold infants confirmed that complete eye movements were available to the observer. In addition, the development of visual acuity as measured by OKN and PL are in good agreement (Dobson & Teller, 1978), suggesting that attentional constraints are of little consequence in OKN measures. Therefore, it seems that differences in attentional demand are unlikely to account for the different developmental functions obtained with uniform and relative motion stimuli. Perhaps a more likely reason for the dissociated development found in motion studies is that OKN may utilize sensory-motor pathways while PL may depend on the integrity of an object perception pathway. These systems are believed to be quite distinct from one another (Goodale & Milner, 1992), and may show differential development. To control for this possibility, a single measure of uniform and relative motion sensitivity should be used. In lieu of this evidence, it should be noted again that OKN and PL measures have previously produced equivalent results in visual acuity testing (Dobson & Teller, 1978). Although methodological issues might account for the differences between this and previous studies, the disparate results are also well described by proposing differential development of two mechanisms in infant vision; one sensitive to uniform and one sensitive to relative motion. A developmental dissociation of sensitivity to uniform and relative motion is consistent with the large body of evidence for separate uniform and relative motion detection mechanisms in adults. In addition, an early differentiation of uniform and relative
INFANTS' SENSITIVITY TO UNIFORM MOTION motion mechanisms is consistent with the importance of distinguishing self motion from object motion at all ages (Bridgeman, 1972; Frost et al., 1990). Whether or not this interpretation is correct remains to be tested directly, and current research in our lab is directed toward this goal. For now, we discuss several physiologically plausible frameworks which would support the differential development of uniform and relative motion sensitivity. One possible framework is a hierarchical scheme in which relative motion mechanisms are built from the outputs of uniform motion mechanisms. Models of other complex motions utilize this type of hierarchy to some extent. Rotation and expansion have been modeled by suggesting that complex motion detectors in MST may be created from the convergence of directionally selective cells with small receptive fields in MT (Tanaka et al., 1989). However, Tanaka et al. (1989) deviated from this framework by further proposing that full-field uniform motion detectors may be created from the convergence of M T cells which are locally sensitive to uniform motion. From a developmental perspective, a hierarchical framework predicts that relative motion mechanisms would follow the development of the simpler uniform motion mechanisms from which they are derived. This prediction is consistent with previous empirical data, which suggests that the rapid increase in sensitivity to uniform motion occurs prior to the improvement in sensitivity to relative motion. Thought of in this way, complex motion detectors may develop after uniform motion detectors. A second physiological framework supporting the dissociated development of uniform and relative motion sensitivity is one in which uniform and relative motion thresholds are constrained by the development of independent cortical spatial frequency mechanisms. Uniform motion may primarily stimulate low spatial frequency mechanisms, while relative motion might stimulate independent higher spatial frequency channels. Furthermore, low spatial frequency channels develop more rapidly than high spatial frequency channels (Atkinson et al., 1974; Banks & Salapatek, 1978), suggesting that a sensitivity to uniform motion may develop prior to a sensitivity to relative motion. In the present study, the plateau in sensitivity to uniform motion between 6 and 18 weeks could reflect early development of low spatial frequency mechanisms which are likely to govern uniform motion. Along these lines, the marked improvements in relative motion sensitivity noted between 10 and 20 weeks of age in previous studies could reflect the later development of high spatial frequency mechanisms. Interestingly, Manny and Fern (1990) found no change in direction discrimination with age when the stimulus spatial frequency was adjusted to that producing peak sensitivity at each age. Additional data regarding the development of size-tuned motion mechanisms are required to properly evaluate this hypothesis. In summary, infants as young as 6 weeks of age possess visual mechanisms sensitive to uniform motion. This finding is important, because it challenges the claim
1639
that direction sensitivity emerges after 8 weeks of age (e.g. Braddick, 1993). We suggest that the development of motion processing may follow multiple courses, depending on the type of motion under investigation.
REFERENCES Allman, J., Miezin, F. & McGuinness, E. (1985). Direction- and velocity-specificresponses from beyond the classical receptive field in the middle temporal visual area (MT). Perception, 14, 105-126. Apkarian, P. (1993). Temporal frequency responsivity shows multiple maturational phases: State-dependent visual evoked potential luminance flicker fusion from birth to 9 months. Visual Neuroscience, 10, 1007-1018. Aslin, R. N. & Shea, S. L. (1990). Velocity thresholds in human infants: Implications for the perception of motion. Developmental Psychology, 26, 589-598. Atkinson, J. & Braddick, O. J. (1981). Development of optokinetic nystagmus in infants: An indicator of cortical binocularity? In Fisher, D. F., Monty, R. A. and Senders, J. W. (Eds), Eye movements: Cognition and visual perception (pp. 53-64). Hiilsdale, NJ: Erlbaum. Atkinson, J., Braddick, O. & Braddick, F. (1974). Acuity and contrast sensitivity of infant vision. Nature, London, 247, 403-404. Banks, M. S. & Salapatek, P. (1978). Acuity and contrast sensitivity in 1-, 2- and 3-month-old human infants. Investigative Ophthalmology and Visual Science, 17, 361-365. Banton, T. & Bertenthal, B. I. (1995). The effect of target size on infant detection of relative motion. Investigative Ophthalmology and Visual Science, 36, $909. Bertenthat, B. I. & Bradbury, A. (1992). Infants' detection of shearing motion in random-dot displays. Developmental Psychology, 28, 1056-1066. Boothe, R. G., Dobson, V. & Teller, D. Y. (1985). Postnatal development of vision in human and nonhuman primates. Annual Review of Neuroscience, 8, 495-545. Braddick, O. (1993). Orientation- and motion-selectivemechanismsin infants. In Simons, K. (Ed.), Early visual development; normal and abnormal (pp. 163-177). New York: Oxford University Press. Bridgeman, B. (1972). Visual receptive fields sensitive to absolute and relative motion during tracking. Science, 178, 1106-1108. Britten, K. H., Shadlen, M. N., Newsome, W. T. & Movshon, J. A. (1993). Responses of neurons in macaque MT to stochastic motion signals. Visual Neuroscience, 10, 1157-1169. Dannemiller, J. L. & Freedland, R. (1989). The detection of slow stimulus movement in 2- to 5-month-olds.Journal of Experimental Child Psychology, 47, 337-355. Dannemiller, J. L. & Freedland, R. (1991). Detection of relative motion by human infants. Developmental Psychology, 27, 67-78. Davidson, R. M. & Bender, D. B. (1991). Selectivity for relative motion in the monkey superior colliculus. Journal of Neurophysiology, 65, 1115-1133. Dobson, V. & Teller, D. Y. (1978). Visual acuity in human infants: A review and comparison of behavioral and electrophysiological studies. Vision Research, 18, 1469-1483. Freedland, R. L. & Dannemiller, J. L. (1987). Detection of stimulus motion in 5-month-old infants.Journal of Experimental Psychology: Human Perception and Performance, 13, 566-576. Frost, B. J. (1978). Moving background patterns alter directionally specific responses of pigeon tectal neurons. Brain Research, 151, 599-603. Frost, B. J., Wylie, D. R. & Wang, Y. C. (1990). The processing of object and self-motion in the tectofugal and accessory optic pathways of birds. Special Issue: Optics, physiology and vision. Vision Research, 30, 1677-1688. Giaschi, D. E. & Regan, D. (1995). Dissociated visual development in the processing of motion-defined and luminance-defined form. Investigative Ophthalmology and Visual Science, 36, $443. Goodale, M. A. & Milner, A. D. (1992). Separate visual pathways for perception and action. Trends in Neurosciences, 15, 20--25.
