The authors were financially supported by the Grants-in-Aid for Scientific Research from Japan Society for the Promotion of Science (18000090 to M.K.Y. and 21830042 to N.S.), and the Grant for Promotion of Research by Faculty of Humanities, Niigata University (to N.S.).
Radial optic flow is one of the crucial cues for the perception of motion-in-depth and contributes to our daily adaptive actions such as locomotion or postural control. Although many researchers have examined the development of radial motion perception and that of adaptive motor actions, no valid developmental model for visuo-motor coordination has been proposed. In the present study, we tried to propose a developmental framework for the interactive process between visual radial motion perception and the adaptive motor actions in infancy, with a brief review of the previous psychophysical, psychophysiological, and neurophysiological studies. The effect of the functional development of the posterior parietal cortex, which might be induced by the experiences of ego motion in particular developmental periods, on visuo-motor coordination was discussed.
A visual radial expansion/contraction motion can create compelling perception of a motion-in-depth event (De Bruyn & Orban, 1990; Johansson, 1964) such as the observer's own forward/backward motion or an approaching/receding object. A number of theoretical and empirical studies have claimed that the perception of radial motion has an important role in our various daily actions related to motion-in-depth perception. For instance, the ability to voluntarily explore the environment is one of the most fundamental and important abilities for animals, including human beings. When we move through the environment we tend to face the front parallel plane. This situation typically causes a large-field radial optic flow onto our retina during locomotion. James J. Gibson, who is the first person to point out the important role of radial flow in controlling ego motion, claimed that radial optic flow contributes to perceive the direction in which one is heading, so that the observer can control her/his locomotion (Gibson, 1950, 1979). Gibson's idea of the relationship between radial optic flow and ego motion has been empirically confirmed later by a number of studies (see the reviews of Lappe, Bremmer, & van den Berg, 1999; Warren, 1998; but see also Harris & Rogers, 1999). The other important function related to radial motion perception in our adaptive actions is detection of approaching objects. We can detect an accurate time to contact (TTC) with an approaching object by calculating “visual tau” (Lee, 1976). The visual tau is the ratio of the visual angle of an approaching object to the rate of change of this angle at an arbitrary moment. Estimating TTC is useful for avoiding a dangerous collision with an approaching obstacle or intercepting an approaching object (Savelsbergh, Whiting, & Bootsma, 1991).
Because radial motion perception and the relevant motor actions seem to be quite fundamental and indispensable abilities for animals, a number of studies have concerned this research topic, as we mentioned above. One of the key areas of interest in this research topic is how an individual acquires the ability to tie up visual radial motion perception with various motor actions. Indeed, many researchers have engaged in studies for the development of adaptive motor actions possibly relevant to radial motion perception, such as avoidant responses to collision events or the effect of optic flow information on postural control, in recent decades (see the sections “Development of defensive motor responses to visual radial motion” and “Development of ego motion perception from visual radial motion” of this review). Although these previous studies have provided detailed trends in the development of adaptive actions, no-one has mentioned how the interaction between visual radial motion perception and motor actions develops in early infancy. The difficulties in reasoning the developmental bases of such visuo-motor interactions might partly due to relatively poor knowledge about the development of radial motion perception in infancy. However, recent psychophysical and psychophysiological studies have revealed the early development of radial motion processing in infancy (see the section “Early development of radial motion perception” of this review). This means that now we can discuss the development of the interaction between visual radial motion perception and the execution of relevant motor actions.
The goal of the present review is to propose a developmental framework for the integrative process of visual radial motion and motor actions. Firstly, we introduce recent studies that have focused on the development of radial motion perception. Understanding the development of radial motion perception would aid in considering the emergence of the interaction between radial motion perception and control of ego motion. After that, we review the previous empirical findings that potentially reflect the development of interactions between radial motion perception and the ability to control ego motion, and finally discuss a possible developmental framework for the interactions between radial motion perception and relevant adaptive actions.
