Direction-selective adaptation from implied motion in infancy

Abstract

We investigated whether adaptation from implied motion (IM) is transferred to real motion using optokinetic nystagmus (OKN) in infants. Specifically, we examined whether viewing a series of images depicting motion shifted infants’ OKN responses to the opposite direction of random dot kinematograms (RDKs). Each RDK was presented 10 times in a pre-test, followed by 10 trials of IM adaptation and test. During the pre-test, the signal dots of the RDK moved left or right. During IM adaptation, 10 randomly selected images depicting leftward (or rightward) IM were presented. In the test, the RDK was presented immediately after the last IM image. An observer, blinded to the motion direction, assessed the OKN direction. The number of matches in OKN responses for each RDK direction was calculated as the match ratio of OKN. We conducted a two-way mixed analysis of variance, with age group (5–6 months and 7–8 months) as the between-participant factor and adaptation (pre-test and test) as the within-participant factor. Only in 7–8 months the OKN responses were shifted in the opposite direction of RDK by viewing a series of images depicting motion, and these infants could detect both IM and RDK motion directions in the pre-test. Our results indicate that detecting the IM and RDK directions might induce direction-selective adaptation in 7–8 months.

Introduction

In this study, we investigate the generalization of implied motion (IM) to real motion (RM) during infancy. To this end, we applied motion after-effects (MAEs) from IM to RM in infants. A previous study used a motion after-effect paradigm to determine whether the same neural circuits are employed by inference and perception in the visual motion domain (Winawer, Huk, & Boroditsky, 2008). MAEs are a well-studied psychophysical phenomenon (Mather, Verstraten, & Anstis, 1998; Wohlgemuth, 1911) and have been used to assess the properties of direction-selective neural mechanisms. The adapted viewing of motion in one direction causes subsequently viewed stationary or ambiguous patterns to appear to move in the opposite direction. This illusion is thought to result from an adaptation-induced reduction in the activity of directionally selective neurons that respond to direction-adapted motion (Barlow & Hill, 1963; Mather et al., 1998; Sutherland, 1961). Winawer et al. (2008) predicted that, if viewing IM images relies on the same direction–selective neurons involved in the perception of physical motion, then viewing a series of photographs depicting same direction motion would adapt direction-selective neurons and produce an MAE. They tested this hypothesis to determine whether viewing an IM in one direction altered the perceived direction of subsequently presented RM. Their results showed that viewing a series of static photographs with implied motion in a particular direction produced MAEs in the opposite direction. This demonstrates that the perception of implied motion activates direction-selective circuits that are also involved in processing real motion. 

In more recent studies (Shirai & Imura, 2014; Shirai & Imura, 2016) demonstrated that infants are sensitive to IM images. They showed that a static figure depicting a running (not standing) person shifted the visual preference toward a target that subsequently appeared in the cued running direction for 5–8 months (Shirai & Imura, 2014), but not for 4-month-olds (Shirai & Imura, 2016). They tested infants’ IM perception using the forced-choice preferential-looking method. Specifically, in their experiment, a full-colored still image of a young male either running or standing facing a lateral side was presented at a rate of 600 ms. After the presentation of the still image, two black disk targets simultaneously appeared on both the left and right sides of the previous image. Their results showed that the running image of a still person significantly enhanced infants’ preference for a visual target that appeared consistently on the same side as the running direction. This result suggests that the implied body motion figure shifted infants’ visual preference in the same direction as the implied running. This means that even infants as young as 5–8 months are sensitive to the IM of static images. The perception of more complex IM was investigated by Friedman and Stevenson (1975) by using various human actions with motion lines for preschoolers and first- and sixth-grade elementary school children. Their results indicated that sixth graders identified movement in the pictures with motion lines, but preschoolers and first-graders did not. Similarly, Carello, Rosenblum, and Grosofsky (1986) compared several pictorial devices to assess their relative effectiveness in depicting events for 3- to 4-year-olds. They showed that the postural expressions were an effective movement device for 3- to 4-year-olds. Thus, the ability to utilize form information to perceive directionality would develop from 5 to 8 months until around 4 years of age. 