1640
T. BANTON and B. I. BERTENTHAL
von Grunau, M. W. & Frost, B. J. (1983). Double opponent-process mechanism underlying receptive field structure of directionally specific cells of cat lateral suprasylvian visual area. Experimental Brain Research, 49, 84-92. Gulyas, B., Orban, G. A., Duysens, J. & Maes, H. (1987). The suppressive influence of moving textured backgrounds on responses of cat striate neurons to moving bars. Journal of Neurophysiology, 57, 1767-1791. Hamer, R. D. & Norcia, A. M. (1994). The development of motion sensitivity during the first year of life. Vision Research, 34, 23872402. Hammond, P., Ahmed, B. & Smith, A. T. (1986). Relative motion sensitivity in cat striate cortex as a function of stimulus direction. Brain Research, 386, 93-104. Hammond, P. & Smith, A. T. (1982). On the sensitivity of complex cells in feline striate cortex to relative motion. Experimental Brain Research, 47, 457~160. Hammond, P. & Smith, A. T. (1984). Sensitivity of complex cells in cat striate cortex to relative motion. Brain Research, 301,287-298. Jouen, F., Lepecq, J. C. & Gapenne, O. (1996). Optical flow sensitivity in neonates. Manuscript submitted for publication. Kaufmann, F., Stucki, M. & Kaufmann-Hayoz, H. R. (1985). Development of infants' sensitivity for slow and rapid motions. Infant Behavior and Development, 8, 89-98. Manny, R. E. & Fern, K. D. (1990). Motion coherence in infants. Vision Research, 30, 1319-1329. Manny, R. E. & Klein, S. A. (1985). A three alternative tracking paradigm to measure vernier acuity of older infants. Vision Research, 25, 1245-1252. Mikami, A. & Fujita, K. (1992). Development of the ability to detect visual motion in infant macaque monkeys. Developmental Psychobiology, 25, 345-354. Nakayama, K. & Tyler, C. W. (1981). Psychophysical isolation of movement sensitivity by removal of familiar position cues. Vision
Orban, G. A., Gulyas, B. & Vogels, R. (1987b). Influence of a moving textured background on direction selectivity of cat striate neurons. Journal of Neurophysiology, 57, 1792-1812. Regan, D., Giaschi, D., Sharpe, J. A. & Hong, X. H. (1992). Visual processing of motion-defined form: Selective failure in patients with parietotemporal lesions. The Journal of Neuroscience, 12, 21982210. Shimojo, S., Bauer, J., O'Connell, K. M. & Held, R. (1986). Prestereoptic binocular vision in infants. Vision Research, 26, 501-510. Shimojo, S., Birch, E. E., Gwiazda, J. & Held, R. (1984). Development of vernier acuity in infants. Vision Research, 24, 721-728. Smeets, J. B. J. & Brenner, E. (1994). The difference between the perception of absolute and relative motion: A reaction time study. Vision Research, 34, 191-195. Snowden, R. J. (1992). Sensitivity to relative and absolute motion.
Perception, 21,563-568. Sterling, P. & Wickelgren, B. (1969). Visual receptive fields in the superior colliculus of the cat. Journal ofNeurophysiology, 32, 1-15. Tanaka, K., Fukada, Y. & Saito, H. (1989). Underlying mechanisms of the response specificity of expansion/contraction and rotation cells in the dorsal part of the medial superior temporal area of the macaque monkey. Journal of Neurophysiology, 62, 642-656. Volkmann, F. C. & Dobson, M. V. (1976). Infant responses of ocular fixation to moving visual stimuli. Journal of Experimental Child Psychology, 22, 86-99. Wattam-Bell, J. (199l). Development of motion-specific cortical responses in infancy. Vision Research, 31,287-297. Wattam-Bell, J. (1992). The development of maximum displacement limits for discrimination of motion direction in infancy. Vision Research, 32, 621-630. Wattam-Bell, J. (1994). Coherence thresholds for discrimination of motion direction in infants. Vision Research, 34, 877-883. Wilson, H. R. (1988). Development of spatiotemporal mechanisms in infant vision. Vision Research, 28, 611-628.
Research, 21,427--433. Orban, G. A., Gulyas, B. & Spileers, W. (1987a). A moving noise background modulates responses to moving bars of monkey V2 cells but not of monkey V1 cells. Investigative Ophthalmology and Visual Science Supplement, 28, 197. Orban, G. A., Gulyas, B. & Spileers, W. (1988). Influence of moving textured backgrounds on responses of cat area 18 cells to moving bars. In Hicks, T. P. & Benedek, G. (Eds), Progress in brain research (pp. 137-145). Amsterdam: Elsevier Science Publishers B.V.
Acknowledgements--We thank our trained observers, Tara Speakman and Heather Tevendale, for their expert assistance in the data collection. A preliminary account of this research was presented at the annual meeting of the Association for Research in Vision and Ophthalmology, Sarasota, Florida, 1993. This work is supported by NIH Grant HD16195 to Bennett I. Bertenthal.