In contrast, the ability to detect radial motion emerges relatively later than that to detect single directional motion. Shirai and his colleagues have examined infants' radial motion perception in various experimental conditions. They have investigated, in early infancy, the detection of radial motion in a visual search display (Shirai, Kanazawa, & Yamaguchi, 2004a), the detection of speed-gradient in a radial motion pattern (Shirai, Kanazawa, & Yamaguchi, 2004b), coherence thresholds for the detection of radial expansion/contraction (Shirai, Kanazawa, & Yamaguchi, 2006), and speed sensitivity to radial motion (Shirai, Kanazawa, & Yamaguchi, 2008a) using the preferential-looking technique. Despite the difference in experimental conditions, they found a common trend in the development of radial motion perception in infancy: the ability to detect radial motion pattern is severely limited at 2 months of age, and rapidly develops until 3 months of age. Moreover, they found that significant cortical responses to radial motion also emerge at approximately 3–4 months of age (Shirai, Birtles, Wattam-Bell, Yamaguchi, Kanazawa, Atkinson, & Braddick, 2009; but see Gilmore, Hou, Pettet, & Norcia, 2007). Soon after the rapid development of radial motion perception at approximately 3 months of age, a developmental plateau has been observed. Gilmore and his colleagues have reported that the ability to discriminate the position of the center of the radial motion patterns did not develop significantly between 3 and 6 months old (Gilmore, Baker, & Grobman, 2004; Gilmore & Rettke, 2003). Summarizing these results, radial motion perception rapidly develops at approximately 3 months of age, but shows little or no change after this period. A recent longitudinal study carried out by Brosseau-Lachaine, Casanova, and Faubert (2008), who measured infants' coherence thresholds for the detection of radial motion patterns, confirmed this developmental trend. They demonstrated that radial motion sensitivity changed significantly between 2 and 4 months of age, but did not change through the developmental period from 4–8 months of age. Note that although infants aged approximately 3–4 months show quite similar properties to adults in radial motion perception (Gilmore et al., 2004; Gilmore & Rettke, 2003; Shirai et al., 2004a, 2004b, 2006), their sensitivity to radial motion (defined by coherence threshold, Brosseau-Lachaine et al., 2008) is significantly lower than that of adults. This implies that even though the fundamental functions of radial motion perception emerge at 3–4 months old, infants' radial motion perception is still immature until at least 8 months of age (see the lower column of Figure 1).
One might attribute the difference in the developmental trends between single directional and radial motion perception to distinct complexities of those motion patterns. A radial motion pattern contains a variety of motion components, while a single directional pattern has uniform components. The difference in the variety of motion components might simply affect the infants' detection of a particular motion pattern: it seems to be quite natural that an ability to process more complex patterns develops more slowly. However, the developmental trend for radial motion perception is different from that of rotational motion perception, which has virtually the same variety of motion components as radial motion. For instance, 4-month-old infants could perceive the complete shape of a partly occluded object behind an approaching/receding (retinally expanding/contracting) occluder (Kellman, Spelke, & Short, 1986), while infants in almost the same age range could not perceive the shape of an object occluded by a rotating occluder (Eizenman & Bertenthal, 1998). More recently, Shirai, Kanazawa, and Yamaguchi (2008b) directly compared the development of rotational motion speed sensitivity to that of radial motion speed sensitivity. They found that 3-month-olds showed poorer performance in the detection of a rotational motion pattern when the pattern had a lower speed value, while the speed value of the radial motion had little effect on the detection performance of infants of the same age. They also revealed that rotational motion speed sensitivity did not improve between 2 and 3 months of age, unlike radial motion sensitivity.
These previous findings suggest that radial motion perception has a unique developmental trend in early infancy compared with the perception of other motion patterns. Because it has been often assumed that different visual mechanisms have different developmental rates (cf. Dannemiller, 2001), the developmental uniqueness of radial motion perception implies that even young infants have the specialized mechanism(s) for radial motion processing. The specialized mechanism(s) may have an important role in controlling various adaptive actions of infants. The point is, which of the neural bases coordinates the interactions between radial motion perception and control of various adaptive actions in early infancy? Before discussing this point, in the following sections we will survey previous findings regarding the development of adaptive actions that are potentially related to radial motion perception.
Development of defensive motor responses to visual radial motion
The earliest studies of the development of radial motion perception were conducted by researchers who investigated defensive motor responses to a radial motion pattern in infancy. The defensive responses were motor responses, such as backward body motion, avoidant head rotation, or blinking, elicited by the presentation of a radial expansion pattern. Such defensive responses seem to be useful for avoiding serious injury caused by a collision with an approaching object that casts a radial expanding image onto the observer's retina.
