- Email: [email protected]
A critique of the neuroecology of learning and memory
426
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
A critique of the neuroecology of learning and memory Johan J. Bolhuis and Euan M. Macphail Recent years have seen the emergence of neuroecology, the study of the neural mechanisms of behaviour guided by functional and evolutionary principles. This research has been of enormous value for our understanding of the evolution of brain- and species-specific behaviour. However, we question the validity of the neuroecological approach when applied to the analysis of learning and memory, given its arbitrary assumption that different ‘problems’ engage different memory mechanisms. Differences in memory-based performance in ‘natural’ tasks do not prove differences in memory capacity; similarly, differences in the use of memory in the natural environment do not provide a sound basis for expecting differences in anatomical structures that subserve learning and memory. This critique is illustrated with examples taken from the study of the neurobiology of food storing and song learning in birds.
Tinbergen formulated the four basic questions that are important in the study of animal behaviour1, based on Julian Huxley’s three questions for biology in general. Tinbergen followed Huxley in distinguishing between the causation (proximal factors), function (survival value) and evolution of behaviour, and he added a fourth factor: development2. Following Tinbergen, it is now generally agreed that the study of animal behaviour should involve all these four aspects. However, within such an integrated approach it should be clear that functional questions and questions about mechanisms are fundamentally different2,3, and furthermore, that results from one domain cannot be used as explanations in the complementary domain. Thus, for example, a functional interpretation of why an animal performs a specific behaviour does not explain the cognitive and neural mechanisms governing that behaviour. Johan J. Bolhuis Dept of Behavioural Biology, Institute of Evolutionary and Ecological Sciences, Leiden University, PO Box 9516, 2300 RA Leiden, The Netherlands. Dept of Behavioural Biology, Utrecht University, Padualaan 14, 3584 CH, Utrecht, The Netherlands. e-mail: [email protected] Euan M. Macphail Dept of Psychology, University of York, Heslington, York, UK YO1 5DD.
Cause and function in animal behaviour and neurobiology
In the study of animal cognition there have recently been several attempts to achieve a synthesis between a functional and a causal approach, in what has been termed the ‘ecological’ or ‘synthetic’ approach4,5, or ‘cognitive ecology’6,7. Shettleworth5 asked what is for us the crucial question: can function explain mechanism? She suggests that the best examples of an affirmative answer to this question5,8,9 come from interpretations that involve modules10, which are ‘distinguishable cognitive mechanisms’ (Ref. 5). However, the idea that ‘memory’ consists of independent domain-specific memory stores is not congruent with Fodor’s account10 (see Box 1). http://tics.trends.com
The ecological or synthetic approach has also been followed in behavioural neuroscience7,11–14. Shettleworth11 coined the phrase ‘inverse neuropsychology’ for this approach, which was recently described as ‘rather than attempting to understand brain function by comparing normal to brain-damaged subjects, as is usually done in neuropsychology, investigators taking a comparative approach can look to examples of evolutionary enhanced function for clues to what specific brain areas do’15. Intuitively, the reverse neuropsychology approach has advantages compared with the traditional approach. Neural manipulations such as lesions are notoriously difficult to interpret, and it would seem eminently sensible to study the neural mechanisms of behavioural processes that are naturally enhanced to serve a specific function. In the context of spatial memory in birds, Smulders and DeVoogd argue that, as food hoarding is ‘a naturally occurring behavior, shaped by natural selection’, it is a particularly good model for the study of memory, as ‘we can therefore assume that the underlying neural mechanisms have been optimized by evolution to perform the required behavior better or more efficiently’14. By way of shorthand – and in an attempt to avoid misunderstanding – the term ‘neuroecology’16 will be used to denote the functional approach to behavioural neuroscience and to distinguish it from neuroethology17, with which it should not be confused. This article examines this approach, concentrating on the neuroecology of learning and memory, an area in which, we believe, the application of the approach has made clear its pitfalls. If functional requirements lead to the evolution of different cognitive modules, so the reasoning goes, then there should also be neural modules that are specialized for a particular function. Such neural modules might be manifest as unique brain regions, or alternatively as brain regions with a specialized mechanism. Specific examples of such putative neural modules in the neurobiological study of food storing and song in birds will be discussed. Nervous ecology: the pitfalls of a functional approach
We argue, first, that the crucial claims of a neuroecological approach to learning and memory are not warranted for logical reasons. The essence of this critique is that questions of mechanism (such as neural mechanisms that underlie behaviour) cannot be solved by functional considerations2,3,18. At most, functional or evolutionary considerations can provide clues as to a possible causal analysis of mechanisms. Second, we argue that neuroecological analyses of learning and memory, by virtue of their attempts to study behaviour in an evolutionary or functionally relevant context, tend to rely on data from experiments ‘provided by nature’ (compare with DeVoogd and Székely13, who suggest that the variation seen in behaviour of animals in the wild can be seen as ‘an immense natural experiment’). As such,
1364-6613/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S1364-6613(00)01753-8
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
Box 1. Brains, minds, modules Much of the current interest in the idea that the mind might be ‘modular’ derives from the work of the philosopher and cognitive scientist Jerry Fodor. Three features of Fodor’s accounta of cognitive modules are particularly relevant here: (1) They are domain specific (2) They are innately specified – that is, they are species specific (3) They are hardwired, and located in specific brain regions Fodor takes as examples of modules various ‘input systems’, such as those involved in visual and auditory perception (including language perception). But Fodor supposes that these modules pass information on to central processes (thought, or ‘problem solving’) that are not domain specific. Similarly, we suppose that memory is a central system that is not domain specific and gathers information from domain-specific perceptual input modules. Reference a Fodor, J.A. (1983) The Modularity of Mind: An Essay on Faculty Psychology, MIT Press
they often involve comparisons between groups of animals that exhibit naturally occurring behaviour. But differences between groups in the use made of learning and memory in the wild do not entail differences between groups in the capacity to learn or remember. These difficulties have led the neurobiological study of learning and memory astray, as will be shown with examples in what follows. Examples of alternative methods to study the neurobiology of learning and memory are filial imprinting in chicks19 and olfactory learning in mice20. These paradigms use naturally occurring behaviours, but have no presuppositions concerning the evolutionary or functional significance of the behaviours, or concerning their neural localization. In general, there are at least three ways in which neuroecological interpretations of learning and memory are problematic: (1) in the attempted localization of a specific brain region as the neural substrate of some adaptive specialization of learning or memory; (2) in treating failures of laboratory experiments to support behavioural predictions of the adaptive specialization view; and (3) in demonstrating causal links between correlated behavioural and neural differences. These general issues will be illustrated with examples from two specific neuroecological paradigms: spatial memory in food hoarding birds and song learning in passerine birds. Space oddities: the neuroecology of spatial learning and memory
Several species of birds, particularly in the families Corvidae, Paridae and Sittidae, store food items in their natural environment, to be retrieved at a later date21,22. The number of items stored varies from tens to hundreds in parids to many thousands in the Clark’s nutcracker (Nucifraga columbiana). The time over which the location of these items is remembered can vary from a few days to four months http://tics.trends.com
427
in parids23,24,25, and up to nine months in Clark’s nutcracker22. The capacity and longevity of the memories involved have led many to suppose that birds that store have evolved a particularly efficient spatial memory, an enhancement of memory that is specific to spatial, as opposed to nonspatial, input. This in turn implies that spatial memory is a specialized adaptation or cognitive module that should be expected to obtain independent representation in a discrete brain region. The idea that spatial memory should be localized in a particular brain region – specifically, the hippocampus – gains support from two findings: first, lesions of the hippocampus of a storing species (the black-capped chickadee, Parus atricapillus) result in severe impairment of retrieval of stored food26; second, species that store food have a larger hippocampus (relative to total telencephalic size) than species that do not store food25,27–30. Localization: why the hippocampus? Parallels with mammals. Proponents of the
hippocampus as a site for spatial memory suggest that the case draws support from parallels with the mammalian hippocampus. This idea depends on two premises: (1) that the issue of homology leads to an expectation of comparability in function; and (2) that the mammalian hippocampus in fact is the site of spatial memory.
‘...there is no evidence from any source to suggest that there are seasonal variations in memory capacity of storing birds...’ Although we do not deny that there is a good case for a general homology between the avian and the mammalian hippocampus based on embryological, histological and hodological data14,31, there are, of course, striking contrasts between avian and mammalian hippocampal anatomy. The gross appearance of the structures is very different and there is, for example, no apparent counterpart in birds of the mammalian trisynaptic pathway. Given in addition the fact that the phylogenetic histories of birds and mammals diverged at least 300 million years ago, there is clearly no confident expectation that avian and mammalian hippocampal function should be closely comparable. The idea that the mammalian hippocampus is selectively involved in spatial memory is also open to question. There is in fact a growing consensus that the hippocampus is not crucial for memory storage, but that it might be involved in processing contextual or spatial input32,33. For example, neurotoxic lesions restricted to the hippocampus (sparing extrahippocampal cortex and the subiculum) impair performance of rats in a water maze not
428
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
through an effect on memory, but on spatial navigation34. In short, the case for the hippocampus as a site of spatial memory gains little support from work on mammals. Hippocampal enlargement and degree of storing.