0042-6989(95) 00216-2
Vision Res., Vol. 36, No. 11, pp. 1633-1640, 1996 Copyright © 1996 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0042-6989/96 $15.00 + 0.00
Infants' Sensitivity to Uniform Motion TOM BANTON,*t BENNETT I. BERTENTHAL* Received 20 January 1995; in revised form 26 July 1995
Uniform motion across the retina is a powerful cue to the perception of self-motion. In spite of its importance for adaptive functioning, little is known about the early development of uniform motion sensitivity. Six-, 12-, and 18-week-old infants viewed random-dot kinematograms depicting leftward or rightward uniform motion. The display induced optokinetic nystagmus (OKN), which a trained observer used to judge the direction of target motion. Both speed of motion and directional coherence were varied to obtain independent motion detection thresholds. Infants of all three ages could detect uniform motion, and their detection thresholds were constant during this period of development. This is in contrast to the clear improvements in relative motion sensitivity noted previously between 6 and 18 weeks of age with a preferential looking (PL) paradigm. The developmental differences between these studies may result from: (1) separate mechanisms for detecting uniform (absolute) and differential (relative) motion; or (2) separate mechanisms underlying OKN and PL response measures. Copyright © 1996 Elsevier Science Ltd.
Development
Infant vision Uniform motion
Relativemotion
INTRODUCTION Recent evidence from neurophysiological, neuropsychological, and psychophysical studies converge to suggest the existence of two visual motion mechanisms; one sensing uniform motion of the retinal image (as occurs with eye movement or when the entire visual field is moved) and one sensing motion discontinuities in the retinal image (as found with motion parallax, for example). Bridgeman (1972) referred to these as "absolute" and "relative" motion mechanisms respectively, and demonstrated their existence neurophysiologically in area V1 of the rhesus monkey. This dissociation is also supported through differential responses to uniform and relative motion in the pigeon tectum (Frost, 1978; Frost et al., 1990), the cat superior colliculus (Sterling & Wickelgren, 1969), lateral geniculate nucleus (Gulyas et al., 1987), lateral suprasylvian gyrus (von Grunau & Frost, 1983), area 17 (Orban et al., 1987b; Gulyas et al., 1987; Hammond & Smith, 1982, 1984; Hammond et al., 1986), area 18 (Orban et al., 1988), monkey superior colliculus (Davidson & Bender, 1991), and extrastriate cortex (Allman et al., 1985; Orban et al., 1987a). Regan et al. (1992) confirmed the separability of uniform and relative motion perception in persons with tempero-parietal lesions. These patients showed an inability to detect and discriminate motiondefined form, but had normal sensitivity to uniform motion. Psychophysical measures in visually normal
observers also point to a distinction between uniform and relative motion sensitivity (Smeets & Brenner, 1994; Snowden, 1992). It is believed that one function of the dissociation is to enable the visual system to distinguish eye movement from object movement (Bridgeman, 1972; Frost et al., 1990). If uniform and relative motion mechanisms are indeed distinct, then they will develop independently, perhaps following different timecourses. Many of the initial developmental investigations of motion sensitivity demonstrated that infants preferred moving to stationary targets (Volkmann & Dobson, 1976; Kaufmann et al., 1985; Freedland & Dannemiller, 1987). Furthermore, these studies suggested that motion detection improved with age. Although some of these stimuli left open the possibility that flicker-sensitive or positionsensitive mechanisms, rather than motion-sensitive mechanisms, were involved (Aslin & Shea, 1990; Braddick, 1993; Nakayama & Tyler, 1981), studies by Aslin and Shea (1990) and Dannemiller and Freedland (1989, 1991) confirmed the earlier findings and showed that flicker- and position-sensitive mechanisms were unlikely to account for infants' motion preferences. Thus, improvements with age were attributed to the development of motion mechanisms. Recently, the development of motion sensitivity has been scrutinized more closely by looking at particular types of motion. In a study of relative motion, Bertenthal and Bradbury (1992) measured infants' sensitivity to the shearing motion of a random-dot sinusoid of 0.16 c/deg. Motion detection improved from 3.5 deg/sec at 13 weeks to 1.2 deg/sec at 20 weeks of age. Wattam-Bell (1992) reported significant improvement in motion-defined square wave detection between 11 and 16 weeks of age. Banton and Bertenthal (1995) found that size thresholds
*Room 171, Gilmer Hall, Department of Psychology, University of Virginia, Charlottesville VA 22903, U.S.A. tTo whom all correspondence should be addressed [Email [email protected]]. 1633
1634
T. BANTON and B. I. BERTENTHAL
for detecting a motion-defined rectangle could not be measured at 8 weeks of age, but showed a significant improvement between 12 and 16 weeks of age. Finally, Giaschi and Regan (1985) showed that mechanisms for motion-defined letter recognition (a high spatial frequency relative motion task) did not reach adult levels until 7-9 yr of age. Taken together, these studies show that relative motion sensitivity improves steadily from at least 11 weeks of age through 7-9 yr of age. It is unclear if this same pattern of development is evident in the few studies investigating infant sensitivity to uniform motion. Hamer and Norcia (1994) used visual evoked potentials (VEPs) to measure infants' thresholds to oscillatory uniform motion of a grating. Motion thresholds were unchanged between 7 weeks and 1 yr of age, suggesting a different developmental pattern than previously found with preferential looking to relative motion. Mikami and Fujita (1992) used preferential looking to investigate the development of uniform motion detection in rhesus and Japanese monkeys. In this case, linearly plotted thresholds for detecting a moving grating improved rapidly between 1 day and 2 weeks of age, after which thresholds began to plateau. After applying a four-to-one correction for life span differences between monkey and man (Boothe et al., 1985), the data predict that human uniform motion detection may improve between 4 days and 8 weeks of age, and might begin to plateau by 8 weeks of age. However, Wattam-Bell (1991) found an insensitivity to oscillatory uniform motion of a random-dot field in human infants prior to 8 weeks of age, consistent with the view that infants are poor at sensing motion prior to this age (Braddick, 1993). Recently, Jouen et al., (1996) found that newborns can make motor responses scaled to the velocity of peripherally presented gratings that move in depth. Although these results suggest that uniform motion sensitivity develops early and then plateaus, it is difficult to establish the age at which this phase occurs, given the differences in methods, species, and stimuli. The purpose of the present study was to conduct a behavioral test of the development of human infants' sensitivity to uniform motion between 6 and 18 weeks of age, a time period when the onset of the developmental plateau may occur. It was anticipated that the results of this study would clarify how human sensitivity to uniform motion changes with age, and would facilitate comparison between the development of uniform and relative motion sensitivity.