The defensive responses of young infants can be interpreted as evidence that young infants have the ability to process visual radial motion in order to execute adaptive motor responses. However, there is the possibility that the defensive responses in early infancy are elicited by simple local physical parameter(s) contained in a large field expanding pattern, but not by global radial motion information as such. For instance, Yonas et al. (1977) have demonstrated that the infants' smooth pursuit looking behavior to the upper edge of a radial expanding pattern resulted in their backward head rotations to an expanding pattern. Moreover, rapid change in the luminance across the large visual field (for instance, the situation in which a bright screen was suddenly turned off) could elicit similar motor actions to the defensive responses (Nanez, 1988; Yonas et al., 1979). These results suggest that the young infants possibly generated these defensive responses without perception of radial motion. Note that we never intend to deny the fact that young infants have the ability to react adaptively to radial motion information. Our position simply means that the ability may be driven by particular combination(s) of some local physical properties of a visual radial motion pattern, such as a rapid change in luminance in the broad central area of the visual field.
The idea that the defensive responses are driven by simple factors contained in a visual radial motion stimulus rather than radial motion information per se corresponds to the findings of recent studies that have focused on the development of radial motion processing in early infancy (see the section “Early development of radial motion perception”). While defensive responses are observed in infants younger than 1 month of age, radial motion perception emerges at approximately 3–4 months old (Brosseau-Lachaine et al., 2008; Shirai et al., 2004a, 2004b, 2006, 2008a). That radial motion perception develops at a later stage than the defensive responses suggests that the defensive responses may be driven by mechanisms independent of the visual processing of radial motion information. Although the defensive responses to radial motion patterns in early infancy potentially reflect the infants' radial motion perception, the responses could also be interpreted as reactions to simple physical properties included in a radial motion pattern (see also Introduction of Shirai et al., 2009).
Development of ego motion perception from visual radial motion
In the previous section, we discussed the development of defensive responses to a radial motion pattern, which may have an important role in reacting to “a movement of an object” in the environment. Although there is a possibility that defensive responses are driven by some local physical properties of a radial motion pattern, but not global radial motion information per se, such defensive responses may contribute to the avoidance of a collision with a moving object. Another important aspect of the radial motion pattern is that it serves as a cue for the perception of the observer's own motion (ego motion). A strong illusion of ego motion (vection) to forward/backward movement occurs when we observe a large field radial expansion/contraction motion pattern. In this section, we review previous studies that have investigated the ability to perceive and control ego motion-in-depth, based on radial optic flow in infancy.
Several studies using the “moving room” paradigm have revealed that human infants perceive an illusory ego motion from an optic flow stimulus that might contain radial motion information. The moving room paradigm was developed by Lee and Aronson (1974), who examined whether dynamic visual information affects infants' postural control. A moving room is a large experimental chamber without a floor and hung from the ceiling, so that the room can be swayed forward/backward independent of the observer's body movement. Such sway-in-depth of the room produces an optic flow containing a radial motion pattern on the observer's retina. In their study, infants aged 13–16 months who could stand independently on their legs participated. They found that infants who stood in the room tilted their body in the same direction as the movement of the room: when the infants were exposed to forward (or backward) motion of the room, they showed forward (or backward) body-tilt. These results indicated that the infants could utilize visual motion to perceive their ego motion. Later, Stoffregen, Schmuckler, and Gibson (1987) found that the interaction of optic flow perception and postural control in infants was relatively similar to that of adults. It has been seen that adult observers show a stronger visuo-motor interaction between optic flow perception and postural control when the optic flow stimuli are presented in the peripheral visual field than when they are presented in the central field (Stoffregen, 1985, 1986). Stoffregen et al. (1987) tested children whose age ranged from 10 months to 5 years and indicated that even the youngest observers showed a stronger visuo-motor interaction when the optic flow stimuli were presented in the peripheral visual field than when they were presented in the central field. This trend of peripheral bias was also observed with infants who could not stand by themselves. Bertenthal and Bai (1989) investigated postural compensations for the visually represented body motion of 5-, 7-, and 9-month-old infants who were sat in a moving room. They found that the 7- and 9-month-olds responded well to the peripherally presented motion patterns, but not the 5-month-olds. These previous studies revealed that the interaction between radial motion perception and ego motion perception that is comparable to that of adults emerges during the second half of the first year of life.
