We comment that covariances between brain divisions may be constraining or facilitating, depending on what is being selected, and that modern quantitative genetic methods provide the tools to discover and manipulate the genetic networks that give rise to the covariances described in the target article.

The human brain does not use more energy than the smaller brains of animals of comparable corporal weight. Uniquely, human functions localized largely in parts of the human brain that show greatest size increase over other animals may be mediated primarily by nonsynaptic neurotransmission, with reduced energy cost per kilogram of brain. This may affect the energetic constraints on evolution.

The opposition set up between co-ordinated and mosaic brain evolution distracts from the fact that the two go hand-in-hand. Here and elsewhere (Barton & Harvey 2000), I show that the patterns of co- ordinated evolutionary change among brain structures fit a mosaic evolution model. The concept of overarching developmental constraints is unnecessary and is not supported by the data.

We make three points. First, even if Finlay et al.'s proposed developmental mechanisms hold, there remains great scope for selection on specific brain structures. Second, the positive covariance among the size of brain structures provides far less support for the proposed developmental mechanisms than Finlay et al. acknowledge. Third, even if the proposed mechanisms are the primary size determinants for most brain structures, these structures should not be considered “spandrels.”

I argue that, while Finlay et al, are correct to suggest that there are developmental regularities (or constraints) acting on brain component evolution, they are incorrect to infer from this that a developmental explanation necessarily implies that structural changes preceded functional use. Developmental and functional (adaptationist) explanations are complementary, not alternatives.

It is suggested that a connection between neurogenesis and brain part size is unsurprising. It is argued that neurogenesis cannot, however, be the only factor contributing to brain size. Highly individual post-natal experience radically shapes individual brains, leading to dramatic increases in brain size. The role of comparatively coarse statistical techniques in addressing these subtle biological issues is questioned.

The article of Finlay et al. is an excellent example of identifying constraints in the development of the brain, and their implications on brain architecture in evolution. Here we further illustrate the importance of constraints by presenting a few examples of how a small number of biophysical mechanisms or even a single life history parameter can have an enormous impact on brain evolution.

Developmental constraints presumably had a major role in channeling evolution. In particular, developmental mechanisms may have coordinated the evolution of neocortex with that of other brain structures. However, the rules determining the differential expansion of different cortical territories remain to be determined as well as the adaptive role of cortical expansion versus that of the structures it is connected to. The high degree of developmental plasticity of neocortex was probably the key to its successful evolution.

Despite the ontogenetic allometric size effects that explain much of phyletic variation in brain components, the residuals of some structures indicates that mosaic brain evolution was an important factor in hominid evolution, and that reorganization of the hominid brain may have occurred as early as 3+ MY. Finlay et al.'s allometric technique masks residual variation around allometric trends, and the patterns of residuals relevant to species-specific departures from strict allometric trends.

Finlay et al. describe a method of examining brain evolution, but it has limits that may hinder extrapolation to all vertebrate taxa or the understanding of how brains work. For example, members of different orders have brain and behavioral organization that are fundamentally different. Future investigations should incorporate a phylogenetic approach and more attention to behavior to further test their conclusions.

Brain parts can scale independently of the whole brain, and an example is given to point out that the authors underestimate variation that can exist in brains of equal size.

Although many aspects of human cognition are likely to be “spandrels” passively affiliated with the primary impetus for hominid brain expansion during the Plio-Pleistocene, that expansion was most likely generated and maintained not by “housekeeping” functions but by improved capacities of reproductive success, especially survivorship.

One of the mammalian groups absent from the Finlay et al. study is cetaceans (dolphins, whales, and porpoises). Inclusion of cetaceans would be useful for assessing the generalizability of the authors' conclusions. Recent findings suggest dolphins may differ from the general pattern observed by Finlay et al. I encourage Finlay and her colleagues to include developmental neurobiological data on cetaceans, when available.

The mechanics of walking restructured the pelvis and narrowed the birth-canal that selected for delays in skeletal ossification. Prolonged phases of fetal maturation increased the mass and volume of the brain relative to adult body-size, as encephalization increased. Thus, bipedal- walking and episodic increases in hominine body size probably triggered selection for neonatal skeletal immaturity that led to encephalization.

Researchers studying primate brain allometry often focus on departures from allometry more than the allometric relationships themselves because only the former reveal what brain regions and behavioral-cognitive abilities were the focus of selection. Allometric departures for the human brain provide insights into hominid brain evolution and cast doubt on the suggestion that the large human cerebral cortex is a “spandrel.”

The existence of linked regularities in size among brain components across species is, by itself, not a strong argument against the importance of behavioral selection in brain evolution. A careful consideration of hominid brain evolution suggests that brain components can change their scaling relationships over time, and that behavioral selection was likely crucial. The best neuroanatomical index of a given behavioral ability can only be determined empirically, not through comparative analysis of brain anatomy alone.

Constraint has played a major role in brain evolution, but cannot tell the whole story. In primates, adaptive specialization is suggested by the existence of a covarying visual system, and may explain some residual variation in the constraint model. Adaptation may also appear at the microstructural level and in the globally integrated system of brain, body, life history and behavior.

Allometric analyses suggest that there are some developmental constraints on brain evolution. However, when one compares animals of similar body size, these constraints do not appear to be very tight. Moreover, the constraints often differ between taxonomic groups. Therefore, one may ask not only what causes developmental constraints but also how (and why) these constraints might be altered (or circumvented) during the course of evolution.

Brains vary in characters other than size. We should consider whether Finlay et al.'s argument, that developmental shifts responding to selection for change in one area yield correlated changes across the brain, must be extended from size differences to other neural characters responsible for the circuitry or physiological differences distinguishing vertebrate brains.

First, we clarify the central nature of our argument: our attempt is to apportion variation in brain size between developmental constraint, system-specific change, and “mosaic” change, underlining the unexpectedly large role of developmental constraint, but making no case for exclusivity. We consider the special cases of unusual hypertrophy of single structures in single species, regressive nervous systems, and the unusually variable cerebellum raised by the commentators. We defend the description of the cortex (or any developmentally-constrained structure) as a potential spandrel, and weigh the implications of the spandrel concept for the course of human evolution. The empirical and statistical objections raised in the commentary of Barton are discussed at length. Finally, we catalogue and comment on the suggestions of new ways to study brain evolution, and new aspects of brain evolution to study.