METHODS
Subjects A total of 177 infants at 6, 12, and 18 weeks of age (_+1 week) participated in the study. Infants were recruited from birth announcements in the local newspaper. All infants who completed at least 10 trials per condition were included in the final analysis (n = 121). Fifty-six of the 177 infants completed an inadequate number of trials due to sleeping, fussing, or inattentive-
ness, and were not included in the analysis. Six adults from the Laboratory for Infant Studies also participated. One adult observer (author TB) was aware of the purpose of the study.
Stimuli Random-dot kinematograms were used as stimuli. The kinematograms were 22 deg wide and 19 deg high at the 62cm viewing distance. Each kinematogram was composed of individual "dots" having a luminance of 70 cd/m 2 and subtending 4.1 by 5.3 min arc. The dots were randomly distributed with a density of 10% on a 0.013 cd/m 2 background. The average screen luminance was 6 cd/m 2. The direction of dot motion defined the display: a proportion of the dots drifted in one direction (left or right) while the remaining dots drifted in random directions. Adults perceived these displays as surfaces drifting to the right or to the left. The stimulus strength was varied by changing the dot velocity (Experiment 1) or by changing the correlation (proportion of dots that drifted in a single direction--Experiment 2). Kinematograms with two-frame dot lifetimes were used to insure that individual elements were not tracked and motion sensitive mechanisms were being studied. The one exception involved displays with 100% correlation, because individual dot lifetimes were necessarily unlimited. A discussion relevant to the spectral properties of random dot displays is found in Britten et al. (1993). Apparatus and procedure Stimuli were presented on a CRT (Zeos CMS-1461/ LR) under the control of a Zeos 486 computer. The CRT was located at the back of a matte-black viewing chamber designed to minimize visual distractions (Fig. 1). Diffuse light was shown from below the CRT to produce enough light to monitor eye movements with a video camera (Panasonic WV-BD400) mounted between the light source and the CRT. Eye movements were displayed on a video monitor (Panasonic WV-5400) located above the viewing chamber. During testing, the infant was seated in the looking chamber, approx. 62 cm from the stimulus CRT and the room lights were turned off. The roof of the chamber shielded the adult holding the infant from the stimuli. Motion detection thresholds were measured using a direction discrimination task. On each trial, a kinematogram containing continuous leftward or rightward motion was presented. When the infant attended to the kinematogram, optokinetic nystagmus (OKN) was induced, in which there was a slow eye movement corresponding to the direction of dot motion followed by a rapid refixation in the opposite direction. Eye movements were captured by the video camera and presented in real-time on the video monitor. A trained observer, who was shielded from the stimulus, viewed the video monitor and made a forcedchoice response as to whether stimulus motion was to the left or right. The observer was given an unlimited viewing duration, but judgments were typically made
INFANTS' SENSITIVITYTO UNIFORM MOTION
1635
Infant Dr )r
~pening
FIGURE 1. A schematic illustration of the infant testing chamber.
after the infant looked for 5-10 sec. Once the observer responded, the correct direction of stimulus movement was provided as feedback. The goal was to present 20 trials per condition. The data from infants completing at least 10 trials in any condition were included in the final analysis. Both the dot velocity and the percent correlation were varied in separate experiments to reach independent motion detection thresholds (defined as 75% correct discrimination). Adult coherence thresholds were obtained by the same procedures as with the infants. As a control, adult coherence and minimum velocity thresholds were also obtained by having subjects report the perceived direction of motion, eliminating variability attributable to the trained observer. Adult performance improved by an average of only 3.8% without an observer, indicating that observer sensitivity was reasonably good.
E X P E R I M E N T 1: SENSITIVITY TO SPEED
In Experiment 1, the sensitivity of 6-, 12-, and 18week-old infants to the speed of uniform motion was measured. The dots moved at one of three speeds (1.2, 3.0, or 7.4 deg/sec), created by displacing each dot 12.3 rain arc per frame at frame durations of 167, 69, and 28 msec, respectively. The experiment was conducted at dot correlations of 50 and 100%. The design and the number of infants per condition is presented in Table 1. Each infant was randomly assigned to the 50 or the 100% correlation condition, and was tested at all three speeds. Speed was randomly assigned on each trial. A total of 46 infants completed 10 trials at all three speeds. Some infants completed 10 trials in only one (n = 10) or two (n = 15) of the three speed conditions, and were included in the appropriate cells. Minimum speed thresholds at 50 and 100% correlation for each age group were determined. Figure 2(A) shows infants' performance on the TABLE 1. Design of Experiment 1 minimum speed task (100% correlation condition). Note that performance did not differ as a function of age Target speed (F[2,92] = 0.50, NS). This was also true when the 1.2 deg/sec 3.0 deg/sec 7.4 deg/sec velocity conditions were analyzed individually (F].2 d/s [2,31] = 0.31, NS; F3.o,1/s[2,38] = 0.45, NS; F7.4d/s (a) 100% correlation [2,23] = 0.47, NS). These data are replotted in Fig. 2(B) 6 weeks n = 11 n = 13 n=9 12 w e e k s n - 15 n = 19 n = 9 to show that performance improved at each age as dot 18 weeks n = 8 n = 10 n = 10 speed increased from 1.2 to 7.4 deg/sec (F[2,92] = 28.97, Adult n = 6 n = 6 P < 0.001). Minimum speed thresholds were obtained (b) 50% correlation from these functions by linear interpolation to the 75% 6weeks n = 8 n = 8 n = 8 correct level. Thresholds were 2.7 deg/sec at 6 weeks, 12 weeks n = 8 n = 12 n = 8 3.7 deg/sec at 12 weeks, and 3.3 deg/sec at 18 weeks of 18 weeks n = 8 n = 8 n = 8 age, and are marked by arrows in Fig. 2(B). Adult n = 6 n = 6 The results were fairly similar with targets displayed at Infants received three speeds of dot motion that were (a) 100% 50% correlation and a two-frame dot lifetime. Figure correlated, or (b) 50% correlated. Adults were tested on two of 3(A) shows that sensitivity to motion did not change as a these dot speeds. Each cell contains the number of subjects tested in each condition. function of age (F[2,67] = 2.02, NS). Further analysis
1636
T. BANTON and B. I. BERTENTHAL A
A
1 lOO% correlat ion ~
1 7 50%
0.8
ILl W W 00 Z
J correlat ion
7.4 deg/sec
___---+
3.0 deg/sec 1.2 deg/sec
0.6-
o
,,,b
0 w
o_
~
10 ,
,
-
,
,
,
15 r
.
