Mammalian encephalization has received considerable scientific attention, probably because the class contains two of the most encephalized orders, primates and cetaceans, and because we humans number among the 5000 or so mammal species (see Encephalization: Comparative Studies of Brain Size and Structure Volume in Mammals). Mammals thus provide a valuable case study for understanding the selection pressures favoring evolutionary changes in brain size, with many hypotheses regarding brain evolution originally applied to mammals and relatively large databases of whole brain and brain component volumes available (for extensive discussion, see the articles on brain evolution in mammals, various mammalian orders, and humans, this volume, e.g., Primate Brain Evolution in Phylogenetic Context, The Evolution of Hemispheric Specializations of the Human Brain, The Evolution of Human Brain and Body Growth Patterns, Mosaic Evolution of Brain Structure in Mammals, Encephalization: Comparative Studies of Brain Size and Structure Volume in Mammals, Evolution of the Cerebellum, Evolution of the Hippocampus). This has made possible large-scale comparative studies. Moreover, experimental data have been combined with comparative studies of brain evolution, and breeding experiments have also addressed the evolution of larger brains.
Mammalian encephalization, like encephalization in all animals, is generally assumed to result from the selective benefits of enhanced cognitive, perceptual, or motor abilities, with most research focusing on cognition (Macphail, 1982; Barton, 1999). Evidence regarding the assumed link between brain volume and cognitive capacity has come from two sources: correlations of behavioral demonstrations of cognitive ability and brain size, and from associations between enlarged brains and lifestyles thought to require increased cognitive demands. The behavioral data provide a more direct test of the assumption that cognitive capacity and brain size are linked, and so are key in this respect (Macphail, 1982; Deaner et al., 2000). Several lines of evidence in mammals have pointed to a correlation between a species' relative brain volume and its cognitive capacity. First, laboratory learning data collated from a variety of sources has been shown to correlate with relative whole brain size, although the number of species tested is small (Riddell and Corl, 1977). Second, various measures assumed to indicate general cognitive capacities correlate with relative neo-cortex size: the frequency of reports of tactical deception, innovation, social learning, and tool use all correlate with relative neocortex volume in primates (Reader and Laland, 2002; Byrne and Corp, 2004). These latter measures have the advantage of covering a larger number of species. Third, it has been proposed that the ability to perform apparently complex cognitive acts such as imitation and understanding the intentions of others are associated with brain enlargement (Byrne, 1992). For example, among mammals, the consensus view is that true imitation has only been experimentally demonstrated in large-brained apes and dolphins, while the smaller-brained mammals tested present equivocal evidence of imitation (Mitchell et al., 1999; Caldwell and Whiten, 2002). However, tests of complex cognition have been the subject of much controversy (e.g., Tomasello and Call, 1997). Moreover, because research has tended to focus on only a few, typically large-brained mammal species, it is difficult to know the true phylogenetic distribution of such traits and to conduct proper comparative analyses.
Problems with comparative studies can be solved by experimental studies of evolution. Studies on rhesus macaques and mice have demonstrated that brain size is heritable (Jensen, 1979; Atchley et al., 1984; Cheverud et al., 1990; Markina et al., 2001), which indicates that it is open to modification by selection. Moreover, selection experiments in mice have bred large- and small-brained lines and have found differences between the lines in performance on learning tasks (Markina et al., 2001 Interpretation of these findings is controversial, not least because learning ability is inferred from performance and a number of other behavioral changes are observed in the selection lines, such as changes in anxiety or exploratory behavior (Jensen, 1979; Johnston, 1982; Markina et al., 2001). A further problem is that selection on learning performance has tended to result in selection for task-specific abilities (Jensen, 1979). Moreover, the critical test of whether selection on learning capacities can lead to evolutionary changes in brain size would be to select for cognitive ability and to examine the effect on brain size (Johnston, 1982). As far as we are aware, such experiments have not been done, although one study reports a decrease in size of a hippocampus area in mice lines exposed to environmental stressors (such as natural predators; Poletaeva et al., 2001).