In this study, we used a motion adaptation paradigm. Adaptation has been previously used in studies of light adaptation in infants (Dannemiller & Banks, 1983; Hamer, Dobson, & Mayer, 1984; Hansen, Hamer, & Fulton, 1992; Rasengane, Palmer, & Teller, 2001). For example, Rasengane et al. (2001) investigated the dependence of photopic contrast thresholds on retinal illuminance in 3-month-old infants. The stimulus was a 0.25 c/° square-wave grating phase alternated at 6 Hz, and the forced-choice preferential looking technique was used. They showed that infants’ contrast thresholds were more than 2 log units higher than those of adults for all retinal illuminances. In infants, the contrast threshold initially decreased with increasing retinal illuminance. Above a critical illuminance of approximately 200 Td, the contrast thresholds remained constant, following Weber's law. However, no infant studies have used the motion adaptation paradigm. 

In this study, an optokinetic nystagmus (OKN) response was used. OKN is a series of reflexive eye movements elicited by a repetitive pattern of movement through the visual field. OKN consists of two alternating eye movement phases: a slow phase or pursuit eye movements in the direction of movement and a fast phase or saccadic eye movements in the opposite direction. OKN has been used in infant vision studies (Lewis, Maurer, Chung, Holmes-Shannon, & Van Schaik, 2000; Manny & Fern, 1990; Mason, Braddick, & Wattam-Bell, 2003; Yu et al., 2013). For example, Lewis et al. (2000) measured OKN to quantify OKN asymmetry for 3- to 24-month-olds. Two testers observed an infant’s one eye through peepholes on either side of the screen. One of four trained testers served as the primary tester, and one of three undergraduate students served as the secondary tester. The primary tester always observed through the peephole closest to the infant's eye. Mason et al. (2003) compared the coherence threshold for motion direction for OKN and preferential-looking responses in infants 6–27 weeks. The observer viewed the infant on a video monitor linked to a camera placed above the center of the display. The observer judged the direction of motion based on the infant's eye movements and was allowed to view the infant's eye movements until they judged the direction of the OKN. Manny and Fern (1990) used OKN to investigate motion coherence in 1-, 2-, and 3-month-old infants. An observer, unaware of the direction of motion, judged the infant's OKN induced by the moving display on a monitor, following an eight-alternative eye movement voting paradigm. Again, the observer was allowed to view the infant's eye movements until they judged the direction of the OKN. Yu et al. (2013) developed and validated a technique for measuring global motion perception in 2-year-old children and demonstrated the relationship between global motion perception and stereoacuity in children. Random dot kinematogram (RDK) stimuli were used to measure motion coherence thresholds at 23–25 months of age. The RDKs of variable coherence were presented. Video recordings of OKN were assessed offline by an experienced assessor who graded the OKN for each trial as leftward, rightward, or no OKN, following a three-alternative forced-choice procedure. In accordance with these studies, in our study, an observer blinded to the direction of motion judged the direction of the OKN from recorded videos. The slow and fast phases of OKN were checked, and the directions of the slow phases were judged as the OKN direction (leftward or rightward) by a two-alternative forced choice. 