Although there have been reports (in corvids30 and parids25,29) of significant positive correlations in storing species between intensity of storing and relative hippocampal size, there are some notable exceptions to the principle. An analysis of 10 food-storing corvid species35 found that (relative to telencephalic volume) the hippocampi of four European corvids were larger than that of Clark’s nutcracker, whose dependence on stored food far exceeds that of any of those birds. Moreover, the relative size of the hippocampus of the pinyon jay, whose dependence on stored food is exceeded only by Clark’s nutcracker, was ranked eighth of the ten corvid species. Similarly, although Volman et al. found that of two food-storing woodpecker species, the scatter-hoarding red-bellied woodpecker Melanerpes carolinus (having, presumably, the greater spatial memory demand) had a relatively larger hippocampus than the larder-hoarding red-headed woodpecker M. erythrocephalus36, the hippocampus of two non-storing woodpeckers (the downy woodpecker, Picoides pubescens and the hairy woodpecker, P. villosus) was comparable in size with the scatter-hoarding species, and larger than that of the food-storing redheaded woodpecker. Dependence on stored food does not, then, give a reliable guide to the relative size of a bird’s hippocampus. Seasonal variation of hippocampal size. The size of the
avian hippocampus might vary with the season, and this has generated the hypothesis that ‘the avian hippocampal formation shows neuroanatomical plasticity associated with seasonal changes in spatial memory demands’37. The evidence in support of the claim is, however, weak. The black-capped chickadee hippocampal formation is largest in autumn, a time that coincides with the most active spell of foodstoring of this bird14. But the hippocampus reverts to its ‘normal’ size relatively soon, and over winter, when the birds still actively recover hidden seeds, the hippocampus is not enlarged. For present purposes, however, our central objection to the hypothesis is that there is no evidence from any source to suggest that there are seasonal variations in memory capacity of storing birds. Behavioural predictions
At the heart of the neuroecological account of spatial memory is the notion that food-storing birds show superior performance to non-storers in tasks that tap spatial memory – and show no memory advantage in non-spatial tasks. Evidence pertinent to this claim has recently been reviewed18, and there was no compelling support for the view. For example, one set of experiments used various analogues of the radial http://tics.trends.com
maze, in which food is stored at several sites at the beginning of a trial, and over the course of a trial birds have to remember which sites have been visited and no longer contain food. The radial maze is a conventional test of spatial memory in rats, and is susceptible to hippocampal damage38. Of four published studies, one used parids and found no advantage for a storing versus a non-storing species39. Three reports compared corvid species that differ markedly in the degree to which they rely upon stored food, and did not find a consistent association between degree of reliance on stored food and performance40–42. In the most recent of these reports, Pinyon jays (who have a high dependence on stored food) and scrub jays (who have a much lesser dependence) performed significantly better than both jackdaws (Corvus monedula; who do not store food) and Clark’s nutcrackers (who have the highest dependence of any corvid on stored food)42. Causal links Hippocampal size. The notion that there is a causal
link between hippocampal size and spatial memory capacity has already been much weakened by the evidence reviewed above, showing that the correlations between hippocampal size and ecological spatial demands are unreliable. Hippocampal damage. Hippocampal damage in blackcapped chickadees26 disrupts retrieval of stored food; but this does not of itself demonstrate a causal link between the hippocampus and spatial memory, because the effects of the lesions were not confined to disruption of spatial learning. When tested on a task in which food was associated with a visual cue, birds with hippocampal lesions made many more re-visits to incorrect cues than did controls. Similarly, one study that explored effects of hippocampal lesions in non-storing species found disruption of nonspatial tasks31. It might, therefore, be the case that disruption of spatial tasks by hippocampal damage in birds is a consequence of a more general deficit that encompasses specific types of spatial learning along with specific types of non-spatial learning. Two reports by Hampton and Shettleworth directly challenge the notion that an enlarged hippocampus should enhance performance in tasks that are susceptible to hippocampal damage43,44. These reports have found that there are spatial memory tasks, which are susceptible to disruption by hippocampal lesions in both a storing (black-capped chickadees) and a nonstoring (juncos, Junco hyemalis) species, in which the performance of (intact) non-storing juncos is either no different from or superior to that of intact chickadees. It is clear, then, that ‘sensitivity of a given task to hippocampal damage does not predict the direction of memory differences between storing and nonstoring species’ (Ref. 44). The junco hippocampus is relatively small compared with that of the black-capped chickadee; the lesion results support the notion that
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
the hippocampus is involved in these spatial tasks. It is therefore clear that an enlarged hippocampus is not reliably associated with superior performance in spatial memory tasks. Singing and the brain: the neurobiology of bird song learning
The study of the neural substrates of bird song learning has become a prominent model system for the investigation of the neural mechanisms of learning and memory45–48 (see Box 2). The basic neuroecological idea is that there is a functional need for a brain system that subserves the adaptive specialization of song learning and storage of auditory memory. Thus, a comparison of the brains of males and females49, and of songbird males with non-songbird males50 should provide important clues as to the brain regions that are important for this behaviour.
‘...evolutionary or functional considerations cannot explain the neural mechanisms of behaviour in general, and of learning and memory in particular...’ Advocates of the neurecological approach to bird song often point out that song learning has the advantage of occurring readily in nature. For example, DeVoogd and Székely advocate an integration of causal and functional approaches to song learning13. They note that there is enormous between-species variation in song structure, as well as in the characteristics of song acquisition and production, and that ‘the variation can be seen as an immense natural experiment: natural selection has modified the song system so as to produce thousands of different patterns for a learned behavior. Comparative study of brain-behavior relations then can be a powerful tool for understanding the neural algorithms used in song learning and expression’13 (italics added). Although we think that functional considerations can provide valuable clues for the analysis of mechanisms, we have difficulty with the statement in italics, which essentially implies that comparative studies of function can be used to understand neural mechanisms. Our concerns will be illustrated with specific examples from the bird song-learning literature. Localization: why the ‘song control nuclei’? The logic of lesions. A series of neuroanatomical and
lesion studies led to the identification of a number of ‘song control nuclei’ as being involved in bird song (Box 2). Nuclei in the rostral pathway, in particular, were thought to be involved in male song learning. The effects of brain lesions are notoriously difficult to interpret, and lesions of brain regions that differ between males and females or songbirds and http://tics.trends.com
429
non-songbirds are no exception. Thus, disruptive effects of lesions to a ‘song control nucleus’, even when these effects are limited to juveniles (Box 2), do not prove that that nucleus is crucial for auditory learning and memory. Neural correlates within and outside the ‘song control nuclei’. Electrophysiological investigations have
revealed that there are neurons in the ‘song control nuclei’ that respond selectively to conspecific song48,51,52. Interestingly, the majority of neurons in these regions responds better to the bird’s own song (BOS) than to that of another conspecific51,52, and even than to the tutor song48,53. Some neurons in these brain nuclei respond equally well to BOS and tutor song, while a small proportion respond more to the tutor song48. This might mean that there is a neural representation of the tutor song in the ‘song control nuclei’. Neurons in these nuclei are activated when the bird is singing46–48,51. In addition, there is increased expression of immediate–early genes in the ‘song control nuclei’ when the bird is singing, but not when it hears song, including tutor song54,55. Thus, it is likely that these brain nuclei are either involved in song production only, or in the auditory feedback of songs that occurs during the sensorimotor phase of song learning, and that can also occur in adult songbirds48. Nottebohm et al. concluded in 1990 that ‘we do not know yet how learned sounds are processed’ and ‘we do not know how and where learned sounds are stored’45, a view that was reiterated in a recent review by Nottebohm46. Likewise, Doupe and her collaborators48, who investigated neuronal responsiveness within the ‘song control nuclei’ in detail, suggested that ‘it seems likely that a pure sensory representation of tutor song is present somewhere in the brain’ and that ‘it seems…plausible that such a representation lies elsewhere in the brain, perhaps in the earlier high-level auditory areas that also process songs of conspecifics’. The latter suggestion, which is similar to one made by Nottebohm46, is a reference to areas outside the ‘song control nuclei’ [in particular the caudal medial neostriatum (NCM) and the caudal medial hyperstriatum ventrale (CMHV; see Box 2)], which were implicated in song learning in recent studies involving the expression of immediate–early genes55,56. Recent findings55 (Box 2) lend support to these suggestions. Of course, these recent findings do not imply that there is an ‘adaptively specialized’ designated brain region specifically involved in song learning. For example, CMHV has an area of overlap with IMHV (intermediate and medical hyperstriatum ventrale), which is part of the neural substrate for filial imprinting19, and also seems to be involved in the memory for stored food in chickadees14. Sex- and season-related effects. Nottebohm found that there were seasonal changes in the size of some ‘song control nuclei’ in the forebrain of the canary, an