It is plausible that the interaction between optic flow and ego motion perception was led by the experience of ego motion accompanied by the development of locomotive skills. Higgins, Campos, and Kermoian (1996) examined whether locomotion experiences effect infants' postural control against the optic flow information. They tested 8-month-old infants who had experienced locomotion by creeping or by using a walker, or who were pre-locomotors. The results indicated that both the creeping and the walker-using infants showed greater body sways synchronized with the direction of the moving room than the pre-locomotor infants. Their results imply that locomotor experience promotes the development of an interaction between optic flow perception and ego motion perception. This view has been also supported indirectly by a study that investigated infants' postural control during sitting. Bertenthal, Rose, and Bai (1997) tested 5-, 7-, 9-, and 13-month-olds and reported that when the infants sat in the moving room, the older infants showed a more systematic body sway that was tuned to the movement of the room. They suggested that the experience of independent sitting might promote the ability to coordinate visual and sensory motor information. These previous findings have shown that experience of independent motor actions such as locomotion or postural control is an important factor for developing the interaction between radial optic flow perception and adaptive motor actions.
As shown by these previous studies, the interaction between visual radial motion perception and motor control skills is elaborated in parallel with the increments in experience in independent ego motion or postural control at approximately the second half of the first year of life. Such visuo-motor interaction would be realized by the functional development of particular neural mechanism(s), which may be related to multisensory processing. Although it is reasonable to assume that experience of ego motion promotes the development of potential multisensory neural bases, there may be some other possible reasons for the development of visuo-motor interaction at that age. That is, basic neural links between the mechanism for radial motion perception and that for motor controls potentially exists even at an earlier stage of life, but it is not remarkable in the earlier stages because of the immaturity of either the radial motion perception or biomechanical functions. For example, remarkable visuo-motor interactions may be led by the development of radial motion perception: the radial motion perception may improve and be rich enough to guide the adaptive motor actions during that age; or the biomechanical functions, such as the strength of the muscles, develop at around that age, so that infants of that age can act explicitly based on visually perceived radial motion. However, those two possibilities are rejected by the findings of the previous studies. First, visuo-motor development should not be a consequence of the improvement of radial motion perception. Note that radial motion perception develops rapidly during the first few months of life, and soon reaches a developmental plateau (see the section “Early development of radial motion perception”). This means that the change in radial motion perception is not a main factor for the emergence of visuo-motor interaction after 6–7 months of age. Second, if a change in the biomechanical skills elicits the emergence of a remarkable visuo-motor interaction, it would be quite unnatural to increase the body sway to the radial optic flow with age. The improvement of biomechanical functions such as muscle strength seems to contribute to cancelling out the body sways caused by visual stimuli; however, it has been reported that the visually guided body sway increases thorough the second half of the first year of life (Bertenthal et al., 1997; Bertenthal & Bai, 1989; but see Butterworth & Hicks, 1977). Based on these discussions, we infer that the development of the interaction between radial optic flow perception and the related motor actions is elicited by the functional development of particular neural bases concerning some multisensory processing, and that the neural development is promoted by the experience of ego motion, such as locomotion. In the following two sections, we will propose a developmental neural framework that potentially elicits the significant change in the visuo-motor interaction during that age range.
Neural bases of visual radial motion perception and ego motion perception
Previous studies have shown that the interactive process between the perception of radial optic flow and the perception of ego motion, such as heading or postural sway, develops around the second half of the first year of life. The visuo-motor interaction may be achieved by the particular neural mechanism that is involved in the integrative process between the visual optic flow and nonvisual information. In this section we will survey the neural studies related to these topics.
Although the functional maturation of the posterior parietal cortex seems to be related to the development of the cortical interaction between radial motion perception and various adaptive actions, no one has directly investigated this issue. However, several recent studies have investigated the development of the cortical activities for radial motion processing. The findings from recent studies will be useful for inferring the developmental relationship between visuo-motor interaction and the functions of the parietal cortex. Shirai et al. (2009) investigated the cortical activity in response to a radial motion pattern in 3- and 4-month-old infants and adults. They found that both the infants aged 4 months and the adults showed significant visual evoked potentials to radial expansion/contraction, but that the 3 month-olds did not. Moreover, they also found that the older infants and adults indicated significant asymmetry between the cortical responses to expansion and contraction stimuli: both the infant and adult brains showed greater responses to contraction than expansion (but see also Gilmore et al., 2007). More recently, Shirai, Imura, Birtles, Anker, Ichihara, Wattam-Bell, Atkinson, and Braddick (2009) retested the findings of Shirai et al. (2009) using a multi-channel high-density VEP recording system to examine the distribution of the cortical activity to a radial motion pattern. In this study, they almost replicated the findings of Shirai et al. (2009): the 4-month-olds and the adults indicated a contraction bias, but not the 3-month-olds. However, the cortical distributions of the contraction bias were completely different between the older infants and the adults. In the adult brain, the contraction bias was localized to the right parietal cortex (P6 and P8). In contrast, in the infant brain there was no significant localization of the contraction bias: the bias was spread across the occipito-parietal areas. This means that the particular function of the cortical radial motion processing, the contraction bias, was mediated by different neural mechanisms in the infants and the adults.
