Developmental Narrowing

Although most theories of behavioral development either ignore developmental narrowing processes altogether or assign relatively little importance to them, a class of what might be referred to as “psychobiological theories of development”, do acknowledge their importance through the concept of behavioral canalization.

This concept was fi rst introduced into the psychological literature by Holt (1931) to account for the emergence of organized motor activity patterns out of the initially diffuse patterns seen during fetal development. According to Holt, the initially diffuse sensorimotor activity observed in early fetal life becomes cana-lized into organized motor patterns through the process of behavioral condition-ing. Later, Kuo (1976) broadened Holt’s limited concept of canalization by proposing that narrowing of behavioral potential was not merely the result of the individual’s history of reinforcement but that the individual’s entire develop-mental history, context, and experience play a crucial role in it. It was Gottlieb (1991), however, who provided the fi rst empirical proof of the importance of narrowing in early development in his studies of the development of species identifi cation in ducks. He showed that the development of mallard hatchlings’

socially affi liative responses toward their conspecifi cs is determined by exposure to their own embryonic vocalizations and, critically, that this experience helps to canalize their subsequent responsiveness. Specifi cally prior to hatching, embryos vocalize and Gottlieb showed that as they do so they actually learn some of the critical features of their species-specifi c call and in the process learn not to respond to the social signals of other species. In other words, experience with their own embryonic vocalizations narrows the embryos’ initially broadly tuned auditory sensitivity. Once this process is completed, the hatchlings end up being sensitive to the acoustic features of the calls of their own, but not to the calls of other, species.

Similar to psychobiological theories of behavioral development, dynamic systems theories of development (Lewis 2000; Thelen and Smith 1994) have called attention to the importance of regressive developmental processes in their account of the development of motor skills. These theories assume that the degrees of freedom that defi ne the critical parameters that control various motor skills are reduced during motor learning and development. For example, when infants are fi rst learning to walk, the many parts of the motor system are free to assemble into many functional patterns and are free to do so in many different ways (i.e., the degrees of freedom are many). As they begin to move and interact with the physical substrate on which they are trying to locomote, however, the

various subparts of their motor system begin to cooperate with one another and begin to assemble into stable and effi cient patterns of motor action. As they do so, the functionally useful patterns are selected from the initially many possible ones through a reduction in the degrees of freedom underlying the various sub-systems that participate in the control of locomotion. Once such patterns are selected, the system achieves a stable attractor state. In other words, effi cient walking emerges from a diffusely organized system that becomes canalized through experience into a more effi cient but, at the same time, less plastic system.

Despite the fact that developmental narrowing processes have to some extent been recognized in some developmental quarters, there has not been widespread recognition of the importance of such processes for understanding the eventual products of behavioral development. This situation has, however, been slowly changing as a new and steadily growing body of empirical evidence has begun to provide some convincing proof that a number of human perceptual functions undergo developmental narrowing in early life and that such narrowing is part-and-parcel of the eventual development of species-specifi c patterns of perceptual expertise. This body of evidence consists of fi ndings from studies of speech and face perception and shows that initially in development some perceptual abilities are broadly tuned and that as development proceeds these abilities become nar-rower in scope.

2.3.1 Developmental Narrowing in Speech Perception

The original studies that provided the fi rst evidence of developmental narrowing effects in human infants showed that initially present sensitivity to nonnative speech contrasts declines during the fi rst year of life as a function of experience.

Specifi cally, these studies have shown that young infants can perceive both native and nonnative phonetic contrasts but that by the time infants reach the end of the fi rst year of life they can no longer perceive nonnative contrasts. Some of the fi rst, though, indirect evidence of this kind of narrowing came from a study by Streeter (1976). Building on the earlier landmark fi ndings of Eimas and col-leagues (Eimas et al. 1971) showing that 1- and 4-month-old infants can perceive a phonologically relevant voice-onset-time contrast (i.e., the distinction between a /ba/ and a /pa/), Streeter showed that 2-month-old Kikuyu infants learning the Kikuyu language can discriminate the voice-onset-time contrast even though this contrast is not used in their native language. Consistent with Streeter’s fi ndings, Aslin et al. (1981) found that English-learning 6–12-month-old infants can dis-criminate a phonologically irrelevant voice-onset-time contrast as well as the phonologically relevant one identifi ed by Eimas et al.

It was Werker and Tees (1984), however, who provided the fi rst direct evi-dence of developmental narrowing. These investigators demonstrated that 6-to-8-month-old English-learning infants can discriminate nonnative consonants, such as the Hindi retrofl ex /Da/ and the dental /da/ and the Thompson glottalized velar /k’i/ versus the uvular /q’i/, but that 10-to-12 month-old infants do not

dis-criminate them. Based on these fi ndings, Werker and Tees concluded that the decline in responsiveness to nonnative phonetic contrasts is due to language-specifi c experience that provides infants with continuing experience with native consonant contrasts and none with the nonnative ones. Subsequent cross-linguis-tic consonant discrimination studies have provided additional evidence of this type of developmental narrowing. For example, Best et al. (1995) tested 6-to-8 month-old and 10-to-22 month-old English-learning American infants’ ability to discriminate native English and Nthlakampx velar vs. uvular ejectives as well as Zulu click sounds. They found that younger infants discriminated the English and Nthlakampx contrasts, older infants only discriminated the English contrasts, and that both age groups discriminated the Zulu click sounds. These authors concluded that the decline in response to Nthlakampx contrasts refl ected similar developmental narrowing effects found by Werker and Tees (1984) and that equivalent levels of responsiveness to the Zulu clicks was due to the failure of such clicks to map onto English phonology. Finally, studies have shown that developmental narrowing is not confi ned to consonants. Kuhl et al. (1992) have shown that responsiveness to nonnative vowels also narrows during infancy and that it actually occurs earlier (by 6 months of age) than in responsiveness to consonants.