430
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
Box 2. Where do bird brains store songs? Of all avian species, approximately half are songbirds (Passeriformes: Oscines). Usually, songbird males learn their song from an adult tutor when they are young. There are two phases in songbird song learning: a memory phase when the vocal information of the tutor song is stored (usually in a sensitive period) and a sensorimotor phase later in life, when the bird’s own vocal output is compared with the stored informationa–d. A distinction can be made between ‘age-limited learners’ and ‘openended learners’a. The former (e.g. the zebra finch, white crowned sparrow) do not alter their songs when they are adult. The latter (e.g. the canary), continue to alter their songs when adult, usually for every new breeding season. Until recently it was thought that two forebrain pathways connecting several ‘song control nuclei’ comprised the neural substrate for bird song (Fig. I). This suggestion arose from a series of neuroanatomical and lesion studiesc–e. The caudal pathway was considered to be involved in the production of song. Lesions to nuclei in this pathway, or to any of its connections result in immediate, profound and irreversible deficits in song in adult birdsd,e. The rostral pathway was thought to play a role in song learning. This suggestion was supported by the finding that bilateral lesions to lMAN or Area X disrupt song acquisition, but have little effect on crystallized song in adultsf. However, despite the obvious involvement of these forebrain pathways in song, it is still not known where in the songbird brain learned sounds are storedc,g,h. Recent studies have analysed the expression of specific immediate–early genes in the forebrain of male songbirds that
HP
LH
HVC
CMHV NCM
lMAN
RA
Cb DLM
Area X
nXIIts TRENDS in Cognitive Sciences
had been exposed to conspecific songi–l. Expression of these genes or their protein products is thought to be a reflection of neuronal activation. It was found that song perception did not lead to neuronal activation in the traditional ‘song control nuclei’, but in different regions, particularly the caudal part of the medial neostriatum (NCM) and the caudal part of the medial
‘open-ended learner’57. In particular, the size of the high vocal centre (HVC) and robust nucleus of the archistriatum (RA) decreases to ~50% at end of the breeding season, and increases thereafter. Seasonal changes in brain nucleus volume are accompanied by changes in neuron size and number in the HVC, and by dendritic growth and synaptogenesis in RA (Ref. 58). Nottebohm suggested that the seasonal changes in the size of these nuclei were related to the learning and forgetting of song repertoire each year57. That is, dendrites grow and new synapses are formed as the bird learns to produce new song elements in the spring. Nottebohm’s suggestion of a ‘brain for all seasons’57 was further investigated by Brenowitz and his colleagues, among others58–62. These authors argued that if seasonal changes in the size of song nuclei are related to learning and forgetting of song repertoires, then such seasonal variation should not occur in age-limited learners, who do not change their repertoire once they are adult. However, in the rufous-sided towhee (Pipilo erythrophtalmus), artificially induced ‘seasonal changes’ lead to changes in the size of HVC, RA and Area X, in the absence of any changes in song repertoire size59. Furthermore, when Brenowitz et al. compared brain nucleus size with song learning and repertoire size in a number of songbird species, they did not find a consistent relationship between these factors59. That http://tics.trends.com
V
is, there are seasonal neural changes in open-ended and age-limited learners alike, and they are independent of the size of the song repertoire58,59. Finally, two groups of marsh wrens (Cistothorus palustris) that had acquired different size song repertoires nevertheless did not differ in the volume, neuronal size, number or density of HVC and RA, two ‘song control nuclei’60. In most songbird species, only the males sing. It is therefore implicitly assumed that in these species only males learn their songs. In general, the ‘song control nuclei’ are much smaller in females than in males, and sometimes they are non-existent in females49. This is consistent with the suggestion that ‘extreme sex differences in the song-control system have co-evolved with extreme sex differences in singing behavior’49. However, these neural sex differences do not prove in any way that they are related to putative differences in vocal learning. For example, in the African duetting bush shrike (Laniarius funebris), the HVC and RA and their numbers of neurons are about twice as large in males as in females. However, both sexes have a similar size repertoire, with songs of similar complexity, and they are able to learn the same song types63. In addition, female songbirds of several species, who do not sing, nevertheless show seasonal changes comparable with those in males, in the volume of ‘song control nuclei’61,62.
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
Fig. I. A parasagittal view of the brain of a songbird, with some brain regions and connections that are thought to be involved in birdsong. (Approximate positions of nuclei and brain regions are shown.) A series of lesion studies in adult and young songbirds has led to the distinction between a caudal pathway (red arrows), including the high vocal centre (HVC) and the robust nucleus of the archistriatum (RA), considered to be involved in the production of songe, and a rostral pathway (purple arrows), including HVC, lateral part of the magnocellular nucleus of the neostriatum (lMAN) and Area X, thought to play a role in song learningf. The NCM and CMHV (green) have been found to be involved in song perceptionj–l, and possibly in the storage of the tutor songm,n. Scale bar, 1 mm. Abbreviations: Cb, cerebellum; CMHV, caudal medial hyperstriatum ventrale; DLM, medial part of the dorsolateral thalamic nucleus; HP, hippocampus; HVC, high vocal centre; lMAN, lateral part of the magnocellular nucleus of the neostriatum; LH, lamina hyperstriatica; nXIIts, tracheosyringeal portion of the nucleus hypoglossus; NCM, caudal medial neostriatum; RA, robust nucleus of the archistriatum; V, ventricle. Modified from Refs m,n.
hyperstriatum ventrale (CMHV, see Fig. I). Song production by itself does lead to immediate–early gene expression in the ‘song control nuclei’k. It has been suggested recently that NCM and CMHV might be the neural substrate for the storage of the tutor songg,h,m. This suggestion has received some support from recent studies that demonstrate a significant correlation between neuronal activation in the NCM and the strength of song learningm,n. References a Marler, P. (1987) Sensitive periods and the roles of specific and general sensory stimulation in birdsong learning. In Imprinting and Cortical Plasticity. Comparative Aspects of Sensitive Periods (Rauschecker, J.P. and Marler, P., eds), pp. 99–135, John Wiley and Sons
b Marler, P. (1991) Song-learning behavior: the interface with neuroethology. Trends Neurosci. 14, 199–206 c Nottebohm, F. et al. (1990) Song learning in birds: the relation between perception and production. Philos. Trans. R. Soc. London Ser. B 329, 115–124 d DeVoogd, T.J. (1994) The neural basis for the acquisition and production of bird song. In Causal Mechanisms of Behavioural Development (Hogan, J.A. and Bolhuis, J.J., eds), pp. 49–81, Cambridge University Press e Nottebohm, F. et al. (1976) Central control of song in the canary. J. Comp. Neurol. 165, 457–486 f Bottjer, S.W. et al. (1984). Forebrain lesions disrupt development but not maintenance of song in passerine birds. Science 224, 901–903 g Nottebohm, F. (2000) The anatomy and timing of vocal learning in birds. In The Design of Animal Communication (Hauser, M.D. and Konishi, M., eds), pp. 63–110, MIT Press h Solis, M.M. et al. (2000) Song selectivity and sensorimotor signals in vocal learning and production. Proc. Natl. Acad. Sci. U. S. A. 97, 11836–11842. i Mello, C.V. et al. (1992) Song presentation induces gene expression in the songbird forebrain. Proc. Natl. Acad. Sci. U. S. A. 89, 6818-6822 j Mello, C. V. and Clayton, D. F. (1994) Song-induced ZENK gene expression in auditory pathways of songbird brain and its relation to the song control system. J. Neurosci. 14, 6652–6666 k Jarvis, E. D. and Nottebohm, F. (1997) Motor-driven gene expression. Proc. Natl. Acad. Sci. U. S. A. 94, 4097–4102 l Clayton, D.F. (2000) The neural basis of avian song learning and perception. In Brain, Perception, Memory: Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 113–125, Oxford University Press m Bolhuis, J.J. et al. (2000) Localized neuronal activation in the zebra finch brain is related to the strength of song learning. Proc. Natl. Acad. Sci. U. S. A. 97, 2282–2285 n Bolhuis, J.J. et al. (2001) Localized immediate early gene expression related to the strength of song learning in socially reared zebra finches. Eur. J. Neurosci. 13, 2165–2170
