Findings from the current study highlight the utility of the NIRS method for studying the neurobiological basis for early perceptual processing. We found reliable hemodynamic changes in the left temporal cortex of infants in the second half of the first year of life in response to audiovisual stimuli relative to visual-only stimuli, as well as relative to changes across stimulus conditions in homologous regions of the right cortex in the same infants. These results are consistent with a long history of behavioral and neurophysiological research with adult humans demonstrating that the left temporal cortex is a primary area for language processing due to its structural and functional characteristics (see Hutsler & Galuske, 2003
, for a review). The results of the current study, along with those of Bortfeld et al. (2007)
, are also consistent with recent reports that very young infants demonstrate significant hemodynamic changes in the left temporal cortex in response to speech-specific auditory stimuli (Dehaene-Lambertz et al., 2002
; Peña et al., 2003
). These current results thus suggest that NIRS can be used to assess infants’ neurophysiological responses in a variety of processing tasks.
Research will need to build on and expand the current findings. For example, it will be important to investigate the extent to which these results replicate in infants of the same age when no visual stimuli accompany the auditory stimuli, a process that is ongoing in our lab currently. Specific to language development, additional research will need to investigate the specific areas within the bilateral temporal cortex, as well as other cortical regions, that respond selectively to specific features of the auditory signal. Finally, it will be important to determine how different experimental manipulations influence infants’ processing biases. For example, contrasts in classes of structured auditory stimuli (e.g., native vs. nonnative speech; speech vs. music) may reveal important changes across the first year of life attributable to environmental experience. Indeed, because the hemispheric lateralization for speech processing observed in infants is not as strong as it is in adults (Holland et al., 2001
), it would seem that consolidation occurs during the course of learning language and that researchers should see changes across the first year, as sensitivity to language-specific features increases. Such findings would bolster our understanding of the neural basis for the variety of behavioral findings from infants, many of which demonstrate increasingly sophisticated perceptual capacities that are quickly modified by the linguistic environment (e.g., Jusczyk, Luce, & Charles-Luce, 1994
; Jusczyk, 1997
). Likewise, and consistent with recent findings using NIRS (Homae et al., 2007
), it is possible that infants’ right temporal cortex, a region that is important to prosodic perception in adults (Baum & Pell, 1999
; Ross, Thompson, & Yenkosky, 1997
; Friederici & Alter, 2004
; Grimshaw, Kwasny, Covell, & Johnson, 2003
), would respond more robustly to particular auditory stimuli than others given a paradigm that highlights that aspect of the signal. Given that infants rely differentially on cues unique to their native language to segment the incoming auditory stream into linguistic units, this type of approach would allow researchers to map the functional development of hemispheric specialization and determine the role it plays in tuning auditory processing during the first year of life.
Another direction for future research will be to combine multiple techniques, both behavioral and neurophysiological, to determine if known developmental sequences reflect biological changes in the neurophysiology underlying perceptual processing. In the current study, looking times were tracked only to verify that infants were attending to the stimuli being presented during any particular trial. The utility of looking-time differences has been made clear by behavioral researchers, who operationalize them as indicative of differential sensitivity (whether characterized as preference or recognition; e.g., Bortfeld et al., 2005
; Singh, Nestor, & Bortfeld, 2008
). Therefore, the development of experimental paradigms that combine measures of differential looking behavior with measures of neurophysiological response will do much to bridge the gap between findings in the behavioral and the neurobiological literatures on human infant development.
Limitations of the current study can be discussed in terms of the particular experimental design employed and the physiological basis for the NIRS technique. The infant population requires certain concessions and creative adaptations of experimental design in order to accommodate infants’ general inattention and their likelihood of moving during experimental trials. In the current study, we used stimulus alternating blocked trials. This design was effective at holding infants’ attention throughout the experiment, while NIRS was able to characterize the corresponding dynamic perceptual processing. Nonetheless, this design necessarily limited our ability to draw conclusions based entirely on the auditory component of the stimuli. A paradigm that would allow us to test infants with auditory stimuli alone would employ probabilistically determined event-related trials (Friston et al., 1998
; Friston, Zarahn, Josephs, Henson, & Dale, 1999
), in which responses to individual stimuli are modeled as brief bursts of neural activity that are convolved with a hemodynamic response function. This design approach likewise would address the second limitation in this experiment, the relatively sluggish temporal resolution of hemodynamic-based measures in general. Techniques such as evoked response potentials (ERP) and electroencephalogram (EEG) can be used to map electrical signals based on action potentials occurring perpendicular to the surface of the scalp and are extremely accurate temporally, but their spatial acuity is quite limited relative to NIRS. The use of an event-related design targeting specific cortical regions using NIRS would provide a relatively sensitive measure of the temporal nature of cortical responses in those relatively specific cortical regions.
Finally, the temporal specificity of processing provided by ERP complements the localization specificity provided by NIRS. Notable advances have been made in recent years in combining a variety of infant behavioral measures (e.g., bimodal preference paradigms; combinations of preferential listening and eye-tracking). Likewise, simultaneous tracking of multiple neurophysiological measures stands to substantially advance our understanding of the processing that underlies infants’ perceptual experience. For example, much progress has been made in the application of ERP to help predict a range of language outcome measures (e.g., Espy, Molfese, Molfese, & Modglin, 2004
; Mills et al., 2004
; Molfese et al., 2006
). Future research that succeeds in combining these two approaches (NIRS and ERP) in the assessment of processing abilities in infants and young children will maximize the data obtainable from these populations. Developmental researchers will gain much from the current push to combine these and other technologies in the measurement of adult neurophysiology (e.g., Roche-Labarbe, Wallois, Ponchel, Kongolo, & Grebe, 2007
The findings reported here continue our extension of NIRS to use with older infants. In particular, they highlight the utility of this technology as a tool for tracking hemodynamic activity in infants, even when they are actively processing relatively complex material. A variety of questions remain, however, about the region specific acoustic representation that underlies humans’ ability to identify subtle distinctions in both spectral and temporal components of a variety of sounds (in both speech and non-speech). It is becoming increasingly clear that distinct subareas of bilateral human auditory cortex have different response properties (e.g., Seifritz et al., 2002
) and that, although ongoing research on adult humans (e.g., Zatorre, 2003
) is expanding our understanding of the nature of these response properties, relatively little is known about auditory processing characteristics specific to infants. Our results are an initial demonstration that NIRS is sufficiently sensitive to assess the neural basis of perceptual processing in awake and behaving infants. Although much additional research is needed to examine a variety of important issues underlying this result, the current study demonstrates the wealth of information available through thoughtful application of this technology to the study of infant development.