The vast majority of work on encephalization has focused on identifying the selection pressures associated with brain enlargement. Hypotheses have tended to fall into two camps, social and ecological explanations for brain enlargement. Social (or Machiavellian) intelligence hypotheses argue that enlarged brains evolved as an adaptation to living in large, complex social groups (Jolly, 1966; Humphrey, 1976; Byrne and Whiten, 1988; Flinn, 1997; Whiten and Byrne, 1997). Byrne and Whiten (1997) distinguish the term ''narrow Machiavellianism,'' the idea that it is selection for strategies of social manipulation or deception that has driven brain evolution, from their own broader use of the term ''Machiavellian intelligence,'' which includes all forms of social intelligence such as social learning. Ecological explanations for the evolution of large brains are also common and include the extractive foraging (Parker and Gibson, 1977) and cognitive/spatiotemporal mapping hypotheses (Milton, 1988; Deaner et al., 2000). Unpredictability and patchiness of resources are often cited as key ecological factors favoring brain enlargement (Eisenberg and Wilson, 1978). For example, fruit is likely to be distributed more patchily in space and time than are leaves, and so a fruiteating diet may be expected to have more cognitive demands than a leaf-eating one. Technical intelligence hypotheses that argue that technology or technical skills drove brain evolution can be considered with the ecological intelligence hypotheses, because they tend to focus on foraging or antipredator tool use (Passingham, 1982; Byrne, 1997a, 1997b).
What evidence is there in support of these hypotheses? Relative neocortex or brain size is positively correlated with social group size in nonhuman primates, carnivores, cetaceans, bats, and some insectivores, consistent with social intelligence hypotheses (Worthy and Hickie, 1986; Sawaguchi and Kudo, 1990; Dunbar, 1992; Marino, 1996; Barton, 1999; but see Connor et al., 1998; van Schaik and Deaner, 2003 on cetaceans). In ungulates, social species tend to have larger relative brain and neocortex sizes than do nonsocial species, (Schultz and Dunbar, 2006). Barton (1993) finds a correlation between group size and neocortex size in haplorhine primates, but not strepsirhines, which may indicate that group living favored brain size evolution among haplorhines only. Adding support to Machiavellian intelligence hypotheses, the frequency of reports of deceptive behavior has been found to correlate with relative neocortex size (Byrne and Corp, 2004). The findings that human and other primates tend to have superior abilities in problems involving social knowledge versus those involving nonsocial knowledge has also been taken as evidence that cognitive abilities developed as a response to social pressures (Cheney and Seyfarth, 1988; see also Baron-Cohen et al., 1999).
In support of ecological intelligence ideas, diets presumed to require increased cognitive demands have been shown to correlate with relative brain volume in several mammalian groups. For example, primate relative whole brain size and neocortex size correlate with frugivory (Barton, 1999) and diet breadth and relative neocortex volume are correlated in African anthropoid primates (Reader and MacDonald, 2003). Similar associations with diet, albeit not based on independent contrasts, have been found in bats (aerial insectivores have smaller brains for their body size than frugivores and nec-tarivores, with piscivores, foliage gleaners, carnivores, and sanguivores falling between the two extremes; Eisenberg and Wilson, 1978), small mammals (in rodents and lagomorphs, folivores and insectivores have small brains relative to their body mass and body length compared with those of other dietary categories; Harvey et al., 1980), and the Carnivora (carnivorous and omnivorous species have larger relative brains, though not significantly so, than insectivorous carnivora; Gittleman, 1986). In a phylogenetically controlled study of 59 bat species, Ratcliffe et al. (2006) have recently reported a larger relative brain size in species that use a flexible combination of gleaning and hawking techniques, compared to those that are specialized on either of the hunting modes. Home range correlates with relative brain size in primates, in support of spatiotemporal mapping ideas (Clutton-Brock and Harvey, 1980; Deaner et al., 2000). Consistent with technical intelligence hypotheses, tool use frequencies correlate with relative neocortex volume in primates (Reader and Laland, 2002; Lefebvre et al., 2004). However, the fact that few primates appear to make regular use of tools in the wild questions the idea that technical intelligence was a driving force for primate brain enlargement (van Schaik et al., 1999).