Behavioural predictions
Songbirds learn their songs and most non-songbirds do not. Similarly, usually male songbirds sing and females do not. The association of these differences with differences in anatomical structures raises the following questions: (1) Are songbirds superior to non-songbirds in auditory learning? (2) Are songbird males better auditory learners than females? The specific question that we are concerned with is whether non-songbirds and female songbirds are able to learn and store auditory information. Some non-songbirds such as parrots and hummingbirds can learn vocalizations, so vocal learning is not restricted to songbirds. Other non-songbirds such as doves and pigeons can learn to discriminate between different sounds64. There are songbirds species where both males and females sing63,65, and where both sexes can learn songs equally well65. Furthermore, females in some songbird species can be made to sing by administering testosterone to adults47,66, or oestrogen to hatchlings67. Females have been shown to sing songs they have learned from an adult conspecific, usually their male partner65–67. Furthermore, in some songbird species where the females do not sing, they do learn the characteristics of tutor songs, for which they show a preference in subsequent tests68,69. The strength of zebra finch female song preferences http://tics.trends.com
431
measured in operant tests is not different from those of their male counterparts70. Taken together, the evidence suggests that non-songbirds and female songbirds are capable of learning and storing auditory information in a similar fashion to songbird males. Causal links
As we have seen, the comparisons of songbird males with non-songbirds or females have yielded ambiguous results. A problem with such analyses is that there are many differences between males and females, and between songbirds and non-songbirds, apart from the fact that one sings and the other does not. Thus, it is difficult to draw conclusions as to whether brain nuclei that occur in songbirds and not in non-songbirds50, or that are larger in males than in females49, are involved in song learning or in some other behaviour that differs between these groups of birds. It is possible that differences between these groups that are in fact related to song learning occur in other regions of the brain that do not show between group differences in occurrence, size or cytoarchitecture. This is not to say that the neuroecological analyses of bird song have been in vain. It is likely (although not certain) that the brain regions that have been found in songbirds and not in non-songbirds (or that are larger in songbird males than in females) have something to do with vocal
432
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
Questions for future research • Are there any examples of ‘cognitive modules’ – or is adaptive specialization limited to perceptual and motor mechanisms? • What is the neural substrate for spatial learning and memory in foodstoring birds? Is the same brain region involved in spatial learning and memory in non-storers? Furthermore, if hippocampal enlargement in food-storing birds is associated neither with memory nor with the motivation or capacity to store food, why does it occur? • Does auditory learning in female songbirds and in non-songbirds involve the same brain regions as those used by male songbirds? What is the neural substrate for the memory of the tutor song in male songbirds? • Are the areas involved in memory storage for a given input determined solely by the modality of the input, or is storage of some types of input of a given modality independent of storage of other types? For example, are areas that store spatial information about food sources distinct from areas that store spatial information about territorial boundaries? Do the areas that store learned songs differ from those that store other learned auditory input? • Is the capacity and longevity of memory for input of a given modality determined solely by such ‘low-level’ factors as salience, or are ‘higher’ factors also involved – factors associated with the nature of the central processing engaged by the input?
Acknowledgements We are grateful to Tom Smulders, Katharina Riebel and Gabriel Beckers for discussion, and to the anonymous referees for their constructive comments.
behaviour. But of course there is more to vocal behaviour than learning and the memorization of songs. Vocal behaviour also involves perception and production of songs, with all the complicated sub-processes that that entails. We have seen that the seasonal changes in size of the ‘song control nuclei’ are unlikely to be related to song learning, as was thought previously57, but might be involved in some aspects of song production. Brenowitz and colleagues have found that these seasonal changes are not related to the learning of new songs, song repertoire or song output58,59. They suggest that the seasonal changes might be related to song stereotypy: songs being more stereotyped (with less song-to-song variability) during the spring breeding season58. Similarly, it is not clear what aspect of song is related to (season invariant) within-species variation in volume of ‘song control nuclei’. Ward et al. have found significant positive correlations between number of song elements copied and HVC volume and neuron number71. These authors
References 1 Tinbergen, N. (1963) On aims and methods in ethology. Z. Tierpsychol. 20, 410–433 2 Bolhuis, J.J. (1999) The development of animal behaviour. From Lorenz to neural nets. Naturwissenschaften 86, 101–111 3 Hogan, J.A. (1994) The concept of cause in the study of behavior. In Causal Mechanisms of Behavioural Development (Hogan, J.A. and Bolhuis, J.J., eds), pp. 3–15, Cambridge University Press 4 Kamil, A.C. (1988) A synthetic approach to the study of animal intelligence. In Comparative Perspectives on Modern Psychology (Leger, D.W., ed.), pp. 230–257, University of Nebraska Press http://tics.trends.com
suggest that ‘naturally occurring variation in neuron number constrains how much song material can be copied or reproduced’. In zebra finch males, MacDougall-Shackleton et al. did not find correlations between the volumes of HVC, lateral part of the magnocellular nucleus of the neostriatum (lMAN), Area X or RA and a number of song parameters72. The only significant correlation was a negative one between lMAN volume and element repertoire size and the number of elements per phrase. On the basis of an evaluation of the literature, Gahr et al. conclude that ‘the songbird model therefore does not support the notion that sex differences in area size and neuron number explain sex differences in a behavior that occurs in both sexes’63. On the basis of an extensive review of the literature on sex differences in the size of the ‘song control nuclei’ in songbirds, MacDougall-Shackleton and Ball suggest that these differences are related to differences in song production49. Concluding remarks: the pros and cons of neuroecology
Our critique of the neuroecological approach should not detract from the undoubted merit of much of the work that has been carried out under this conceptual umbrella. For example, the comparative analysis of relative hippocampal size in food-storing birds and related species has provided us with important information regarding the evolution of brain and behaviour. In addition, the analysis of the ‘song control nuclei’ in songbirds has given us important insights into the neuroanatomy and neurophysiology of the songbird forebrain, and the circuits that play a role in vocalization. It is possible that these brain regions are involved in some way in the storage of vocal information, possibly of the BOS, in singing or in the processing of vocal feedback during song learning. Nevertheless, we maintain that evolutionary or functional considerations cannot explain the neural mechanisms of behaviour in general, and of learning and memory in particular. In addition, our review of the literature demonstrates that, although an application of such functional considerations to the mechanisms of learning and memory might give clues for research, it is often misleading and might provide us with the wrong answers to these important questions.
5 Shettleworth, S.J. (1998) Cognition, Evolution, and Behavior, Oxford University Press 6 Real, L.A. (1993) Toward a cognitive ecology. Trends Ecol. Evol. 8, 413–417 7 Healy, S. and Braithwaite, V. (2000) Cognitive ecology: a field of substance? Trends Ecol. Evol. 15, 22–26 8 Sherry, D.F. and Schacter, D. L. (1987) The evolution of multiple memory systems. Psychol. Rev. 94, 439–454 9 Cosmides, L. and Tooby, J. (1995) From function to structure: the role of evolutionary biology and computational theories in cognitive neuroscience. In The Cognitive Neurosciences (Gazzaniga, M., ed.), pp. 1199–1210, MIT Press
10 Fodor, J.A. (1983) The Modularity of Mind: An Essay on Faculty Psychology, MIT Press 11 Shettleworth, S.J. (1995) Comparative studies of memory in food storing birds: from the field to the Skinner box. In Behavioral Brain Research in Naturalistic and Semi-Naturalistic Settings (Alleva, E. et al., eds), pp. 159–194, Kluwer Academic 12 Clayton, N.S. and Krebs, J.R. (1995) Memory in food-storing birds: from behaviour to brain. Curr. Opin. Neurobiol. 5, 149–154 13 DeVoogd, T.J. and Székely, T. (1998) Causes of avian song: using neurobiology to integrate proximate and ultimate levels of analysis. In Animal Cognition in Nature (Balda, R.P. et al., eds), pp. 337–380, Academic Press
