Sensitive periods in functional brain development: Problems and prospects

The articles in this special issue discuss and describe evidence from several sensory domains relating to sensitive periods during postnatal brain development. I begin this commentary with discussion of some of the major themes that emerge from the collection of articles. Then I focus on the implications of the evidence presented in this special issue for three different perspectives on mammalian functional brain development. I conclude that the available evidence is most consistent with an ‘‘interactive specialization’’ view of functional brain development in which the closing of sensitive periods are a consequence of achieving functional specialization within a brain region or cortical area. Several authors in the collection prefer to use terms such as ‘‘sensitive period’’ or ‘‘optimum period’’ (Werker & Tees, this issue) instead of the more traditional ‘‘critical period.’’ As noted in the historical introduction by Michel and Tyler (this issue), early ethologists such as Konrad Lorenz described particular age periods during which experience must occur for it to have an effect on subsequent development and behavior. In the best known example, imprinting, it was thought that young precocial birds had to have exposure to conspecifics within hours of hatching to develop adaptive filial attachments and appropriate later mate preferences. Later work with birdsong, cats, dogs, monkeys, and language development (in humans) led to the view that critical periods were major phenomena in brain and behavioral development (for review, see Michel & Tyler, this issue); however, more in-depth studies of some of these critical periods revealed that these periods were not as critical in terms of their timing and specificity as previously supposed. For example, the critical period for imprinting in domestic chicks was shown to be extendible in time in the absence of appropriate stimulation, and the learning to be reversible under certain circumstances (for review, see Bolhuis, 1991). Evidence accrued that the plasticity underlying imprinting was ‘‘self-terminating’’ (Bateson, 1987) in a way that can be simulated by computer neural-network models that incorporate features of the known architecture of the chick forebrain (discussed later). In other words, the closing of the critical period was a natural consequence of the learning processes itself. This line of evidence, along with several others (see Michel & Tyler, this issue), has led many researchers to prefer the term ‘‘sensitive period.’’ To further clarify that neither the onset nor the offset of the period is absolute and invariant, Werker and Tees (this issue) suggest the alternative term ‘‘optimum period.’’ However, I will adopt the more commonly used ‘‘sensitive period’’ (SP) for the rest of this commentary.