Because these psychophysiological studies conducted by Shirai and colleagues focused on only a limited aspect of radial motion cortical processing, such as the contraction bias, we must apply their findings carefully to construct a hypothesis about the development of radial motion cortical processing. However, the fact that the infants' and adults' cortical areas showed different specializations for the asymmetric radial expansion/contraction motion processing implies that infant and adult brains have different cortical mechanisms for radial motion processing. This leads to the possibility that the infant's ability to coordinate visual radial motion perception and the related adaptive actions, such as locomotion or controlling the posture, may be differentiated from that of the adults.
Developmental framework for the interaction between visual radial motion perception and adaptive actions
Summarizing the discussions of the previous sections, the main points of this review are as follows (see also Figure 1). First, fundamental radial motion perception rapidly develops at approximately 3–4 months of age, and then reaches a developmental plateau. The plateau lasts until at least 8 months of age (see the section “Early development of radial motion perception”). Although defensive motor responses are observed before the rapid development of radial motion perception, the defensive responses may be driven by the simpler physical properties typically contained in an expanding retinal image of an approaching object, but not by radial motion as such (see the section “Development of defensive motor responses to visual radial motion”). The interaction of radial motion perception and the related adaptive actions (locomotion or postural sway) first develops in the second half of the first year of life. This period almost corresponds to the developmental plateau for radial motion perception (see the section “Development of ego motion perception from visual radial motion”). The posterior parietal cortex is a candidate for the neural basis of the interaction between visual radial motion and ego motion. However, the infant neural mechanism for processing radial motion is potentially different from the adult one. Therefore, the mechanisms related to the interaction between the perception of visual radial motion and the perception of ego motion should also be distinct between infants and adults (see the section “Neural bases of visual radial motion perception and ego motion perception”).
Based on these discussions, here we propose a hypothetical developmental framework for the interaction between visual radial motion perception and the relevant adaptive actions (see also Figure 1). First, the basic ability to perceive radial motion emerges by 3–4 months old. The radial motion perception is relatively immature, but may be enough to control various rudimentary adaptive actions, such as creeping or postural control during sitting, at the earlier phases of the developmental plateau of radial motion perception (from 3–4 months to approximately 8 months old). Keeping radial motion perception in an immature state before acquiring skilled adaptive actions may be reasonable for the later development of visuo-motor interactions, because radial motion perception should be calibrated according to the developmental changes of the adaptive actions. For instance, the appropriate optic flow sensitivity for controlling locomotion may differ between creepers and walkers. In other words, radial motion perception changes from an immature and flexible state to a solid and mature state through the later developmental stages. At the earlier phases of the developmental plateau, the visuo-motor interactions are coordinated by the relatively broad areas of the dorsal pathway, including MST and VIP, unlike in adults. This earlier visuo-motor interaction would provide infants with experience of ego motion accompanied by the execution of independent adaptive actions. The ego motion experiences would aid infants in calibrating and optimizing their radial motion perception for controlling adaptive actions. This calibrated radial motion perception would also be used to refine the execution of adaptive actions. This interactive feedback loop between radial motion perception and ego motion perception would shape the specialization of the posterior parietal cortex (e.g. VIP) in order to integrate visual radial motion and related motor actions. Undergoing these visuo-motor interactions in infancy would enable the ability to control various adaptive actions that are related to radial motion perception to be approximated to the adult level.
This is only a rough hypothetical sketch of the early development of the interaction between radial motion perception and related adaptive actions. However, this hypothesis can be tested and modified by future empirical investigations. Particularly, it is important to examine whether there is any significant correlation between the development of radial motion perception and that of adaptive actions, such as locomotion or postural control, with a consideration for individual difference. There have been only a few studies examining the effect of individual difference in ability on motor actions related to radial motion perception in infancy. Our hypothesis suggests that radial motion perception is improved by the experience of ego motion accompanied by various adaptive actions, such as locomotion or postural sways. Because there are substantial individual differences in the development of such adaptive actions, longitudinal investigations of the development of radial motion perception that consider individual differences in adaptive actions would be helpful for understanding the relationship between radial motion perception and various adaptive actions. Additionally, infant brain-imaging studies focusing on the activity of parietal cortical areas such as VIP will reveal the critical changes in brain activity required for radial motion perception and related adaptive actions in infancy.