As shown above, recent studies (Shirai & Imura, 2014; Shirai & Imura, 2016) have shown that viewing still images depicting motion shifted visual preference in the same direction in 5- to 8-month-olds. Studies in adults (Winawer et al., 2008) used the adaptation of IM to show that viewing IM altered the perceived direction of the subsequently presented RM. This indicates that adaptation from IM is transferred to RM in adults. A main contribution of this study is that we used this adaptation paradigm in infants. We also examined whether viewing a series of images depicting motion altered the OKN response to the RDK in infants and assessed the OKN response to the RDK before and after IM adaptation (pre-test and test) as follows. After IM adaptation, the OKN response to the RDK shifted in the opposite direction. First, we conducted a two-way mixed analysis of variance (ANOVA) with age group (5–6 and 7–8 months old) as the between-participant factor and adaptation (pre-test and test) as the within-participant factor. Second, we tested whether the OKN response to RDK was observed more frequently than chance. Third, we tested whether the OKN responses to IM images were observed more frequently than by chance. Generally, the OKN responses appeared to follow a series of moving objects, so we predicted that no OKN response would be induced by IM images. However, a previous study (Castellotti, Francisci, & Del Viva, 2021) reported a pupil dilation change in response to IM; therefore, we measured the OKN response to IM. Finally, we investigated the generalization of IM to RM, using OKN during infancy. 

In our experiment, RDK with coherence and velocity was selected, based on previous studies on infants. Previous studies have demonstrated the minimum velocity and coherence thresholds of the RDK in infants (Banton & Bertenthal, 1996; Bertenthal & Bradbury, 1992; Mason et al., 2003; Wattam-Bell, 1994). Specifically, Bertenthal and Bradbury (1992) used RDK with a coherence of 100% and showed that the minimum velocity thresholds in 13- and 20-week-olds were 3.57°/s and 1.2°/s, respectively, using the forced-choice preferential looking technique (FPL). Wattam-Bell (1994) used RDK at a velocity of 8°/s, and showed that the coherence thresholds were approximately 50% in 3 months using FPL. Banton and Bertenthal (1996) used RDK with a coherence of 50% and showed that the minimum velocity thresholds in 6-, 12-, and 18-week-olds were 4.8°/s, 4.1°/s, and 6.9°/s, respectively. They also used RDK at a velocity of 7.4°/s and found that the minimum velocity thresholds in 6-, 12-, and 18-week-olds were 36%, 29%, and 37%, respectively, using OKN responses. Likewise, Mason et al. (2003) used RDK at a velocity of 9.27°/s. They determined that the mean coherence thresholds in 6- to 27-week-olds were 19.8% by using OKN responses and that preferential looking thresholds were significantly higher than OKN thresholds. In our experiment, we used a RDK with a coherence of 50% and a velocity of 6.32°/s, based on the thresholds of around 5- to 6-month-olds. In adult adaptation studies, the individual's threshold is usually used; however, for infants, because threshold measurements require time, data from previous studies were generally used in our study. 

Methods

Participants

A final total of 40 infants participated. The infants were divided into two age groups: 5–6 months (seven males and 13 females; mean ± SD age, 160.6 ± 18.39 days), and 7–8 months (14 males and six females; mean age, 228.7 ± 15.67 days). An additional five infants were excluded from the analysis because they could not complete all experimental trials because they were crying. The infants were recruited through local community flyers. All infants were born at full term and were healthy at the time of the experiment. This study was approved by the ethics committee of Chuo University. Prior to testing, written informed consent was obtained from the parents of the participating infants. 

Apparatus

All experiments were controlled using PsychoPy 3.0. The luminance of stimuli was determined using a chroma meter (CS-100A; Konica Minolta, Tokyo, Japan). The infants sat on the lap of one of their parents in front of the display at a viewing distance of 40 cm in a dark room. The viewing distance was established at 40 cm with the infant's head resting on the parent's chest. Two speakers were placed on each side of the monitor. A charge-coupled device (CCD) camera was set below the display to record the infant’s behavior so the experimenter could observe it. The experimenter initiated each trial as soon as the infant began paying attention to the fixation. 

Stimuli

The IM images were 10 full-color photographs (633 × 356 pixels, 32.5° × 18.61°) with either rightward or leftward IM. Images depicted animals in motion and were sourced from Internet searches (Figure 1). None of the images contained text, and all images were mirror-reversed so that they could be used in both directions. 