.
.
.
-- . . . .
0.2
1.2 deg/sec
2'0
i 5
10
(weeks)
AGE
*3.0deo,see
04
on 0 n
0
7.4 deg/sec
I
,
E
0.2
±
0° 0.6-] Q
0.4-
15
20
AGE (weeks)
B
B
1 50%
1- lOO%
c
bILl
0.8
O O Z O
0.6
o
r
r
e
l
rr W 0 0
•
•. . . . . . . .
correlation
~
rr n-
j_~
0.6
Z _o
/ 0.4
0.4 O e_ O Cc ct
0.0 t ~
2
I • adult l -E:b-- 18 wk
0.2
-~ a-~u,t i
m
--O - 12 wk 6 wk
--~ -
0
0
;~
4
-
o~
i
1
6
8
0
10
VELOCITY (deg/sec)
FIGURE 2. Sensitivity to dot velocity at a correlation of 101)%. (A) Direction discrimination plotted as a function of age. Solid lines are the best linear fits for each speed of motion tested. In all conditions, infants' performance does not improve significantly with age. (B) Direction discrimination replotted as a function of dot velocity. Minimum speed thresholds interpolatedfrom these data are markedby arrows. Each datum is the average of at least eight infants or six adults.
showed no change with age across individual velocity conditions (F1.2 d/s[2,21] = 0.99, NS; F3.o ,Vs[2,25] = 0.15, NS; F7.4 d/s[2,21] = 1.39, NS). The replotted data in Fig. 3(B) show that at each age performance improved as dot speed increased (F[2,67] = 20.44, P < 0.001), and minimum speed thresholds were 4.8 deg/sec at 6 w e e k s , 4.1 deg/sec at 12 weeks, and 6.9 deg/sec at 18 weeks of age. Although performance appears to be poorer at 50% than at 100% correlation, these thresholds show again that sensitivity to the speed of uniform motion remains constant between 6 and 18 weeks of age. This plateau in uniform motion sensitivity is consistent with the early development of uniform motion sensitivity found in infant monkeys (Mikami & Fujita, 1992) and humans (Hamer & Norcia, 1994). The plateau in the infants' data does not suggest that adult levels of performance were reached. Adults performed at the 100% level on the two speeds at which they were tested [see Figs 2(B) and 3(B)]. It was not practical to measure adult minimum speed thresholds with this method, because large increases in the frame duration to produce very slow speeds gave the display the
.... , ~ ........ 2
i
'
~-,-
4 VELOCITY
,
_~
---A-- 6 wk
. . . . ,
6
8
7 10
(deg/sec)
3. Sensitivity to dot velocity at a c o r r e l a t i o n o f 5()9~, Direction discrimination plotted a s a function of age. Solid lines are the best linear fits for each speed of motion tested. In all conditions, infants" performance does not improve significantly with a g e . ( B ) Direction discrimination replotted as a function of dot velocity, Minimum speed thresholds interpolated from these data are marked by arrows. Each datum is the average of at least eight infants or six adults. FIGURE
(A)
appearance of unsmooth motion. A comparison of infant and adult minimum speed thresholds was therefore not made, but it is clear that infants' sensitivity to uniform motion is less acute than adult sensitivity. This is likely to reflect retinal and cortical limitations (Wilson, 1988), as well as attentional constraints. The timecourse of the development of motion sensitivity may be rather long, as adult speed thresholds for recognizing motion-defined targets are not reached until 7 - 9 yr of age (Giaschi & Regan, 1995). EXPERIMENT 2: SENSITIVITY TO COHERENCE To provide a second measure of sensitivity to uniform motion, coherence thresholds were determined for adults and 6-, 12-, and 18-week-old infants. The percent correlation of uniform motion necessary for discriminating left vs right motion, i.e., the minimum correlation that will drive the OKN response, was measured. Four correlation levels (12, 25, 50, 100%) were investigated. The dot speed was fixed at 7.4deg/sec, closely approximating the dot speed used by Wattam-Bell (1994) to measure coherence thresholds in 12-week-old
INFANTS' SENSITIVITYTO UNIFORM MOTION TABLE 2. Design of Experiment 2
A 1
Target correlation 3%
6%
12%
25%
50%
100%
7.4 deg/sec 6weeks 12 weeks 18 weeks Adult
n=6
n=6
n=9 n=8 n=9 n=6
n=8 n=8 n=8 n=6
1637
n=8 n=8 n=8 n=6
n=9 n=9 n=10 n=6
Subjects received different dot correlations at a dot speed of 7.4 deg/sec. Each cell contains the number of subjects tested in each condition.
7.4 deg/sec
0.8 n.. nO
, 50% 25%
I
,I,
0
4~
0.6-
4~ 12%
Z 0
0.4O n~
0.2-
0
. . . .