A number of other lifestyle correlates of relative brain size have been described. For example, in didelphid marsupials, preference for arboreal activity is associated with relative brain volume (Eisenberg and Wilson, 1981). Cerebellum volume has also been linked to locomotion mode in primates, bats, and cetaceans (Stephan and Pirlot, 1970; Rilling and Insel, 1998; Marino et al., 2000). Females of carnivore species where the females provide the sole parental care have larger brains than those of biparental or communal species (Gittleman, 1994). Neocortex volume has also been linked to sexual selection, being correlated with mating competition in frugivorous primates (Sawaguchi, 1997).
Potential costs and constraints of mammalian encephalization have received less attention, though there is support for the idea that brain enlargement carries metabolic and developmental costs, with brain size negatively correlated with litter size in marsupials (Eisenberg and Wilson, 1981), and positively correlated with the age of sexual maturity and life span, but not gestation length, in primates (Barton, 1999; Allman et al., 1993; Kaplan and Robson, 2002; van Schaik and Deaner, 2003). In odontocetes, relative brain size is also associated with relative time to sexual maturity and life span. Within life span, however, length of the adult period is more closely correlated with relative brain size than is length of the juvenile period, suggesting that the temporal costs of delayed maturity might be compensated by a longer time as a reproducing adult (Lefebvre et al., 2006). Primate brain size is not correlated with basal metabolic rate, and comparative analysis indicates that improved diet quality, by allowing reduction in gut mass relative to body size, is one possible mechanism allowing the energetic constraints on the evolution of the meta-bolically expensive large brain to be lifted (Aiello and Wheeler, 1995; Barton, 1999; Fish and Lockwood, 2003).
Given that there is support for a number of hypotheses regarding mammalian encephalization, can any consensus be formed? The findings described above are consistent with the idea that several selective pressures are responsible for the evolution of encephalization. An alternative view is that one factor is driving brain evolution, but that the cognitive abilities afforded by a large brain are applied to other domains. Determining causation is difficult since most comparative studies are based on correlational evidence. Moreover, the divisions between social and ecological intelligence may be fuzzy and social and ecological demands on cognition may evolve together, making it difficult to consider social and ecological intelligence hypotheses as alternatives (Barton, 1999; Deaner et al., 2000; Reader and Laland, 2002; Seyfarth and Cheney, 2002). It is also difficult to separate perception from cognitive demands. For example, primate brain size variation is associated with visual specialization. Visual processing may be critically involved in the treatment of both social and ecological information; relative expansion of parts of the visual system is correlated with both frugivory and group size (Barton, 1999). What is clear, however, is that evolution has shaped mammalian brains in response to the demands of their lifestyles, with convergent evolution of brain structures in several groups (De Winter and Oxnard, 2001; Kaas, 2002).
It was long believed that the avian forebrain was composed of hypertrophied basal ganglia with only meager pallial derivatives, whereas in mammals, the pallium had grown into a highly parcellated laminar neocortex (Ariens-Kappers et al., 1936). The recent revised nomenclature of the avian telencephalon (Reiner et al., 2004) recognizes many anatomical and functional similarities between the avian and mammalian forebrains. In particular, the newly named avian nidopallium, mesopallium, and arco-pallium are considered homologous to mammalian pallial derivatives, the neocortex, claustrum, and pallial amygdala (Karten, 1969, 1991; Giintiirkiin, 1991; Wild et al., 1993; Butler, 1994; Veenman et al., 1995; Striedter, 1997; Reiner et al., 1998; Smith-Fernandez et al., 1998; Medina and Reiner, 2000; Puelles et al., 2000).