Figure 1.

 

Experimental procedure. RDKs were presented 10 times in pre-test, following 10 trials of IM adaptation and test. In IM adaptation, 10 randomly selected leftward (or rightward) IM images were presented at the center of the monitor at a rate of 600 ms per image, with no ISI. In test, RDKs were presented immediately following the last IM image. Test RDK and IM images had the same directions.

All stimuli were presented on a 32-inch liquid-crystal display (LCD) monitor with a refresh rate of 60 Hz and a resolution of 1920 × 1080 pixels. The RDK was presented at the center of the monitor for 2000 ms within 633 × 356 pixels (32.5° × 18.61°) and was composed of 100 white dots (diameter = 1.06°, 81 cd/m²) placed randomly on a gray background (45.8 cd/m²). Dots were randomly assigned to either the signal or noise dots and remained for the duration of the trial. The coherence of the dots was 50%. The signal dots had a speed of 2.0 pixels/frame (6.32°/s) and a lifetime of 367.4 ms (22 frames). Over their lifetimes, the signal dots were initialized and repositioned within the motion field. The noise dots disappeared in each frame and randomly reappeared at other locations. All stimuli were viewed binocularly and presented with beeping sounds. 

Procedure

First, RDKs were presented 10 times in the pre-test; the signal dots of the RDK moved to the left five times and to the right another five times. Then, the RDK was presented immediately after IM adaptation in the same direction for 10 trials. For the IM adaptation, 10 randomly selected leftward (or rightward) IM images were presented at the center of the monitor at a rate of 600 ms per image, with no interstimulus interval (ISI). The directions of the IMs and RDKs in the tests were to the left in half of the 10 trials and to the right in the other half. The directions were selected pseudo-randomly, with the same direction not being presented more than three consecutive times (Figure 1). 

An observer who was blinded to the direction of motion judged the direction of OKN based on the recorded videos. The slow and fast phases of OKN were checked, and the direction of the slow phases was judged as OKN (leftward or rightward) using a two-alternative forced choice. The observer judged the directions of OKN to the RDKs and IM images in the same way. In the pre-test, OKN directions were judged 10 times, and we counted the number of OKN response matches for each RDK direction. The match ratio of OKN was calculated as the number of matches of OKN responses in each RDK direction/10 times. The match ratio of OKN was calculated in the same way for IM adaptation and test. 

Results

We expected that IM adaptation would shift the OKN response to the opposite direction of the RDK in infants. The match ratios of OKN in the pre-test and test for each age group are displayed in Figure 2. We conducted a two-way mixed ANOVA with age group (5–6 and 7–8 months) as the between-participant factor and adaptation (pre-test and test) as the within-participant factor. The two-way ANOVA showed a significant interaction between age group and adaptation, F(1, 78) = 7.73, p < 0.01, η_p² = 0.09, but no other effects. The tests for the simple effects of adaptation in each age group showed that the match ratio of OKN in the pre-test was higher than in the test for 7–8 months, F(1, 39) = 4.72, p < 0.05, η_p² = 0.10, but there was no other effect of adaptation for 5–6 months, F(1, 39) = 0.34, p = 0.56, η_p² = 0.01. The strength of adaptation was calculated by subtracting the match ratio of OKN in the pre-test from that of the test in each age group (Figure 3). 

Figure 2.

 

Match ratios of OKN in pre-test and test. The dotted line represents the chance level. Error bars represent standard errors; *p < 0.05, **p < 0.01. The light gray dots indicate the individual data, and the light gray lines indicate the change from pre-test to test within individuals.

Figure 3.

 

Strength of IM adaptation in each age group. The dotted line represents the chance level, error bars represent standard errors, and the gray dots indicate individual data.