b
5
. . . . AGE
infants. Table 2 shows the design and number of infants used to obtain these motion detection thresholds. Since data for the 50 and 100% conditions were collected in Experiment 1, only the 12 and 25% correlation conditions were measured in Experiment 2. Each infant was tested in only one of these two conditions, because task difficulty and testing duration increased at lower correlations. Figure 4(A) shows infants' direction discrimination at four levels of directional coherence. As in Experiment 1, there was no systematic improvement in uniform motion sensitivity as a function of age (F[2,65] = 2.75, NS)*. Further analysis showed the lack of a developmental effect at each coherence level (F12%[2,23] = 3.04, NS; F25%[2,21] = 0.38, NS; F5o%[2,21] = 1.39, NS; Ftoo% [2,23] = 0.47, NS). Figure 4(B) replots the data to show that performance improved as the percent correlation increased (F[2,65] = 12.67, P < 0.001). Coherence thresholds (the percentage of uniformly moving dots required to correctly discriminate the direction of motion) were calculated from these functions by linear interpolation to the 75% correct level. Coherence thresholds were 36% at 6 weeks, 29% at 12 weeks, 37% at 18 weeks of age, and 11% by adulthood, and are marked by arrows in Fig. 4(B). The adult coherence threshold is higher than previously reported (5-7%; Wattam-Bell, 1994). It is important to remember that the adults were unpracticed, like the infants. The one adult observer with previous psychophysical experience (TB) had a coherence threshold of 7.7%. Interestingly, infant thresholds were lower than those reported by Wattam-Bell (1994). Lower thresholds may be due to either the large integration region in the uniform motion display resulting from the increased display size, threshold variability resulting from the shallow psychometric functions in Fig. 4(B), or
• Note that all omnibus analyses of these data do not include the 100% correlation condition, because the data in this condition were drawn from the same sample as the data in the 50% condition. Thus, inclusion of these data would have violated the between-subjects design used in the analyses of variance. The results from the 100% condition are nevertheless informative because they represent an upper limit on performance, and are therefore included whenever possible.
i 15
1
. . . .
i 20
(weeks)
B 1 7 7.4 deg/sec
~
~
o
•
LU 112 n"
8 Z O
n
oil O.
g
-~
-~ ~
°0w .
0
20
40
2 wk
• adult P-, 18 wk
.
.
6'0
.
i
80
100
% CORRELATION
FIGURE 4. Sensitivity to dot coherence at a speed of 7.4 deg/sec. (A) Direction discrimination plotted as a function of age. Solid lines are the best linear fits for each coherencelevel tested. In all conditions, infants' performance does not improve significantly with age. (B) Directiondiscrimination replotted as a function of dot coherence. Minimum coherence thresholds interpolated from these data are marked by arrows. Each datum is the averageof at least eight infants or six adults, unless otherwise noted.
a true processing difference between uniform and relative motion.
GENERAL DISCUSSION The present research shows that infants' sensitivity to uniform motion undergoes little change between 6 and 18 weeks of age. Since developmental functions generally show a period of rapid improvement that tapers into a relatively fiat slope as maximum performance levels are reached, the present data may reflect a developmental process that is near completion. A second possibility is that the fiat slope is reflecting a period of slow development prior to a stage of further improvement. In either case, the flat slope is a surprising result because it is in striking contrast to studies of acuity and contrast sensitivity (Banks & Salapatek, 1978), vernier localization (Manny & Klein, 1985; Shimojo et al., 1984), temporal resolution (Apkarian, 1993), binocularity (Shimojo et al., 1986), and other visual functions, all
1638
T. BANTON and B. I. BERTENTHAL
1 . 2 -
1
. -
_..-
U.I
rr" 0
0.8
5
°°l,-
Frr
0 rr
•
adult Macaque MT neurons (Britten et al, 1993) >,: adult human •--,~-.-18 week human (shifted) --4>- 12 week human (shifted) -~6 week human (shifted)
0.4 0.2
,
0
,
2~0
,
j
40
,
,
6'0
,
8'0
10 0
% CORRELATION
FIGURE 5. Comparison of uniform motion sensitivity in monkey and man with the coherence paradigm. Direction selectivity of six macaque MT neurons replotted from Britten et al. (1993), and direction discrimination by infants and adults in Experiment 2. The infant data have been shifted upward by a factor of 0.22. Power functions are fit to each data set. Note the similarity in the fits of monkey cortical motion mechanisms and the present data.
of which show improvement between 6 and 18 weeks of age. It is especially noteworthy that the developmental trajectory for uniform motion sensitivity differs from that of relative motion sensitivity, because it suggests that motion sensitivity is not a singular process: there may be separate mechanisms for sensing uniform and relative motion. However, important methodological differences between studies still remain and must be considered before drawing this conclusion. For example, Atkinson and Braddick (1981) argued that OKN is derived from subcortical structures until 2-3 months of age. Therefore, one might suspect that the OKN response measure invoked a subcortical motion mechanism in the present study of uniform motion. This contrasts with the cortical motion mechanisms presumably probed by VEP and preferential looking (PL) procedures (see Wattam-Bell, 1991) in studies investigating relative motion. Thus, the different developmental timecourses between this and previous studies may be unrelated to the development of uniform and relative motion, reflecting instead the emergence of subcortical mechanisms prior to cortical mechanisms. Although this hypothesis is plausible, we reject it for two reasons. First, the present response pattern using an OKN measure is similar to the cortical response to uniform motion obtained by Hamer and Norcia (1994) using VEPs. Second, there are striking similarities between the present coherence data and single unit recordings from macaque cortical area MT for the same task (Britten et al., 1993). Figure 5 highlights the similar shapes of the infant, adult, and neuronal functions. All the infant data were shifted upward by a constant (0.22), showing that only sensitivity changes by adulthood. Mechanism shifts, such as the transition from rod to cone activity, are classically accompanied by changes in function shape. The present data do not indicate that a shift (presumably from subcortical to
cortical processing) has occurred. In sum, it seems unlikely that our stimuli induced a motion response reflecting primarily subcortical activity. The different methodologies also impose different requirements on infants' attention, which might explain the current findings. The present uniform motion study used an eye movement measure requiring sufficient time to observe a slow following eye movement. This attentional constraint was not present in previous relative motion studies because a forced-choice PL procedure was typically used. The importance of a constraint on visual attention is that it could conceivably mask a change in the detectability of motion with age. If infants' attention spans decrease with age, their eye movements might be interrupted by attentional shifts, and the observer would have less information from which to extract the direction of eye movement. Thus, infants' performance might appear unchanged with age, but would actually reflect suboptimal performance by the older infants due to their decreased attention span. We raise this possibility because we noticed informally that 18-week-old infants often viewed the stimulus with several short looks in succession, while 6-week-old infants were usually content to stare at the screen for prolonged periods of time. Counter to this argument, our trained observer did not find it difficult to extract complete eye movements from 18-week-old infants, and videotape review of several sessions of 18-weekold infants confirmed that complete eye movements were available to the observer. In addition, the development of visual acuity as measured by OKN and PL are in good agreement (Dobson & Teller, 1978), suggesting that attentional