A large amount of variance in the size of both adult and hatchling avian brains can be explained by the altricial versus precocial dichotomy in development mode. Birds that develop slowly and require extensive parental care are born with relatively smaller brains than birds that are mobile only a few minutes or hours after birth. The reverse applies to adult brain size, where altricial birds have larger brains than do precocial ones (Portmann, 1946; Bennett and Harvey, 1985a, 1985b). In Bennett and Harvey's study, most of the ecological variables (e.g., diet, habitat) that showed a relationship with relative forebrain size in univariate analyses became nonsignificant when development mode was included in multivariate statistics. Only mating system (monogamous ! polygynous) and mode of prey capture (moving from a perch vs. other categories) remained significant predictors of relative size of the brain and forebrain. All large-brained avian clades develop slowly, but the reverse is not true. The altricial Columbiformes (pigeons and doves) and Caprimulgi (nightjars) are, in relative terms, not much more encephalized that the preco-cial Galliformes and ratites. On average, growing a large brain may impose some limits on incubation energetics, as well as the length of intrauterine growth. It might not be possible for birds to stock sufficient energy in the egg for extensive brain growth. Stark (1993) offers some interesting observations of the comparison of brain growth in altricial and precocial birds. He found that in the freshly hatched buttonquail, a precocial species, all telencephalic areas and fiber pathways have undergone differentiation and started myelinization. There are no more areas of cell proliferation in the hatchling, indicating that the number of neurons is definite and that postnatal volume increase can be exclusively attributed to growth. Similarly, the optic tectum of the hatchling muscovy duck has almost reached adult size and differentiation level. In contrast, the three altricial species that Stark (1993) studied have a significant posthatch cell proliferation in the periventricular zone. In the budgerigar and Java sparrow, cell proliferation continues until the 10th postnatal day. Stark suggests that a large postnatal increase in the volume of the brain depends on a persistence of this large periventricular proliferation zone, which can be maintained as long as there are no functional demands on the developing systems. This is possible in altricial hatchlings freed from the need for a functional forebrain by extended parental care. Stark (1993) proposed that to arrive at a larger brain volume, more cells have to divide during the proliferation phase. Theoretically, the increase in cell numbers can be achieved in three ways: increasing the rate of periventricular cell division, increasing the area of cell proliferation, or lengthening the proliferation period. There is no empirical support for the first two options in birds; the third option is possible only in altricial species where parental care compensates for the lack of functional independence by the chicks. Stark (1993) also suggests that nutritional constraints on the hatchlings affect the options: precocial species tend to eat foods that can be easily obtained, while altricial ones eat foods that are widely dispersed or difficult to find. In a recent study, Iwaniuk and Nelson (2003) corroborate the conclusions of Stark (1993) using continuous development time measures in addition to the dichotomous precocial/ altricial classification. They divide development into four periods: incubation, fledging, postfledging parental care, and total period of parental care. All developmental periods except time to fledging are significantly correlated with brain size, once common allometric correlates are removed. The relationships vary with development mode, and Iwaniuk and Nelson (2003) suggest that factors such as diet and foraging techniques interact with development in determining brain size.
If the relationship between ecology and encepha-lization appears to be confounded by developmental constraints, this is not the case for more direct measures of cognition. Lefebvre and collaborators have quantified avian cognition in the field by measuring the frequency of novel, unusual, or rare feeding behavior in over 800 species in five areas of the world (see Lefebvre et al., 1997, 1998 for examples). They have collated over 2300 cases of innovative feeding and 130 cases of tool use and shown that both measures of cognition show taxo-nomic distributions that are positively correlated with relative size of the brain and forebrain (Lefebvre et al., 1997, 1998, 2001, 2002, 2004; Nicolakakis and Lefebvre, 2000). Nine potential confounding variables have been included in these analyses, to ensure that biases inherent to the quantification of anecdotal judgments on novelty and cognitive complexity do not affect the biological trends. One of these confounding variables was development mode, which does not account for the correlation between encephalization and either innovation rate or tool use frequency.
