Neuroconstructivism (Mareschal et al. Reference Mareschal, Johnson, Sirois, Spratling, Thomas and Westermann2007a) is an ambitious, brave, pioneering attempt to integrate system neuroscience and cognition across their current boundaries, within a developmental perspective. Computational modeling might be the best tool for such a complex task. This project is what we need if neuroscience has to have any future at all. I shall certainly go back to this book in the coming years, to refine my understanding of some of the rich, theoretical framework it explores.

I had some difficulties in grasping whether constructivism applies to the way cognition emerges from the brain (as in the subtitle), to how the brain and cognition codevelop, or to both. From a developmental point of view, I shall argue about the second interpretation. Let us go to facts. Accumulated evidence overwhelmingly demonstrates the generalized occurrence of exuberance and regression in brain development. Adult neural (and, I will argue, cognitive structures) emerge by selection from a richer (but constrained) juvenile stock. This applies not only to neuronal overproduction and death (as mentioned in ch. 2) but also to the generation of long axons, axonal branches, synapses, dendritic branches, number of spines, neurotransmitter receptors and/or expression, and so on. (reviewed in Innocenti & Price, Reference Innocenti and Price2005). In cognitive development the establishment of phonemic boundaries early in the postnatal period (Kuhl & Melzoff Reference Kuhl and Melzoff1996), and the evidence of a progressively more restricted activation of neural structures during the acquisition of skills, documented by imaging and electrophysiological data are clear examples of selectionism, as well as, I believe, Johnson's concept of “interactive specialization” (Mareschal 2007a).

The authors make only weak attempts (p. 7) to resolve the apparent opposition between selectionism and constructivism. I feel this dyad lies at the core of understanding how genes and environment (via neural activity) interact in evolution and in development. The opposition between selectionism and constructivism can be overcome by precise reference to the time-frames considered. In evolution, the power of selection is unchallenged and two-sided. One side is the selection of the genetic networks responsible for the harmonious emergence of brains/bodies and cognitions/behaviors most appropriate to the ecological constrains of a given environment (“context” in the book). The second is the selection of permissive developmental algorythms, in particular development by exuberance and regression of neurons, axons and so forth, mentioned above, which allows crossed adjustments of brain structures (e.g., Kingsbury et al. Reference Kingsbury, Lettman and Finlay2002), bodies, cognitions, and behaviors, and which actually allows evolution itself (Innocenti Reference Innocenti and Kaas2006). Now consider the developmental time-frame. Evolutionarily selected neural and/or cognitive structures are validated by experience in a context-dependent manner. Competition, cooperation, and chronotopy, which the authors identified as general developmental mechanisms, become relevant. Active exploration (experimenting or proactivity) is the expression of the evolutionary heritage, which, at the cognitive level might well consist of “partial representations” to be tested for their relevance in the real world. Only part of the structures selected in evolution survives this second wave of selection. Most important, they will not be tested and selected simultaneously, thus providing the framework for construction by sequential selections, a scenario we proposed for the development of axonal arbors (Zufferey et al. Reference Zufferey, Jin, Nakamura, Tettoni and Innocenti1999).

I loved the concept of partial representations proposed by authors, although representations seem difficult to define rigorously (pp. 4 and 5) and to test experimentally (pp. 221–22). Several questions spring to mind, for example.: How complex is a representation? Is it reducible to elemental parts? How far? In any case, partial representations seem to be economical in terms of neural resources and are open to completion by developmental experience and/or learning, in adult life.

In conclusion, I feel that one should accept the overwhelming evidence that environment and activity, in development are sculpting-out brain, body, cognition and behavior by selection, within the boundaries established, for each species, by evolution. Some further elaboration of the evolutionary heritage may occur based on cycles of generation, testing and validation/rejection of elemental structures were they axonal branches, synapses or representations. I would prefer to call the whole process selectionistic neurocostrutivism. The opposition then is not so much between selectionism and constructivism as between permissive and instructive morphogenetic factors, including experience. I have struggled with some of the concepts before, although not to my full satisfaction (Zufferey et al. Reference Zufferey, Jin, Nakamura, Tettoni and Innocenti1999).

One crucial question is, what happens when “context,” in development, fails to validate the neural and/or cognitive structures that evolution has selected. This is equivalent to asking whether the information provided by environment is permissive or instructive. Many examples of abnormal development are specified in the book all of which illustrate genetic abnormalities (Table 11.1, p. 164). Examples of abnormal experience in early development due purely to environmental factors, such as sensory or social deprivation (reviewed in Innocenti Reference Innocenti, Von Hofsten and Rosander2007), toxic insults (e.g., fetal alcohol syndrome), thyroid disturbances (Berbel et al. Reference Berbel, Obrego, Bernal, Escobar del Rey and Morreale de Escobar2007), and others, may be more informative. Some of them illustrate the response of the developing organism to the abnormal, and therefore potentially instructive, environmental conditions. The evidence for an instructive role of development is slim. Often the outcomes are nothing more than arrested development with no adaptive consequences whatsoever, providing strong evidence that development is rather tightly constrained. Thus, if the environmental conditions which drive selection in evolution are not met, the evolutionary heritage is deleted. This recalls Wiesel's statement that: “Innate mechanisms endow the visual system with highly specific connections but visual experience early in life is necessary for their maintenance and full development” (Wiesel Reference Wiesel1982).

This book is a very stimulating achievement. Its organization in Part 1, Foundations, explicating the main principles and the reasons for attempting the great synthesis of system neuroscience with cognition and development; the Case Studies section demonstrating how the principles can be applied; and the final section, Conclusions, specifying at least some of the open issues and the future directions is excellent. The “Boxes” provide essential information for readers not familiar with the territories explored and are particularly useful for a broad readership. The book provides a solid attempt to establishing a new field of intellectual and experimental investigations which will hopefully grow in the coming years.