A quiet revolution has been taking place in the thinking about the development of the nervous system and behaviour. Only 30 years ago E.O. Wilson (Reference Wilson1976) felt able to write that development was a “module” that could be decoupled when correlating differences in genomes with differences in the adult phenotypes. Few would agree with that view nowadays. The objective to discover the relative importance of genes and the environment at the population level has a distinctly old-fashioned feel about it now, based as the approach is on the assumption that sources of variation add together. “Systems analysis” is the phrase on everybody's lips, although this particular fashion has rarely been translated into an understanding of process. What we need is a coherent way of rendering usable the intuition about the interplay between the developing organism and its environment. The two volumes, of which I shall focus on the first (Mareschal et al. Reference Mareschal, Johnson, Sirois, Spratling, Thomas and Westermann2007a), represent a serious attempt to provide such a road map.

Setting out their stall in the first volume, Mareschal et al. argue that the challenge is to incorporate information from the brain when explaining the development of behaviour without lapsing into reductionism. Quite right. As Robert Hinde put it many times, we have to cross and recross the boundaries between the different levels of analysis (Bateson Reference Bateson1991). The case Mareschal et al. make for some degree of formalization is that an explicit working model brings with it mental discipline and may expose easily missed weaknesses in a verbal argument. Such models can also serve several other valuable functions. They can show how we are easily misled by the dynamics of development into supposing that the processes are so complicated that they are beyond comprehension. From the point of future empirical research, they can suggest profitable new lines of enquiry. Finally, they bear directly on the general arguments about the interplay between internal and external factors. As far as linking levels of analysis is concerned, the models point to the parameters that are features of the organism as opposed to ones that might be manipulated externally by experiment. One example is the constant affecting the rate at which plastic changes occur. These hypothetical features of the organism, if real, must have some correspondence to underlying neural mechanisms.

Mareschal et al. emphasize the active role of the individual in its own development and pick out three principles which they regard as crucial: cooperation between different systems, competition, and chronotopy (by which they mean relative temporal ordering of expression). Competition is defined as the process by which one of several components wins out and becomes the relevant functional unit. Despite the high level of abstraction, I believe that they are generally correct in much of what they argue for, although I think competition has an additional side to it. By the process of competitive exclusion, experience of one kind can shut out the effects of subsequent experiences of another kind – very important in terminating sensitive periods in development.

Other important features of development are also left out of the authors' analysis. First, many manifestations of the young individual's brain and behaviour represent specializations to the particular ecological conditions of that stage of development; suckling by young mammals is one well-analysed case. Second, while I applaud the authors' emphasis on the active role of the individual, they do not draw attention to the scaffolding-like character of many instances of such behaviour; as in the assembly of a tall building, the scaffolding can be dismantled once the structure is in place. Third, the conditional responses to local circumstances mean that, viewed globally, phenotypic characteristics of adults can be extremely variable, but in each case well adapted to the particular conditions of the local environment –unless the individual has been “misinformed” by the events that initially triggered its developmental trajectory in early life (Bateson et al. Reference Bateson, Barker, Clutton-Brock, Deb, D'Udine, Foley, Gluckman, Godfrey, Kirkwood, Lahr, McNamara, Metcalfe, Monaghan, Spencer and Sultan2004). Finally, despite all their plasticity, many aspects of fully developed individuals are highly robust and their development across the population is seemingly impervious to great differences in experience of individuals. We need theories of how and why.

The authors are rather more impressed with their own sense of originality than is really justified. I suspect that I will not be alone in being irritated by some of their claims that their approach is so novel. Many distinguished neuroscientists, behavioural biologists, and developmental psychobiologists who have made major contributions to the understanding of development have seemingly been air-brushed out of the picture. So for example, no references are given to the work of Andrew, Blakemore, Fentress, Gray, Griffiths, Hinde, Johnston, Konishi, Lehrman, Marler, Michel, Moore, Nottebohm, Rosenblatt, Schneirla, Oyama, Trevarthen, Rogers, Rose, and many other important figures who have shaped understanding of behavioural development. When it comes to behavioural imprinting, Gabriel Horn is mentioned but nothing is stated about his long-term collaboration with me or about the neural net model we developed together (Bateson & Horn Reference Bateson and Horn1994). This model was developed with precisely the same objectives as those stated by the authors of the first volume, namely aiming to provide an account that was plausible at both the behavioural and neural levels. We set out many important implications for development. A general characteristic of our three-layered, unsupervised model was that it exhibited a well-known feature of development, simulating the competitive exclusion process that generates a descriptive sensitive period. While tending to settle into familiar habits, it was also able to build with increasing elaboration on the basis of previous perceptual experience. It was able to classify together different views of the same object and simulated well the perceptual learning process by which low stimulus value cues become more salient as a result of being paired with high stimulus value cues.

The authors argue admirably for a multidisciplined approach to the study of brain and behavioural development. It is pity, therefore, that they have seemed to settle into a silo of their own making and have not drawn more explicitly on many of the important conceptual and empirical advances in subjects close to their own. It shows how much more dialogue is necessary if the cooperation needed for the enterprise in which they are engaged is to proceed satisfactorily in the future.