constraints are of little consequence in OKN measures. Therefore, it seems that differences in attentional demand are unlikely to account for the different developmental functions obtained with uniform and relative motion stimuli. Perhaps a more likely reason for the dissociated development found in motion studies is that OKN may utilize sensory-motor pathways while PL may depend on the integrity of an object perception pathway. These systems are believed to be quite distinct from one another (Goodale & Milner, 1992), and may show differential development. To control for this possibility, a single measure of uniform and relative motion sensitivity should be used. In lieu of this evidence, it should be noted again that OKN and PL measures have previously produced equivalent results in visual acuity testing (Dobson & Teller, 1978). Although methodological issues might account for the differences between this and previous studies, the disparate results are also well described by proposing differential development of two mechanisms in infant vision; one sensitive to uniform and one sensitive to relative motion. A developmental dissociation of sensitivity to uniform and relative motion is consistent with the large body of evidence for separate uniform and relative motion detection mechanisms in adults. In addition, an early differentiation of uniform and relative
INFANTS' SENSITIVITY TO UNIFORM MOTION motion mechanisms is consistent with the importance of distinguishing self motion from object motion at all ages (Bridgeman, 1972; Frost et al., 1990). Whether or not this interpretation is correct remains to be tested directly, and current research in our lab is directed toward this goal. For now, we discuss several physiologically plausible frameworks which would support the differential development of uniform and relative motion sensitivity. One possible framework is a hierarchical scheme in which relative motion mechanisms are built from the outputs of uniform motion mechanisms. Models of other complex motions utilize this type of hierarchy to some extent. Rotation and expansion have been modeled by suggesting that complex motion detectors in MST may be created from the convergence of directionally selective cells with small receptive fields in MT (Tanaka et al., 1989). However, Tanaka et al. (1989) deviated from this framework by further proposing that full-field uniform motion detectors may be created from the convergence of M T cells which are locally sensitive to uniform motion. From a developmental perspective, a hierarchical framework predicts that relative motion mechanisms would follow the development of the simpler uniform motion mechanisms from which they are derived. This prediction is consistent with previous empirical data, which suggests that the rapid increase in sensitivity to uniform motion occurs prior to the improvement in sensitivity to relative motion. Thought of in this way, complex motion detectors may develop after uniform motion detectors. A second physiological framework supporting the dissociated development of uniform and relative motion sensitivity is one in which uniform and relative motion thresholds are constrained by the development of independent cortical spatial frequency mechanisms. Uniform motion may primarily stimulate low spatial frequency mechanisms, while relative motion might stimulate independent higher spatial frequency channels. Furthermore, low spatial frequency channels develop more rapidly than high spatial frequency channels (Atkinson et al., 1974; Banks & Salapatek, 1978), suggesting that a sensitivity to uniform motion may develop prior to a sensitivity to relative motion. In the present study, the plateau in sensitivity to uniform motion between 6 and 18 weeks could reflect early development of low spatial frequency mechanisms which are likely to govern uniform motion. Along these lines, the marked improvements in relative motion sensitivity noted between 10 and 20 weeks of age in previous studies could reflect the later development of high spatial frequency mechanisms. Interestingly, Manny and Fern (1990) found no change in direction discrimination with age when the stimulus spatial frequency was adjusted to that producing peak sensitivity at each age. Additional data regarding the development of size-tuned motion mechanisms are required to properly evaluate this hypothesis. In summary, infants as young as 6 weeks of age possess visual mechanisms sensitive to uniform motion. This finding is important, because it challenges the claim
1639
that direction sensitivity emerges after 8 weeks of age (e.g. Braddick, 1993). We suggest that the development of motion processing may follow multiple courses, depending on the type of motion under investigation.
REFERENCES Allman, J., Miezin, F. & McGuinness, E. (1985). Direction- and velocity-specificresponses from beyond the classical receptive field in the middle temporal visual area (MT). Perception, 14, 105-126. Apkarian, P. (1993). Temporal frequency responsivity shows multiple maturational phases: State-dependent visual evoked potential luminance flicker fusion from birth to 9 months. Visual Neuroscience, 10, 1007-1018. Aslin, R. N. & Shea, S. L. (1990). Velocity thresholds in human infants: Implications for the perception of motion. Developmental Psychology, 26, 589-598. Atkinson, J. & Braddick, O. J. (1981). Development of optokinetic nystagmus in infants: An indicator of cortical binocularity? In Fisher, D. F., Monty, R. A. and Senders, J. W. (Eds), Eye movements: Cognition and visual perception (pp. 53-64). Hiilsdale, NJ: Erlbaum. Atkinson, J., Braddick, O. & Braddick, F. (1974). Acuity and contrast sensitivity of infant vision. Nature, London, 247, 403-404. Banks, M. S. & Salapatek, P. (1978). Acuity and contrast sensitivity in 1-, 2- and 3-month-old human infants. Investigative Ophthalmology and Visual Science, 17, 361-365. Banton, T. & Bertenthal, B. I. (1995). The effect of target size on infant detection of relative motion. Investigative Ophthalmology and Visual Science, 36, $909. Bertenthat, B. I. & Bradbury, A. (1992). Infants' detection of shearing motion in random-dot displays. Developmental Psychology, 28, 1056-1066. Boothe, R. G., Dobson, V. & Teller, D. Y. (1985). Postnatal development of vision in human and nonhuman primates. Annual Review of Neuroscience, 8, 495-545. Braddick, O. (1993). Orientation- and motion-selectivemechanismsin infants. In Simons, K. (Ed.), Early visual development; normal and abnormal (pp. 163-177). New York: Oxford University Press. Bridgeman, B. (1972). Visual receptive fields sensitive to absolute and relative motion during tracking. Science, 178, 1106-1108. Britten, K. H., Shadlen, M. N., Newsome, W. T. & Movshon, J. A. (1993). Responses of neurons in macaque MT to stochastic motion signals. Visual Neuroscience, 10, 1157-1169. Dannemiller, J. L. & Freedland, R. (1989). The detection of slow stimulus movement in 2- to 5-month-olds.Journal of Experimental Child Psychology, 47, 337-355. Dannemiller, J. L. & Freedland, R. (1991). Detection of relative motion by human infants. Developmental Psychology, 27, 67-78. Davidson, R. M. & Bender, D. B. (1991). Selectivity for relative motion in the monkey superior colliculus. Journal of Neurophysiology, 65, 1115-1133. Dobson, V. & Teller, D. Y. (1978). Visual acuity in human infants: A review and comparison of behavioral and electrophysiological studies. Vision Research, 18, 1469-1483. Freedland, R. L. & Dannemiller, J. L. (1987). Detection of stimulus motion in 5-month-old infants.Journal of Experimental Psychology: Human Perception and Performance, 13, 566-576. Frost, B. J. (1978). Moving background patterns alter directionally specific responses of pigeon tectal neurons. Brain Research, 151, 599-603. Frost, B. J., Wylie, D. R. & Wang, Y. C. (1990). The processing of object and self-motion in the tectofugal and accessory optic pathways of birds. Special Issue: Optics, physiology and vision. Vision Research, 30, 1677-1688. Giaschi, D. E. & Regan, D. (1995). Dissociated visual development in the processing of motion-defined and luminance-defined form. Investigative Ophthalmology and Visual Science, 36, $443. Goodale, M. A. & Milner, A. D. (1992). Separate visual pathways for perception and action. Trends in Neurosciences, 15, 20--25.