Opinion 14 Smulders, T. V. and DeVoogd, T. J. (2000) The avian hippocampal formation and memory for hoarded food: spatial learning out in the real world. In Brain, Perception, Memory. Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 127–148, Oxford University Press 15 Shettleworth, S.J. and Hampton, R.R. (1998) Adaptive specializations of spatial cognition in food-storing birds? Approaches to testing a comparative hypothesis. In Animal Cognition in Nature (Balda, R.P. et al., eds), pp. 65–98, Academic Press 16 Bolhuis, J.J. (2000) Introduction to Part two. In Brain, Perception, Memory. Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 87–91, Oxford University Press 17 Ewert, J.P. (1997) Neural correlates of key stimulus and releasing mechanism: a case study and two concepts. Trends Neurosci. 20, 332–339 18 Macphail, E.M. and Bolhuis, J.J. (2001) The evolution of intelligence: adaptive specialisations versus general process. Biol. Rev. 76, 341–364 19 Horn, G. (1998) Visual imprinting and the neural mechanisms of recognition memory. Trends Neurosci. 21, 300–305 20 Brennan, P.A. and Keverne, E.B. (2000) Neural mechanisms of olfactory recognition memory. In Brain, Perception, Memory. Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 93–112, Oxford University Press 21 Sherry, D.F. et al. (1981). Memory for the location of stored food in marsh tits. Anim. Behav. 29, 1260–1266 22 Balda, R.P. and Kamil, A.C. (1992) Long-term spatial memory in Clark’s Nutcracker, Nucifraga columbiana. Anim. Behav. 44, 761–769 23 Brodin, A. (1994) The disappearance of caches that have been stored by naturally foraging willow tits. Anim. Behav. 47, 730–732 24 Hitchcock, C.L. and Sherry, D.F. (1990) Long-term memory for cache sites in the black-capped chickadee. Anim. Behav. 40, 701–712 25 Healy, S.D. and Krebs, J.R. (1996) Food storing and the hippocampus in Paridae. Brain Behav. Evol. 47, 195–199 26 Sherry, D.F. and Vaccarino, A.L. (1989) Hippocampus and memory for food caches in Black-Capped Chickadees. Behav. Neurosci. 103, 308–318 27 Sherry, D.F. et al. (1989) The hippocampal complex of food-storing birds. Brain Behav. Evol. 34, 308–317 28 Sherry, D. F. (1998) The ecology and neurobiology of spatial memory. In Cognitive Ecology: The Evolutionary Ecology of Information Processing and Decision Making (Dukas, R., ed.), pp. 261–296, Chicago University Press 29 Krebs, J.R. et al. (1989) Hippocampal specialization of food-storing birds. Proc. Natl. Acad. Sci. U. S. A. 86, 1388–1392 30 Healy, S.D. and Krebs, J.R. (1992) Food storing and the hippocampus in corvids: amount and volume are correlated. Proc. R. Soc. London Ser. B 248, 241–245 31 Colombo, M., and Broadbent, N. (2000). Is the avian hippocampus a functional homologue of the mammalian hippocampus? Neurosci. Biobehav. Rev. 24, 465–484 32 Brown, M.W. (2000) Neuronal correlates of recognition memory. In Brain, Perception, Memory. Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 185–208, Oxford University Press http://tics.trends.com
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
33 Buckley, M.J. and Gaffan, D. (2000) The hippocampus, perirhinal cortex and memory in the monkey. In Brain, Perception, Memory. Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 279–298, Oxford University Press 34 Bolhuis, J.J. et al. (1994) Retrograde amnesia and memory reactivation in rats with ibotenate lesions to the hippocampus or subiculum. Q. J. Exp. Psychol. 47B, 129–150 35 Basil, J.A. et al. (1996) Differences in hippocampal volume among food storing corvids. Brain Behav. Evol. 47, 156–164 36 Volman, S.F. et al. (1997) Relative hippocampal volume in relation to food-storing behavior in four species of woodpeckers. Brain Behav. Evol. 49, 110–120 37 Clayton, N.S. et al. (1997). Seasonal changes of hippocampus volume in parasitic cowbirds. Behav. Processes 41, 237–243 38 Cassel, J.C. et al. (1998). Fimbria-fornix vs selective hippocampal lesions in rats: effects on locomotor activity and spatial learning and memory. Neurobiol. Learn. Mem. 69, 22–45 39 Hilton, S.C. and Krebs, J.K. (1990). Spatial memory of four species of Parus: performance in an open-field analogue of a radial maze. Q. J. Exp. Psychol. 43B, 345–368 40 Kamil, A.C. et al. (1994). Performance of 4 seedcaching corvid species in the radial-arm maze analog. J. Comp. Psychol. 108, 385–393 41 Balda, R.P. et al. (1997). Species differences in spatial memory performance on a threedimensional task. Ethology 103, 47–55 42 Gould-Beierle, K.L. (2000). A comparison of four corvid species in a working and reference memory task using a radial maze. J. Comp. Psychol. 114, 347–356 43 Hampton, R.R. and Shettleworth, S.J. (1996). Hippocampal lesions impair memory for location but not color in passerine birds. Behav. Neurosci. 110, 831–835 44 Hampton, R.R. and Shettleworth, S.J. (1996). Hippocampus and memory in a food-storing and in a nonstoring bird species. Behav. Neurosci. 110, 946–964 45 Nottebohm, F. et al. (1990) Song learning in birds: the relation between perception and production. Philos. Trans. R. Soc. London Ser. B 329, 115–124 46 Nottebohm, F. (2000) The anatomy and timing of vocal learning in birds. In The Design of Animal Communication (Hauser, M.D. and Konishi, M., eds), pp. 63–110, MIT Press 47 DeVoogd, T.J. (1994) The neural basis for the acquisition and production of bird song. In Causal Mechanisms of Behavioural Development. (Hogan, J.A. and Bolhuis, J.J., eds), pp. 49–81, Cambridge University Press 48 Solis, M.M. et al. (2000) Song selectivity and sensorimotor signals in vocal learning and production. Proc. Natl. Acad. Sci. U. S. A. 97, 11836–11842 49 MacDougall-Shackleton, S.A. and Ball, G.F. (1999) Comparative studies of sex differences in the song-control system of songbirds. Trends Neurosci. 22, 432–436 50 Gahr, M. (1997) Hormones make songs sexually attractive: hormone-dependent neural changes in the vocal control system of songbirds. Zoology 100, 260–272 51 Margoliash, D. (1986). Preference for autogenous song by auditory neurons in a song system nucleus of the white-crowned sparrow. J. Neurosci. 6, 1643–1661
433
52 Doupe, A.J. and Konishi, M. (1991). Songselective auditory circuits in the vocal control system of the zebra finch. Proc. Natl Acad. Sci. U. S. A. 88, 11339–11343 53 Volman, S.F. (1993) Development of neural selectivity for birdsong during vocal learning. J. Neurosci. 13, 4737–4747 54 Jarvis, E.D. and Nottebohm, F. (1997) Motordriven gene expression. Proc. Natl. Acad. Sci. U. S. A. 94, 4097–4102 55 Bolhuis, J.J. et al. (2000) Localized neuronal activation in the zebra finch brain is related to the strength of song learning. Proc. Natl. Acad. Sci. U. S. A. 97, 2282–2285 56 Mello, C. V. et al. (1992) Song presentation induces gene expression in the songbird forebrain. Proc. Natl. Acad. Sci. U. S. A. 89, 6818–6822 57 Nottebohm, F. (1981) A brain for all seasons: cyclical anatomical changes in song control nuclei of the canary brain. Science 214, 1368–1370 58 Tramontin, A.D. and Brenowitz, E. (2000) Seasonal plasticity in the adult brain. Trends Neurosci. 23, 251–258 59 Brenowitz, E.A. et al. (1991) Seasonal changes in avian song nuclei without seasonal changes in song repertoire. J. Neurosci. 11, 1367–1374 60 Brenowitz, E.A. et al. (1995) Brain space for learned song in birds develops independently of song learning. J. Neurosci. 15, 6281–6286 61 Kirn, J.R. et al. (1989) Song-related brain regions in the red-winged blackbird are affected by sex and season but not repertoire size. J. Neurobiol. 20, 139–163 62 Deviche, P. and Gulledge, C.C. (2000) Vocal control region sizes of an adult female songbird change seasonally in the absence of detectable circulating testosterone concentrations. J. Neurobiol. 42, 202–211 63 Gahr, M. et al. (1998) Sex difference in the size of the neural song control regions in a dueting songbird with similar song repertoire size of males and females. J. Neurosci. 18, 1124–1131 64 Cynx, J. (1995) Similarities in absolute and relative pitch perception in songbirds (starling and zebra finch) and a nonsongbird (pigeon). J. Comp. Psychol. 109, 261–267 65 Baptista, L.F. et al. (1993) Singing and its functions in female white-crowned sparrows. Anim. Behav. 46, 511–524 66 Chilton, G. and Lein, M.R. (1996) Songs and sexual responses of female white-crowned sparrows (Zonotrichia leucophrys) from a mixeddialect population. Behaviour 133, 173–198 67 Simpson, H.B. and Vicario, D.S. (1991) Early estrogen-treatment alone causes female zebra finches to produce learned, male-like vocalizations. J. Neurobiol. 22, 755–776 68 Brenowitz, E. A. (1986) Altered perception of species-specific song by female birds after lesions of a forebrain nucleus. Science 251, 303–305 69 Riebel, K. (2000) Early exposure leads to repeatable preferences for male song in female zebra finches. Proc. R. Soc. London Ser. B. 267, 2553–2558 70 Riebel, K. et al. (2001) Female and male zebra finch siblings do not differ in their adult preferences for the father’s song. Ethology (Suppl. 36), 251 71 Ward, B.C. et al. (1998) Individual variation in neuron number predicts differences in the propensity for avian vocal imitation. Proc. Natl. Acad. Sci. U. S. A. 95, 1277–1282 72 MacDougall-Shackleton, S.A. et al. (1998) Neural correlates of singing behavior in male zebra finches (Taeniopygia guttata). J. Neurobiol. 36, 421–430
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
A critique of the neuroecology of learning and memory Johan J. Bolhuis and Euan M. Macphail Recent years have seen the emergence of neuroecology, the study of the neural mechanisms of behaviour guided by functional and evolutionary principles. This research has been of enormous value for our understanding of the evolution of brain- and species-specific behaviour. However, we question the validity of the neuroecological approach when applied to the analysis of learning and memory, given its arbitrary assumption that different ‘problems’ engage different memory mechanisms. Differences in memory-based performance in ‘natural’ tasks do not prove differences in memory capacity; similarly, differences in the use of memory in the natural environment do not provide a sound basis for expecting differences in anatomical structures that subserve learning and memory. This critique is illustrated with examples taken from the study of the neurobiology of food storing and song learning in birds.