Second, we conducted a two-tailed one-sample t-test (vs. a chance level of 0.5) for the match ratios of OKN in the pre-test and test. For 7–8 months, the match ratio of OKN was higher than chance in the pre-test, t(19) = 3.73, p < 0.01, d = 1.67, but did not reach significance in the test, t(19) = –0.22, p = 0.82, d = 0.10. These results were not obtained for 5–6 months; the match ratio of OKN did not reach significance in the pre-test and test, t(19) = 1.76, p = 0.09, d = 0.79 and t(19) = 2.04, p = 0.06, d = 0.91, respectively. The match ratio of OKN was higher than chance only in the pre-test of 7–8 months. 

To confirm that a 2000-ms RDK presentation is sufficient to induce the OKN response, we obtained the OKN response using RDK with coherence of 80% for 5–8 months. Because this coherence is over the 3-month-olds’ threshold, this RDK certainly induced an OKN response in the 5–8 months. The results showed that the match ratio of OKN was higher than chance for both 5–6 and 7–8 months (5–6 month: t(19) = 6.37, p < 0.001, d = 2.85; 7–8 months: t(19) = 6.53, p < 0.001, d = 2.92; Figure 4). This indicates that 2000-ms RDK presentation would induce an OKN response to RDK in infants. 

Figure 4.

 

Match ratios of OKN to RDK with 80% coherence. The dotted line represents the chance level. Error bars represent standard errors; ***p < 0.001. The light gray dots indicate individual data.

Third, we also conducted a two-tailed one-sample t-test (vs. a chance level of 0.5) for the match ratios of OKN in the IM adaptation. The match ratio of OKN was higher than chance for 7–8 months, t(19) = 2.87, p < 0.01, d = 1.28, but did not reach significance for 5–6 months, t(19) = 0.21, p = 0.84, d = 0.09) (Figure 5). 

Figure 5.

 

Match ratio of OKN in IM adaptation. The dotted line represents the chance level. Error bars represent standard errors; **p < 0.01.

We conducted a correlation analysis between strength of adaptation and OKN response in IM (Figure 6). The strength of adaptation was plotted as a function of OKN response in IM for each participant. The strength of adaptation did not correlate with OKN response in IM for all age groups: for all age groups, r = 0.04, p = 0.81; for the 5–6 months age group, r = 0.02, p = 0.95; for the 7–8 months age group, r = –0.17, p = 0.48. 

Figure 6.

 

Correlation between strength of adaptation and OKN response in IM. Data showing strength of adaptation are plotted as a function of OKN response in IM. The solid line is the regression line fitted to the data for all age groups, and the dotted lines are the regression lines for each age group.

To examine gender difference, we conducted a three-way mixed ANOVA with age group (5–6 and 7–8 months) and gender group (males and females) as the between-participant factor and adaptation (pre-test and test) as the within-participant factor. The three-way ANOVA showed a significant main effect of age group, F(1, 76) = 7.22, p < 0.01, η_p² = 0.09, and a significant interaction between age group and adaptation, F(1, 76) = 6.05, p < 0.05. The tests for the simple effects of adaptation in each age group showed that the match ratio of OKN in the pre-test was higher than in the test for 7–8 months, F(1, 38) = 8.04, p < 0.01, η_p² = 0.17, but there was no other effect of adaptation for the 5–6 months, F(1, 38) = 0.37, p = 0.55, η_p² = 0.01. The match ratios of OKN for females were higher than those for males in both pre-test and test. Our results suggest gender differences in OKN responses to the RDK. However, previous infant OKN studies have not reported any gender differences (Atkinson, 1979; Hainline, Lemerise, Abramov, & Turkel, 1984; Naegele, 1983). Gender differences in infants’ OKN responses should be investigated in future studies. 

Last, to check infants’ habituation to our task in each age group, we compared looking times in the first two and the last two trials in test and pre-test for each age group. For all age groups, a two-tailed paired t-test revealed no differences between looking time in the first two and last two trials in pre-test and test: (5–6 months: pre-test; t(19) = 1.96, p = 0.06, test; t(19) = 1.04, p = 0.31 and 7–8 months: pre-test; t(19) = 1.29, p = 0.21, test; t(19) = 1.44, p = 0.17) (Figure 7). This means that all of the infants were not habituated and attentive to the task throughout the experiment. An infant's condition did not influence the results of the age difference. 