1640
T. BANTON and B. I. BERTENTHAL
von Grunau, M. W. & Frost, B. J. (1983). Double opponent-process mechanism underlying receptive field structure of directionally specific cells of cat lateral suprasylvian visual area. Experimental Brain Research, 49, 84-92. Gulyas, B., Orban, G. A., Duysens, J. & Maes, H. (1987). The suppressive influence of moving textured backgrounds on responses of cat striate neurons to moving bars. Journal of Neurophysiology, 57, 1767-1791. Hamer, R. D. & Norcia, A. M. (1994). The development of motion sensitivity during the first year of life. Vision Research, 34, 23872402. Hammond, P., Ahmed, B. & Smith, A. T. (1986). Relative motion sensitivity in cat striate cortex as a function of stimulus direction. Brain Research, 386, 93-104. Hammond, P. & Smith, A. T. (1982). On the sensitivity of complex cells in feline striate cortex to relative motion. Experimental Brain Research, 47, 457~160. Hammond, P. & Smith, A. T. (1984). Sensitivity of complex cells in cat striate cortex to relative motion. Brain Research, 301,287-298. Jouen, F., Lepecq, J. C. & Gapenne, O. (1996). Optical flow sensitivity in neonates. Manuscript submitted for publication. Kaufmann, F., Stucki, M. & Kaufmann-Hayoz, H. R. (1985). Development of infants' sensitivity for slow and rapid motions. Infant Behavior and Development, 8, 89-98. Manny, R. E. & Fern, K. D. (1990). Motion coherence in infants. Vision Research, 30, 1319-1329. Manny, R. E. & Klein, S. A. (1985). A three alternative tracking paradigm to measure vernier acuity of older infants. Vision Research, 25, 1245-1252. Mikami, A. & Fujita, K. (1992). Development of the ability to detect visual motion in infant macaque monkeys. Developmental Psychobiology, 25, 345-354. Nakayama, K. & Tyler, C. W. (1981). Psychophysical isolation of movement sensitivity by removal of familiar position cues. Vision
Orban, G. A., Gulyas, B. & Vogels, R. (1987b). Influence of a moving textured background on direction selectivity of cat striate neurons. Journal of Neurophysiology, 57, 1792-1812. Regan, D., Giaschi, D., Sharpe, J. A. & Hong, X. H. (1992). Visual processing of motion-defined form: Selective failure in patients with parietotemporal lesions. The Journal of Neuroscience, 12, 21982210. Shimojo, S., Bauer, J., O'Connell, K. M. & Held, R. (1986). Prestereoptic binocular vision in infants. Vision Research, 26, 501-510. Shimojo, S., Birch, E. E., Gwiazda, J. & Held, R. (1984). Development of vernier acuity in infants. Vision Research, 24, 721-728. Smeets, J. B. J. & Brenner, E. (1994). The difference between the perception of absolute and relative motion: A reaction time study. Vision Research, 34, 191-195. Snowden, R. J. (1992). Sensitivity to relative and absolute motion.
Perception, 21,563-568. Sterling, P. & Wickelgren, B. (1969). Visual receptive fields in the superior colliculus of the cat. Journal ofNeurophysiology, 32, 1-15. Tanaka, K., Fukada, Y. & Saito, H. (1989). Underlying mechanisms of the response specificity of expansion/contraction and rotation cells in the dorsal part of the medial superior temporal area of the macaque monkey. Journal of Neurophysiology, 62, 642-656. Volkmann, F. C. & Dobson, M. V. (1976). Infant responses of ocular fixation to moving visual stimuli. Journal of Experimental Child Psychology, 22, 86-99. Wattam-Bell, J. (199l). Development of motion-specific cortical responses in infancy. Vision Research, 31,287-297. Wattam-Bell, J. (1992). The development of maximum displacement limits for discrimination of motion direction in infancy. Vision Research, 32, 621-630. Wattam-Bell, J. (1994). Coherence thresholds for discrimination of motion direction in infants. Vision Research, 34, 877-883. Wilson, H. R. (1988). Development of spatiotemporal mechanisms in infant vision. Vision Research, 28, 611-628.
Research, 21,427--433. Orban, G. A., Gulyas, B. & Spileers, W. (1987a). A moving noise background modulates responses to moving bars of monkey V2 cells but not of monkey V1 cells. Investigative Ophthalmology and Visual Science Supplement, 28, 197. Orban, G. A., Gulyas, B. & Spileers, W. (1988). Influence of moving textured backgrounds on responses of cat area 18 cells to moving bars. In Hicks, T. P. & Benedek, G. (Eds), Progress in brain research (pp. 137-145). Amsterdam: Elsevier Science Publishers B.V.
Acknowledgements--We thank our trained observers, Tara Speakman and Heather Tevendale, for their expert assistance in the data collection. A preliminary account of this research was presented at the annual meeting of the Association for Research in Vision and Ophthalmology, Sarasota, Florida, 1993. This work is supported by NIH Grant HD16195 to Bennett I. Bertenthal.