2.3.2 Developmental Narrowing in Face Perception

Perceptual narrowing in human infants is not limited to the auditory modality.

Studies have shown that infant responsiveness to native and nonnative faces follows a pattern of decline similar to that found in speech perception studies.

These studies have found that human as well as monkey adults are better at rec-ognizing faces from their own species than from other species (Dufour et al. 2006;

Pascalis and Bachevalier 1998). Once again, the native-nonnative response dif-ference observed in adults turns out to be the result of experience-dependent narrowing processes that occur during infancy. Direct evidence of this fact was fi rst provided by Pascalis et al. (2002) who investigated discrimination of human and monkey faces in 6 and 9-month-old infants. These researchers found that 6-month-old infants discriminated human and monkey faces but that 9-6-month-old infants only discriminated human faces. In a subsequent study, Pascalis et al.

(2005) replicated these fi ndings and, like Kuhl et al. (2003) who showed that responsiveness to nonnative speech contrasts can be maintained into later infancy through exposure to nonnative speech, also demonstrated that the decline in nonnative discriminative ability can be reversed with continued experience. That is, Pascalis et al. demonstrated that infants who were exposed to monkey faces during the 3 month period between 6 and 9 months of age continued to exhibit the ability to discriminate monkey faces at 9 months of age.

Another line of evidence that provides further proof of the contribution of developmental narrowing processes comes from studies of the “other race effect”

(ORE). The ORE refl ects the fact that adults fi nd it more diffi cult to distinguish between the faces of people from other races than the faces of people from their

own race. This fi nding has been interpreted as refl ecting experience-dependent effects of greater exposure to individuals from one’s own race than to individuals from a different race. Developmental studies have shown that the roots of the ORE can be found quite early in infancy. For example, Sangrigoli and de Schonen (2004b) tested 3-month-old Caucasian infants’ preference for Caucasian and Asiatic faces following habituation to a single face from each race. They found that when infants were habituated to a Caucasian face they successfully recog-nized a different Caucasian face but that when they were habituated to an Asian face they did not recognize a novel Asian face. In addition, Sangrigoli and de Schonen (2004a) obtained evidence consistent with an experience-dependent interpretation of the ORE. When they gave Caucasian infants more extensive experience with other-race faces by familiarizing them with three different faces within each race the infants successfully detected differences in both Caucasian and Asiatic faces. More recently, Kelly et al. (2005) replicated the ORE in 3-month-old infants and, in addition, provided even more direct evidence of the experience-dependent nature of the ORE by showing that the ORE is absent in newborn infants. These authors concluded that selective exposure to the faces from one’s own ethnic group during the fi rst months of life leads to developmen-tal narrowing of visual preferences to the faces from one’s own ethnic group.

2.3.3 Developmental Narrowing in Face/Voice Integration

As is evident from the fi ndings of perceptual narrowing in the speech and face processing domains, experience plays an important role in refi ning processing in the auditory and visual modalities. It was also noted earlier that experience plays an important role in the development of multisensory perception in that greater experience with one’s own mother’s face and voice increases infants’ ability to integrate the two. Could it be that experience can also have the opposite effect on multisensory integration? Could it be that the lack of particular experiences can lead to a decline of an initially broadly tuned sensitivity to face-voice rela-tions? The perceptual narrowing effects found in the speech and face processing domains show clearly that we begin life with relatively broad sensitivity in each modality and suggest that perceptual narrowing in face-voice integration may take place as well. If that is the case then this would suggest that perceptual nar-rowing is a domain-general developmental process.

Recently, we investigated the possibility that perceptual narrowing also plays a role in the development of multisensory perception (Lewkowicz and Ghazanfar 2006). We did so by testing the hypothesis that the range of salient multisensory relations should gradually narrow in early development and that, as a result, infants’ ability to integrate nonnative faces and vocalization should be initially present early in infancy and then decline later in infancy. We put this hypothesis to empirical test by using a multisensory matching technique and stimuli from another primate species: the rhesus monkey. We presented to human infants pairs of movies in which the same macaque monkey’s face could be seen on side-by-side computer monitors mouthing a “coo” call on one monitor and a “grunt”

call on the other monitor. During the fi rst two trials, infants watched these movies in silence (the side of call presentation was switched on the second trial). During the next two trials we presented the movies again but this time accompanied by either the coo or the grunt vocalization. The coo call is longer than the grunt and, thus, given that the onset of the vocalization was synchronized with the onset of both visible calls, infants could make multisensory matches on the basis of synchrony and/or duration cues. To determine when during the fi rst year of life the decline in multisensory integration of nonnative faces and vocalizations might occur, we tested groups of 4, 6, 8, and 10 months of age. We found that 4- and 6-month old infants exhibited signifi cantly greater looking at the matching visible call in the presence of the matching call than in its absence but that the 8- and 10-month-old infants did not. These fi ndings are fully consistent with the previous fi ndings of a developmental decline in responsiveness to nonnative faces and speech and show for the fi rst time that perceptual narrowing of multisensory integration also occurs during early development and suggests that narrowing is a domain-general process.