Philosophers of science point out that good models should suggest experiments, allow predictions about the consequences of interventions or manipulations, and prompt new questions or directions for further research (e.g., Keller Reference Keller2002). In this light, Neuroconstuctivism is a significant step forward in the use of modeling within developmental science. Particularly noteworthy, the authors have provided a conceptual framework that takes seriously the notion that the minimal unit of analysis necessary for understanding behavior and cognition is the “brain in a body in an environment” system (Chiel & Beer Reference Chiel and Beer1997; Nelson Reference Nelson2007). Although the argument against the usefulness of radical reductionism has been made by a number of developmentalists in recent years, it is certainly the case that many neuroscientists and psychologists continue to speak as if understanding the brain will yield a full understanding of behavioral and cognitive development. Neuroconstructivism provides a cogent argument against this reductionistic view and outlines a well articulated example of how to integrate research from cognitive studies, computational modeling, neuroimaging, and developmental psychology to model some of the complexities and multidimensional dynamics of the developmental process. I am in agreement with the spirit of this integrative effort, particularly its emphasis on context dependence, multilevel isomorphism, and developmental trajectory, and its acknowledgment that explanations of cognitive development must be consistent with our knowledge of developmental and evolutionary biology (Lickliter & Honeycutt Reference Lickliter and Honeycutt2003). However, I have several reservations about the scope of the present version of the neuroconstructive approach. These include (1) that the authors have downplayed the nonlinear causality often at play in the developmental process, and (2) have effectively ignored the importance of value and meaning in the formation and maintenance of cognitive representations.

It is surprising that a framework designed to address the development of cognitive representations, based on the assumption that “cognition is essentially a process involving the manipulation of information” (p. 208), effectively sidesteps the topic of meaning and the issue of subjective experience. Value and meaning must be addressed in any theory of cognitive development; information takes form only when it is personally meaningful. As recently reviewed by Tucker (Reference Tucker2007), the formation of cognitive representations is achieved by the same limbic networks that provide the motivational control of behavior. In other words, the expectancies that guide cognition are personal and motivated. Neuroconsructivism does not adequately address the importance of meaning or the related issue of subjectivity, despite its emphasis on the “proactive” nature of development. This is an obvious shortcoming of the current framework and a significant challenge to computational modeling. In this light, the omission of reference to the work of Walter Freeman, which provides an important foundation for thinking about and modeling the links between meaning and representation, is surprising. As Freeman (Reference Freeman2001) has pointed out, meaning derives from the entire brain and body, with the history of the organism built into muscles, endocrine glands, and neural connections. The chapter by Sporns (Reference Sporns, Mareschal, Sirois, Westermann and Johnson2007) that appears in the companion volume does provide an initial step in addressing the psychobiology of meaning by emphasizing the importance of value systems in neuromodulation and learning. The scope and application of the neuroconstructive approach would benefit from additional efforts focused on how to incorporate the role of value and meaning into its models of cognitive development.

With regard to the larger issue of modeling the developmental process, while development can sometimes be viewed as a chain of cause/effect relationships (a common assumption in computational simulations), these chains often turn out to be loops, where cause becomes consequence and consequence becomes cause. Indeed, this nonlinear and distributed nature of causality is a fundamental feature of the developmental process (Gottlieb Reference Gottlieb1997; Gottlieb & Halpern Reference Gottlieb and Halpern2002) and the authors have not gone far enough in coming to terms with how to address this aspect of developmental complexity. Although Neuroconstuctivism does make a strong case for refocusing our attention from making simple causal attributions (to the genes or to the environment, for example), it does not provide a clear road map for how to model the varied systematic processes and interactions that take place within and between levels (i.e., genes, neurons, hormones, social interaction, culture) of the human developmental system (Johnston & Edwards Reference Johnston and Edwards2002). Recent evidence from the rapidly growing field of epigenetics demonstrating the intricate regulatory networks of genes, their products, and the features of the organism's internal and external environment highlights the mind-boggling combinatorial complexity of the bidirectional traffic inherent in the process of development.

In sum, I found the authors overly optimistic as regards how far connectionist models of development will get us. Early in the text, the authors state their aim in Neuroconstructivism is to “explain development.” This is an ambitious goal, to say the least, given that the process of development is a dizzyingly complex multidetermined phenomenon and the number of variables, interactions, and contingencies involved in the journey from fertilized egg to functional adult may well put its full understanding beyond human comprehension. This is not due to a lack of trying – thousands of scientists have worked on or are working on various aspects of this challenge. Keller (Reference Keller2002) has recently argued that there are likely some phenomena in the natural world that extend beyond our grasp, too complex to fit neatly into any of our models, theories, or explanations. The process of development may well turn out to be one of these phenomena. That said, it is also the case that a better understanding of the ways and means of development can make a significant contribution to how we understand nature and ultimately, how we understand ourselves. Despite its shortcomings, Neuroconstuctivism is a step forward on this quest. While it does not succeed in explaining development, it does succeed in providing an integrative model of the development of cognition that will likely suggest experiments, allow predictions about the consequences of interventions, and prompt new questions for research. That seems plenty to expect of this or any other functional model of development.