Using the detailed brain data on 32 species from 17 parvorders gathered by Boire (1989) and Rehkamper et al. (1991a), and the innovation and tool use rates from previous papers, Timmermans et al. (2000) and Lefebvre et al. (2002) were able to pinpoint the avian telencephalic areas most closely associated with cognition. Rehkamper and Zilles (1991) and Boire and Baron (1994) have suggested that it is the disproportionate increase of the size of the nidopallium and mesopallium that drove the enlargement of the avian telencephalon. Consistent with these predictions, Timmermans et al. (2000) showed that it is the relative size of the mesopallium that correlates most closely with innovation rate. In simple regressions, the nidopallium, hyperpallium (Wulst), and components of the striatopallidal complex were also all correlated with innovation rate, with or without phylogenetic corrections. In multiple regressions, however, these structures dropped out of the model because of their strong correlations with the size of the mesopallium, which explains a larger proportion of the common variance. Lefebvre et al. (2002) repeated a similar analysis with two types of tool use, true tools and proto-tools, which are described in over 100 avian species (see also Boswall, 1977, 1978, 1983a, 1983b for comprehensive reviews). Proto-tools involve the use of objects that are part of a substrate, e.g., anvils on which prey are battered or dropped, or wedges and thorns with which food is held. True tools are detached from the substrate, e.g., hammers, probes, scoops, sponges, and levers held directly in the beak or foot;
their use is presumed to require more complex cognition that that of proto-tools. Lefebvre et al. (2002) confirmed that true tool users have larger brains than do proto-tools users and that, within the fore-brain, relative size of the nidopallium and mesopallium are the best predictors of avian tool use frequency. The mesopallium comprises higherorder, multimodal processing areas. The nidopal-lium features tertiary areas of this type, but also includes primary projection fields from both soma-tosensory and visual pathways, as well as secondary areas that receive input from these primary fields (Rehkamper et al., 1985). The nidopallium thus has the necessary features for both the cognitive and sensorimotor aspects of tool use, in particular the integration of visual and somatosensory information involved in the fine manipulation of objects.
Beyond these comparative studies of the whole avian spectrum, a few authors have concentrated on encephalization patterns within particular orders such as Anseriformes, Trochiliformes, and Psittaciformes. Iwaniuk and Nelson (2001) recently examined a large number of waterfowl. This group is of particular interest because it is precocial, keeping constant the main confounding variable identified by Bennett and Harvey (1985a). Iwaniuk and Nelson worked from endocasts of museum specimens of 354 individuals representing 55 species. Their analysis did not show any significant relationship between foraging mode or diet and relative brain size in Anseriformes, which does not preclude that further analyses on finer brain structures might not reveal clearer trends. One interesting species in their sample is the musk duck, Biziura lobata. It has a large brain compared to its sister species and also shows a much more altricial mode of development than do other Anseriformes, raising only one or two offspring that do not feed themselves right after hatching, but rely instead on the mother and slowly become independent. Trochiliformes (hummingbirds) were studied by Rehkamper et al., (1991b); they show a level of encephalization intermediate between that of Galliformes and Passeriformes. It is not clear if hummingbirds' encephalization level is a product of relative brain enlargement or selection for small bodies. Boire (1989) and Boire and Baron (1994) suggest that it is cerebellum size that might be the main component of brain enlargement in this order, in line with the complex motor control required for hovering. Terns and swifts, which have more complex flight behavior than other birds, also show an enlarged cerebellum (Boire, 1989; Boire and Baron, 1994). Psittaciformes have recently been examined by Iwaniuk et al. (2005), who measured whole brain size in 180 species, as well as the size of brain regions in 19 species. Their study confirms previous work (Portmann, 1946, 1947a; Boire, 1989, Boire and Baron, 1994) showing that this order, which shows complex cognitive abilities (Pepperberg, 1999, 2002; Borsari and Ottoni, 2005), has a larger telencephalon than other nonpasserine birds, while subtelencephalic brain components show a much smaller range of variation. Psittaciformes are among the birds showing transactional social behavior in the classification proposed by Burish et al. (2004). This complex form of sociality is associated with larger ratios of tele-ncephalon to total brain size.