Tinbergen formulated the four basic questions that are important in the study of animal behaviour1, based on Julian Huxley’s three questions for biology in general. Tinbergen followed Huxley in distinguishing between the causation (proximal factors), function (survival value) and evolution of behaviour, and he added a fourth factor: development2. Following Tinbergen, it is now generally agreed that the study of animal behaviour should involve all these four aspects. However, within such an integrated approach it should be clear that functional questions and questions about mechanisms are fundamentally different2,3, and furthermore, that results from one domain cannot be used as explanations in the complementary domain. Thus, for example, a functional interpretation of why an animal performs a specific behaviour does not explain the cognitive and neural mechanisms governing that behaviour. Johan J. Bolhuis Dept of Behavioural Biology, Institute of Evolutionary and Ecological Sciences, Leiden University, PO Box 9516, 2300 RA Leiden, The Netherlands. Dept of Behavioural Biology, Utrecht University, Padualaan 14, 3584 CH, Utrecht, The Netherlands. e-mail: [email protected] Euan M. Macphail Dept of Psychology, University of York, Heslington, York, UK YO1 5DD.
Cause and function in animal behaviour and neurobiology
In the study of animal cognition there have recently been several attempts to achieve a synthesis between a functional and a causal approach, in what has been termed the ‘ecological’ or ‘synthetic’ approach4,5, or ‘cognitive ecology’6,7. Shettleworth5 asked what is for us the crucial question: can function explain mechanism? She suggests that the best examples of an affirmative answer to this question5,8,9 come from interpretations that involve modules10, which are ‘distinguishable cognitive mechanisms’ (Ref. 5). However, the idea that ‘memory’ consists of independent domain-specific memory stores is not congruent with Fodor’s account10 (see Box 1). http://tics.trends.com
The ecological or synthetic approach has also been followed in behavioural neuroscience7,11–14. Shettleworth11 coined the phrase ‘inverse neuropsychology’ for this approach, which was recently described as ‘rather than attempting to understand brain function by comparing normal to brain-damaged subjects, as is usually done in neuropsychology, investigators taking a comparative approach can look to examples of evolutionary enhanced function for clues to what specific brain areas do’15. Intuitively, the reverse neuropsychology approach has advantages compared with the traditional approach. Neural manipulations such as lesions are notoriously difficult to interpret, and it would seem eminently sensible to study the neural mechanisms of behavioural processes that are naturally enhanced to serve a specific function. In the context of spatial memory in birds, Smulders and DeVoogd argue that, as food hoarding is ‘a naturally occurring behavior, shaped by natural selection’, it is a particularly good model for the study of memory, as ‘we can therefore assume that the underlying neural mechanisms have been optimized by evolution to perform the required behavior better or more efficiently’14. By way of shorthand – and in an attempt to avoid misunderstanding – the term ‘neuroecology’16 will be used to denote the functional approach to behavioural neuroscience and to distinguish it from neuroethology17, with which it should not be confused. This article examines this approach, concentrating on the neuroecology of learning and memory, an area in which, we believe, the application of the approach has made clear its pitfalls. If functional requirements lead to the evolution of different cognitive modules, so the reasoning goes, then there should also be neural modules that are specialized for a particular function. Such neural modules might be manifest as unique brain regions, or alternatively as brain regions with a specialized mechanism. Specific examples of such putative neural modules in the neurobiological study of food storing and song in birds will be discussed. Nervous ecology: the pitfalls of a functional approach
We argue, first, that the crucial claims of a neuroecological approach to learning and memory are not warranted for logical reasons. The essence of this critique is that questions of mechanism (such as neural mechanisms that underlie behaviour) cannot be solved by functional considerations2,3,18. At most, functional or evolutionary considerations can provide clues as to a possible causal analysis of mechanisms. Second, we argue that neuroecological analyses of learning and memory, by virtue of their attempts to study behaviour in an evolutionary or functionally relevant context, tend to rely on data from experiments ‘provided by nature’ (compare with DeVoogd and Székely13, who suggest that the variation seen in behaviour of animals in the wild can be seen as ‘an immense natural experiment’). As such,
1364-6613/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S1364-6613(00)01753-8
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
Box 1. Brains, minds, modules Much of the current interest in the idea that the mind might be ‘modular’ derives from the work of the philosopher and cognitive scientist Jerry Fodor. Three features of Fodor’s accounta of cognitive modules are particularly relevant here: (1) They are domain specific (2) They are innately specified – that is, they are species specific (3) They are hardwired, and located in specific brain regions Fodor takes as examples of modules various ‘input systems’, such as those involved in visual and auditory perception (including language perception). But Fodor supposes that these modules pass information on to central processes (thought, or ‘problem solving’) that are not domain specific. Similarly, we suppose that memory is a central system that is not domain specific and gathers information from domain-specific perceptual input modules. Reference a Fodor, J.A. (1983) The Modularity of Mind: An Essay on Faculty Psychology, MIT Press
they often involve comparisons between groups of animals that exhibit naturally occurring behaviour. But differences between groups in the use made of learning and memory in the wild do not entail differences between groups in the capacity to learn or remember. These difficulties have led the neurobiological study of learning and memory astray, as will be shown with examples in what follows. Examples of alternative methods to study the neurobiology of learning and memory are filial imprinting in chicks19 and olfactory learning in mice20. These paradigms use naturally occurring behaviours, but have no presuppositions concerning the evolutionary or functional significance of the behaviours, or concerning their neural localization. In general, there are at least three ways in which neuroecological interpretations of learning and memory are problematic: (1) in the attempted localization of a specific brain region as the neural substrate of some adaptive specialization of learning or memory; (2) in treating failures of laboratory experiments to support behavioural predictions of the adaptive specialization view; and (3) in demonstrating causal links between correlated behavioural and neural differences. These general issues will be illustrated with examples from two specific neuroecological paradigms: spatial memory in food hoarding birds and song learning in passerine birds. Space oddities: the neuroecology of spatial learning and memory
Several species of birds, particularly in the families Corvidae, Paridae and Sittidae, store food items in their natural environment, to be retrieved at a later date21,22. The number of items stored varies from tens to hundreds in parids to many thousands in the Clark’s nutcracker (Nucifraga columbiana). The time over which the location of these items is remembered can vary from a few days to four months http://tics.trends.com
427
in parids23,24,25, and up to nine months in Clark’s nutcracker22. The capacity and longevity of the memories involved have led many to suppose that birds that store have evolved a particularly efficient spatial memory, an enhancement of memory that is specific to spatial, as opposed to nonspatial, input. This in turn implies that spatial memory is a specialized adaptation or cognitive module that should be expected to obtain independent representation in a discrete brain region. The idea that spatial memory should be localized in a particular brain region – specifically, the hippocampus – gains support from two findings: first, lesions of the hippocampus of a storing species (the black-capped chickadee, Parus atricapillus) result in severe impairment of retrieval of stored food26; second, species that store food have a larger hippocampus (relative to total telencephalic size) than species that do not store food25,27–30. Localization: why the hippocampus? Parallels with mammals. Proponents of the
hippocampus as a site for spatial memory suggest that the case draws support from parallels with the mammalian hippocampus. This idea depends on two premises: (1) that the issue of homology leads to an expectation of comparability in function; and (2) that the mammalian hippocampus in fact is the site of spatial memory.
‘...there is no evidence from any source to suggest that there are seasonal variations in memory capacity of storing birds...’ Although we do not deny that there is a good case for a general homology between the avian and the mammalian hippocampus based on embryological, histological and hodological data14,31, there are, of course, striking contrasts between avian and mammalian hippocampal anatomy. The gross appearance of the structures is very different and there is, for example, no apparent counterpart in birds of the mammalian trisynaptic pathway. Given in addition the fact that the phylogenetic histories of birds and mammals diverged at least 300 million years ago, there is clearly no confident expectation that avian and mammalian hippocampal function should be closely comparable. The idea that the mammalian hippocampus is selectively involved in spatial memory is also open to question. There is in fact a growing consensus that the hippocampus is not crucial for memory storage, but that it might be involved in processing contextual or spatial input32,33. For example, neurotoxic lesions restricted to the hippocampus (sparing extrahippocampal cortex and the subiculum) impair performance of rats in a water maze not
428
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
through an effect on memory, but on spatial navigation34. In short, the case for the hippocampus as a site of spatial memory gains little support from work on mammals. Hippocampal enlargement and degree of storing.