Figure 7.

 

Mean total looking time for RDK during pre-test and test. The first two and the last two trials in pre-test and test for each age group are shown. The error bars represent standard errors.

Discussion

This study investigated whether the adaptation from IM is transferred to RM using OKN in infants. Recent studies (Shirai & Imura, 2014; Shirai & Imura, 2016) have shown that viewing still images depicting motion shifted visual preference in the same direction for 5–8 month-olds. Studies on adults (Winawer et al., 2008) have shown that viewing IM altered the perceived direction of subsequently presented RM. We used this adaptation paradigm for infants and expected that viewing a series of images depicting motion shifted the OKN response to the opposite direction of the RDK in infants. 

The results showed that IM adaptation shifted the OKN response to the opposite direction of the RDK in 7–8 months, but not in 5–6 months. For 7–8 months, the match ratio of OKN in the pre-test was higher than chance (0.5), but not in the test. However, for 5–6 months, the match ratios of OKN in pre-test and test were not different from chance. In the IM adaptation, the match ratio of OKN was higher than chance for 7–8 months, but not for 5–6 months. This means that, using OKN responses, only 7–8 months could detect the IM direction. Thus, we consider that detecting both the RDK in pre-test and IM directions induces direction-selective adaptation in 7–8 months. It is possible that direction-selective adaptation might develop around the age of 7 months. 

To the best of our knowledge, this is the first study to investigate direction-selective adaptation in infants. Visual motion areas, such as the human middle temporal/medial superior temporal complex (MT+), respond more strongly when subjects view photographs or silhouettes of animals, people, objects, or natural scenes containing IM than when they view similar images not implying motion (e.g., a cup falling off a table compared to a cup resting on a table or a running athlete compared to an athlete at rest) (Kourtzi & Kanwisher, 2000; Lorteije et al., 2006; Peuskens, Vanrie, Verfaillie, & Orban, 2005; Senior et al., 2000). These studies demonstrated that IM can activate brain areas that are engaged in real-image motion. The 7–8 months infants who demonstrated IM adaptation may be mature in these brain areas. 

In our study, we used IM (one direction animal motion) as an adaptation. Perceiving IM is related to the biological motion (BM) perception. Body-related motion was perceived from the movements of point lights attached to major joints of an individual's body, even in the absence of form or textural information regarding body structure (e.g., Johansson, 1973). In developmental studies of BM, Bertenthal, Proffitt, and Kramer (1987) and Fox and McDaniel (1982) showed that the ability to discriminate BM figures from non-BM motion patterns does not develop until 3 to 4 months of age. A more recent study showed that even newborn babies could discriminate between BM and non-BM figures (Simion, Regolin, & Bulf, 2008). Hirai and Hiraki (2005) reported that, through the use of event-related potentials, brain activity in the right hemisphere in response to BM figures occurs at 8 months but not at 6 months of age. The IM was perceived by 8 months. 

For infants to perceive IM, maturation of the neural mechanisms involved in form–motion interaction is necessary. Visual motion perception from form information is related to the interactions between the dorsal and ventral pathways. Several neural studies have reported that static images activate motion-sensitive areas in the dorsal pathway, such as the hMT/MST+ (e.g., Fawcett, Hillebrand, & Singh, 2007; Kourtzi & Kanwisher, 2000; Lorteije et al., 2006; Lorteije et al., 2011; Senior et al., 2000). Krekelberg, Vatakis, and Kourtzi (2005) reported that the motion perception of the dynamic glass pattern, another type of directional perception from form information, is mediated by the interaction between the ventral (V4 and LOC) and dorsal (hMT+/V5, V3, V3a, and KO) streams. Therefore, infants’ ability to perceive IM may be related to the maturation of these cortical pathways. The ability to process global motion patterns emerges around 3 months of age, and the ability to process global form patterns develops by 4–5 months of age, which may reflect the maturation of relevant cortical visual pathways and the dorsal and ventral pathways, respectively (Braddick & Atkinson, 2007). This suggests that IM perception, which is related to the interaction between form- and motion-sensitive visual pathways, emerges at approximately 4–5 months of age. 