Besides the avian equivalents of the mammalian neocortex, the mesopallium and nidopallium, areas such as the olfactory bulb and the hippocampus have also been subject to comparative studies. In general, birds are considered microsmatic, but there is increasing evidence that many of them use smell in foraging, orientation, and homing, as well as site and individual recognition (see references in Healy and Guilford, 1990). There is a large database on the size of avian olfactory bulbs (Bang and Cobb, 1968) showing considerable taxonomic variation; unfortunately, these data are not actual volumes, but the ratio between the largest diameter of the olfactory bulb and that of the longest length of the cerebral hemispheres. This measure is not independent of the size and shape of the cerebral hemispheres, and the data should therefore be interpreted with care. Initial interpretations of these data led to the conclusion that large olfactory bulbs are associated with aquatic habitats (Bang and Cobb, 1968; Bang, 1971). A more careful statistical analysis suggested that nocturnal birds have larger olfactory bulbs (Healy and Guilford, 1990). The hypothesis was that olfaction might be useful for birds in low-light conditions for tasks such as site recognition and location of predators and slow-moving or stationary prey. For the moment, this proposed association between activity pattern and olfactory bulb size is interesting, but awaits a more reliable database. It should be noted that large olfactory bulbs in birds are not generally associated with the enlargement of the telencepha-lon. However, it has been suggested that in Anseriformes, the increased telencephalization is in part correlated with enlarged olfactory structures (see Rehkamper et al., 2001). This is shown by the considerable expansion of telencephalic targets of olfactory projections (Ebinger et al., 1992). For example, the olfactory structures in Anseriformes are twice the size of those in the pigeon (Ebinger et al., 1992).
In absolute size, the avian hippocampus is quite small compared to that of mammals, but several studies suggest a correlation between the size of this structure and lifestyles implying more spatial cognition. The hippocampus is larger in food-storing birds than in nonstorers (Krebs et al., 1989; Sherry et al., 1989; Healy and Krebs, 1992, 1996; Healy et al., 1994; Hampton et al., 1995; Basil et al., 1996; Volman et al., 1997). Spatial cognition is not only relevant to food gathering, but also to homing abilities (Rehkamper etal., 1988) and spatial abilities in finding host nests in brood-parasitic cowbirds (Sherry et al., 1993; Reboreda et al., 1996). Some authors (e.g., Bolhuis and Macphail, 2001) have criticized this literature, but Lucas et al. (2004) have recently shown that despite differences between species from North America and Europe (Brodin and Lundborg, 2003), there is a clear correlation between the degree of food-caching specialization and hippocampus size in Corvidae and Paridae. In pigeons, breeds that were artificially selected for homing have a larger telencephalon than nonhoming breeds, and this seems to be the result of an enlarged hippocampus (Rehkamper et al., 1988). In food-caching birds, most studies conducted at the species level report no correlation between the size of the hippocampus and that of the telencephalon (Healy and Krebs, 1992,1996; Healy et al., 1994; Hampton et al., 1995; Basil et al., 1996), but others have found a correlation at the level of the subfamily and family within Passerines (Sherry et al., 1989). It is interesting to note that food-caching experience leads to neurogenesis both in the hippocampus of young marsh tits and in the mesopallium (Patel et al., 1997). This could mean that the more specialized, hippocampal, component of spatial memory may be linked to more generalized problem-solving processes in the mesopallium. This might explain the fact that species differences in food caching in the field are often stronger than those seen in spatial memory tests in captivity. Lefebvre and Bolhuis (2003) report a negative or zero correlation between innovation rate and reliance on food caching in corvids and parids. If the captive tests solicit both specialized spatial memory and more general problem-solving ability, interspe-cifc differences would be magnified by a positive correlation between the two processes, but dampened by a negative or zero correlation. More research is clearly needed on this point. In the other intensively studied avian specialization, imitated song (Jarvis et al., 2000), correlated evolution of small, specialized nuclei and larger telencephalic structures has been suggested by DeVoogd and co-workers. Airey et al. (2000) have shown that zebra finches have heritable variation in both the size of their song repertoire and the size of nucleus HVC, the control center for syllable organization (Yu and Margoliash, 1996). HVC size is positively correlated with whole telencephalon size in zebra finches, leading DeVoogd (2004) to suggest that song repertoire might be an honest signal for general cognitive ability.
Was this article helpful?