Although there have been reports (in corvids30 and parids25,29) of significant positive correlations in storing species between intensity of storing and relative hippocampal size, there are some notable exceptions to the principle. An analysis of 10 food-storing corvid species35 found that (relative to telencephalic volume) the hippocampi of four European corvids were larger than that of Clark’s nutcracker, whose dependence on stored food far exceeds that of any of those birds. Moreover, the relative size of the hippocampus of the pinyon jay, whose dependence on stored food is exceeded only by Clark’s nutcracker, was ranked eighth of the ten corvid species. Similarly, although Volman et al. found that of two food-storing woodpecker species, the scatter-hoarding red-bellied woodpecker Melanerpes carolinus (having, presumably, the greater spatial memory demand) had a relatively larger hippocampus than the larder-hoarding red-headed woodpecker M. erythrocephalus36, the hippocampus of two non-storing woodpeckers (the downy woodpecker, Picoides pubescens and the hairy woodpecker, P. villosus) was comparable in size with the scatter-hoarding species, and larger than that of the food-storing redheaded woodpecker. Dependence on stored food does not, then, give a reliable guide to the relative size of a bird’s hippocampus. Seasonal variation of hippocampal size. The size of the
avian hippocampus might vary with the season, and this has generated the hypothesis that ‘the avian hippocampal formation shows neuroanatomical plasticity associated with seasonal changes in spatial memory demands’37. The evidence in support of the claim is, however, weak. The black-capped chickadee hippocampal formation is largest in autumn, a time that coincides with the most active spell of foodstoring of this bird14. But the hippocampus reverts to its ‘normal’ size relatively soon, and over winter, when the birds still actively recover hidden seeds, the hippocampus is not enlarged. For present purposes, however, our central objection to the hypothesis is that there is no evidence from any source to suggest that there are seasonal variations in memory capacity of storing birds. Behavioural predictions
At the heart of the neuroecological account of spatial memory is the notion that food-storing birds show superior performance to non-storers in tasks that tap spatial memory – and show no memory advantage in non-spatial tasks. Evidence pertinent to this claim has recently been reviewed18, and there was no compelling support for the view. For example, one set of experiments used various analogues of the radial http://tics.trends.com
maze, in which food is stored at several sites at the beginning of a trial, and over the course of a trial birds have to remember which sites have been visited and no longer contain food. The radial maze is a conventional test of spatial memory in rats, and is susceptible to hippocampal damage38. Of four published studies, one used parids and found no advantage for a storing versus a non-storing species39. Three reports compared corvid species that differ markedly in the degree to which they rely upon stored food, and did not find a consistent association between degree of reliance on stored food and performance40–42. In the most recent of these reports, Pinyon jays (who have a high dependence on stored food) and scrub jays (who have a much lesser dependence) performed significantly better than both jackdaws (Corvus monedula; who do not store food) and Clark’s nutcrackers (who have the highest dependence of any corvid on stored food)42. Causal links Hippocampal size. The notion that there is a causal
link between hippocampal size and spatial memory capacity has already been much weakened by the evidence reviewed above, showing that the correlations between hippocampal size and ecological spatial demands are unreliable. Hippocampal damage. Hippocampal damage in blackcapped chickadees26 disrupts retrieval of stored food; but this does not of itself demonstrate a causal link between the hippocampus and spatial memory, because the effects of the lesions were not confined to disruption of spatial learning. When tested on a task in which food was associated with a visual cue, birds with hippocampal lesions made many more re-visits to incorrect cues than did controls. Similarly, one study that explored effects of hippocampal lesions in non-storing species found disruption of nonspatial tasks31. It might, therefore, be the case that disruption of spatial tasks by hippocampal damage in birds is a consequence of a more general deficit that encompasses specific types of spatial learning along with specific types of non-spatial learning. Two reports by Hampton and Shettleworth directly challenge the notion that an enlarged hippocampus should enhance performance in tasks that are susceptible to hippocampal damage43,44. These reports have found that there are spatial memory tasks, which are susceptible to disruption by hippocampal lesions in both a storing (black-capped chickadees) and a nonstoring (juncos, Junco hyemalis) species, in which the performance of (intact) non-storing juncos is either no different from or superior to that of intact chickadees. It is clear, then, that ‘sensitivity of a given task to hippocampal damage does not predict the direction of memory differences between storing and nonstoring species’ (Ref. 44). The junco hippocampus is relatively small compared with that of the black-capped chickadee; the lesion results support the notion that
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
the hippocampus is involved in these spatial tasks. It is therefore clear that an enlarged hippocampus is not reliably associated with superior performance in spatial memory tasks. Singing and the brain: the neurobiology of bird song learning
The study of the neural substrates of bird song learning has become a prominent model system for the investigation of the neural mechanisms of learning and memory45–48 (see Box 2). The basic neuroecological idea is that there is a functional need for a brain system that subserves the adaptive specialization of song learning and storage of auditory memory. Thus, a comparison of the brains of males and females49, and of songbird males with non-songbird males50 should provide important clues as to the brain regions that are important for this behaviour.
‘...evolutionary or functional considerations cannot explain the neural mechanisms of behaviour in general, and of learning and memory in particular...’ Advocates of the neurecological approach to bird song often point out that song learning has the advantage of occurring readily in nature. For example, DeVoogd and Székely advocate an integration of causal and functional approaches to song learning13. They note that there is enormous between-species variation in song structure, as well as in the characteristics of song acquisition and production, and that ‘the variation can be seen as an immense natural experiment: natural selection has modified the song system so as to produce thousands of different patterns for a learned behavior. Comparative study of brain-behavior relations then can be a powerful tool for understanding the neural algorithms used in song learning and expression’13 (italics added). Although we think that functional considerations can provide valuable clues for the analysis of mechanisms, we have difficulty with the statement in italics, which essentially implies that comparative studies of function can be used to understand neural mechanisms. Our concerns will be illustrated with specific examples from the bird song-learning literature. Localization: why the ‘song control nuclei’? The logic of lesions. A series of neuroanatomical and
lesion studies led to the identification of a number of ‘song control nuclei’ as being involved in bird song (Box 2). Nuclei in the rostral pathway, in particular, were thought to be involved in male song learning. The effects of brain lesions are notoriously difficult to interpret, and lesions of brain regions that differ between males and females or songbirds and http://tics.trends.com
429
non-songbirds are no exception. Thus, disruptive effects of lesions to a ‘song control nucleus’, even when these effects are limited to juveniles (Box 2), do not prove that that nucleus is crucial for auditory learning and memory. Neural correlates within and outside the ‘song control nuclei’. Electrophysiological investigations have
revealed that there are neurons in the ‘song control nuclei’ that respond selectively to conspecific song48,51,52. Interestingly, the majority of neurons in these regions responds better to the bird’s own song (BOS) than to that of another conspecific51,52, and even than to the tutor song48,53. Some neurons in these brain nuclei respond equally well to BOS and tutor song, while a small proportion respond more to the tutor song48. This might mean that there is a neural representation of the tutor song in the ‘song control nuclei’. Neurons in these nuclei are activated when the bird is singing46–48,51. In addition, there is increased expression of immediate–early genes in the ‘song control nuclei’ when the bird is singing, but not when it hears song, including tutor song54,55. Thus, it is likely that these brain nuclei are either involved in song production only, or in the auditory feedback of songs that occurs during the sensorimotor phase of song learning, and that can also occur in adult songbirds48. Nottebohm et al. concluded in 1990 that ‘we do not know yet how learned sounds are processed’ and ‘we do not know how and where learned sounds are stored’45, a view that was reiterated in a recent review by Nottebohm46. Likewise, Doupe and her collaborators48, who investigated neuronal responsiveness within the ‘song control nuclei’ in detail, suggested that ‘it seems likely that a pure sensory representation of tutor song is present somewhere in the brain’ and that ‘it seems…plausible that such a representation lies elsewhere in the brain, perhaps in the earlier high-level auditory areas that also process songs of conspecifics’. The latter suggestion, which is similar to one made by Nottebohm46, is a reference to areas outside the ‘song control nuclei’ [in particular the caudal medial neostriatum (NCM) and the caudal medial hyperstriatum ventrale (CMHV; see Box 2)], which were implicated in song learning in recent studies involving the expression of immediate–early genes55,56. Recent findings55 (Box 2) lend support to these suggestions. Of course, these recent findings do not imply that there is an ‘adaptively specialized’ designated brain region specifically involved in song learning. For example, CMHV has an area of overlap with IMHV (intermediate and medical hyperstriatum ventrale), which is part of the neural substrate for filial imprinting19, and also seems to be involved in the memory for stored food in chickadees14. Sex- and season-related effects. Nottebohm found that there were seasonal changes in the size of some ‘song control nuclei’ in the forebrain of the canary, an