Previous IM studies on infants (Shirai & Imura, 2014; Shirai & Imura, 2016) showed IM direction detection in 5–6 months, whereas we did not show IM adaptation in 5–6 months. A possible reason for this difference is that IM direction detection and IM adaptation depend on different processes. In IM direction detection, the motion perception from form information is mediated by the interaction between the ventral and dorsal streams. On the other hand, in IM adaptation, generalization of motion information from IM to RM is required, and the visual motion area should be active in both IM and RM. That is, the human middle temporal/medial superior temporal complex (MT+) should respond to IM and RM. This IM adaptation processing may require development around 6 months of age. Also, the previous study and our study differed in terms of the objects of the IM images. Shirai and Imura (2014); Shirai and Imura (2016) used human IM images, whereas we used animal IM images. Different infant experiences with the motion of humans and animals may be involved with our results. This possible effect of visual experience on IM should be clarified in future studies by using various IM images, such as humans, animals, and cars. 

Recent IM adaptation studies in adults have debated the involvement of visual awareness in motion-direction decisions. Our infant study provides new evidence for this argument. For example, Faivre and Koch (2014) found that visible but not invisible IM adaptors biased the perception of RM by using continuous flash suppression. They concluded that, although invisible IM undergoes some form of nonconscious processing, visual awareness is necessary to make inferences about motion direction. Also, Gallagher, Suddendorf, and Arnold (2021) recently proposed that the implied motion after-effect produces a bias in direction decision-making. Such adult studies have shown that awareness is involved in IM adaptation. Compared to these studies, in our infant IM adaptation experiments for 7–8 months, IM adaptation was found, but not for 5–6 months. Based on adults’ IM studies, this age difference should reflect the differences in conscious processing; that is, some sort of conscious processing may be acquired at approximately 7 to 8 months of age. Our results are consistent with recent evidence. Tsurumi, Kanazawa, and Yamaguchi (2023) suggested that the visual world of infants younger than 6 months is functionally different from that of older infants and adults. Moreover, Nakashima, Kanazawa, and Yamaguchi (2021) showed that recurrent processing is immature in infants under 6 months of age and that infants can perceive visual stimuli without mature recurrent processing, based on anatomical studies showing immature recurrent pathways in early infancy (Burkhalter 1993; Burkhalter, Bernardo, & Charles, 1993). 

Unexpectedly, we found evidence of OKN responses to IM images in 7–8 months. Although static images would not induce an OKN response, in our experiment an OKN response to IM was observed. It cannot be definitively identified as an OKN response, but it is an OKN-like eye movement. Castellotti et al. (2021) reported that involuntary eye responses are involved in IM. They measured the pupil diameter by viewing RM and IM. Their results showed that IM induced more pupil dilation change than no IM static images, but RM elicited larger pupillary dilation than IM. This suggests that involuntary pupil responses are involved in IM. Based on this study, the OKN response in IM adaptation among the 7–8 months in our study should be related to this involuntary pupil response. The relationship between infants’ OKN responses in the IM and eye movements may require further investigation. 

Acknowledgments

Supported by a grant from the Japan Society for the Promotion of Science (JSPS) Kakenhi (JP 20H00597). 

Commercial relationships: none. 

Corresponding author: Riku Umekawa. 

Email: [e-mail]. 

Address: Department of Psychology, Chuo University, Hachioji, Tokyo, Japan.