430
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
Box 2. Where do bird brains store songs? Of all avian species, approximately half are songbirds (Passeriformes: Oscines). Usually, songbird males learn their song from an adult tutor when they are young. There are two phases in songbird song learning: a memory phase when the vocal information of the tutor song is stored (usually in a sensitive period) and a sensorimotor phase later in life, when the bird’s own vocal output is compared with the stored informationa–d. A distinction can be made between ‘age-limited learners’ and ‘openended learners’a. The former (e.g. the zebra finch, white crowned sparrow) do not alter their songs when they are adult. The latter (e.g. the canary), continue to alter their songs when adult, usually for every new breeding season. Until recently it was thought that two forebrain pathways connecting several ‘song control nuclei’ comprised the neural substrate for bird song (Fig. I). This suggestion arose from a series of neuroanatomical and lesion studiesc–e. The caudal pathway was considered to be involved in the production of song. Lesions to nuclei in this pathway, or to any of its connections result in immediate, profound and irreversible deficits in song in adult birdsd,e. The rostral pathway was thought to play a role in song learning. This suggestion was supported by the finding that bilateral lesions to lMAN or Area X disrupt song acquisition, but have little effect on crystallized song in adultsf. However, despite the obvious involvement of these forebrain pathways in song, it is still not known where in the songbird brain learned sounds are storedc,g,h. Recent studies have analysed the expression of specific immediate–early genes in the forebrain of male songbirds that
HP
LH
HVC
CMHV NCM
lMAN
RA
Cb DLM
Area X
nXIIts TRENDS in Cognitive Sciences
had been exposed to conspecific songi–l. Expression of these genes or their protein products is thought to be a reflection of neuronal activation. It was found that song perception did not lead to neuronal activation in the traditional ‘song control nuclei’, but in different regions, particularly the caudal part of the medial neostriatum (NCM) and the caudal part of the medial
‘open-ended learner’57. In particular, the size of the high vocal centre (HVC) and robust nucleus of the archistriatum (RA) decreases to ~50% at end of the breeding season, and increases thereafter. Seasonal changes in brain nucleus volume are accompanied by changes in neuron size and number in the HVC, and by dendritic growth and synaptogenesis in RA (Ref. 58). Nottebohm suggested that the seasonal changes in the size of these nuclei were related to the learning and forgetting of song repertoire each year57. That is, dendrites grow and new synapses are formed as the bird learns to produce new song elements in the spring. Nottebohm’s suggestion of a ‘brain for all seasons’57 was further investigated by Brenowitz and his colleagues, among others58–62. These authors argued that if seasonal changes in the size of song nuclei are related to learning and forgetting of song repertoires, then such seasonal variation should not occur in age-limited learners, who do not change their repertoire once they are adult. However, in the rufous-sided towhee (Pipilo erythrophtalmus), artificially induced ‘seasonal changes’ lead to changes in the size of HVC, RA and Area X, in the absence of any changes in song repertoire size59. Furthermore, when Brenowitz et al. compared brain nucleus size with song learning and repertoire size in a number of songbird species, they did not find a consistent relationship between these factors59. That http://tics.trends.com
V
is, there are seasonal neural changes in open-ended and age-limited learners alike, and they are independent of the size of the song repertoire58,59. Finally, two groups of marsh wrens (Cistothorus palustris) that had acquired different size song repertoires nevertheless did not differ in the volume, neuronal size, number or density of HVC and RA, two ‘song control nuclei’60. In most songbird species, only the males sing. It is therefore implicitly assumed that in these species only males learn their songs. In general, the ‘song control nuclei’ are much smaller in females than in males, and sometimes they are non-existent in females49. This is consistent with the suggestion that ‘extreme sex differences in the song-control system have co-evolved with extreme sex differences in singing behavior’49. However, these neural sex differences do not prove in any way that they are related to putative differences in vocal learning. For example, in the African duetting bush shrike (Laniarius funebris), the HVC and RA and their numbers of neurons are about twice as large in males as in females. However, both sexes have a similar size repertoire, with songs of similar complexity, and they are able to learn the same song types63. In addition, female songbirds of several species, who do not sing, nevertheless show seasonal changes comparable with those in males, in the volume of ‘song control nuclei’61,62.
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
Fig. I. A parasagittal view of the brain of a songbird, with some brain regions and connections that are thought to be involved in birdsong. (Approximate positions of nuclei and brain regions are shown.) A series of lesion studies in adult and young songbirds has led to the distinction between a caudal pathway (red arrows), including the high vocal centre (HVC) and the robust nucleus of the archistriatum (RA), considered to be involved in the production of songe, and a rostral pathway (purple arrows), including HVC, lateral part of the magnocellular nucleus of the neostriatum (lMAN) and Area X, thought to play a role in song learningf. The NCM and CMHV (green) have been found to be involved in song perceptionj–l, and possibly in the storage of the tutor songm,n. Scale bar, 1 mm. Abbreviations: Cb, cerebellum; CMHV, caudal medial hyperstriatum ventrale; DLM, medial part of the dorsolateral thalamic nucleus; HP, hippocampus; HVC, high vocal centre; lMAN, lateral part of the magnocellular nucleus of the neostriatum; LH, lamina hyperstriatica; nXIIts, tracheosyringeal portion of the nucleus hypoglossus; NCM, caudal medial neostriatum; RA, robust nucleus of the archistriatum; V, ventricle. Modified from Refs m,n.
hyperstriatum ventrale (CMHV, see Fig. I). Song production by itself does lead to immediate–early gene expression in the ‘song control nuclei’k. It has been suggested recently that NCM and CMHV might be the neural substrate for the storage of the tutor songg,h,m. This suggestion has received some support from recent studies that demonstrate a significant correlation between neuronal activation in the NCM and the strength of song learningm,n. References a Marler, P. (1987) Sensitive periods and the roles of specific and general sensory stimulation in birdsong learning. In Imprinting and Cortical Plasticity. Comparative Aspects of Sensitive Periods (Rauschecker, J.P. and Marler, P., eds), pp. 99–135, John Wiley and Sons
b Marler, P. (1991) Song-learning behavior: the interface with neuroethology. Trends Neurosci. 14, 199–206 c Nottebohm, F. et al. (1990) Song learning in birds: the relation between perception and production. Philos. Trans. R. Soc. London Ser. B 329, 115–124 d DeVoogd, T.J. (1994) The neural basis for the acquisition and production of bird song. In Causal Mechanisms of Behavioural Development (Hogan, J.A. and Bolhuis, J.J., eds), pp. 49–81, Cambridge University Press e Nottebohm, F. et al. (1976) Central control of song in the canary. J. Comp. Neurol. 165, 457–486 f Bottjer, S.W. et al. (1984). Forebrain lesions disrupt development but not maintenance of song in passerine birds. Science 224, 901–903 g Nottebohm, F. (2000) The anatomy and timing of vocal learning in birds. In The Design of Animal Communication (Hauser, M.D. and Konishi, M., eds), pp. 63–110, MIT Press h Solis, M.M. et al. (2000) Song selectivity and sensorimotor signals in vocal learning and production. Proc. Natl. Acad. Sci. U. S. A. 97, 11836–11842. i Mello, C.V. et al. (1992) Song presentation induces gene expression in the songbird forebrain. Proc. Natl. Acad. Sci. U. S. A. 89, 6818-6822 j Mello, C. V. and Clayton, D. F. (1994) Song-induced ZENK gene expression in auditory pathways of songbird brain and its relation to the song control system. J. Neurosci. 14, 6652–6666 k Jarvis, E. D. and Nottebohm, F. (1997) Motor-driven gene expression. Proc. Natl. Acad. Sci. U. S. A. 94, 4097–4102 l Clayton, D.F. (2000) The neural basis of avian song learning and perception. In Brain, Perception, Memory: Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 113–125, Oxford University Press m Bolhuis, J.J. et al. (2000) Localized neuronal activation in the zebra finch brain is related to the strength of song learning. Proc. Natl. Acad. Sci. U. S. A. 97, 2282–2285 n Bolhuis, J.J. et al. (2001) Localized immediate early gene expression related to the strength of song learning in socially reared zebra finches. Eur. J. Neurosci. 13, 2165–2170
Behavioural predictions
Songbirds learn their songs and most non-songbirds do not. Similarly, usually male songbirds sing and females do not. The association of these differences with differences in anatomical structures raises the following questions: (1) Are songbirds superior to non-songbirds in auditory learning? (2) Are songbird males better auditory learners than females? The specific question that we are concerned with is whether non-songbirds and female songbirds are able to learn and store auditory information. Some non-songbirds such as parrots and hummingbirds can learn vocalizations, so vocal learning is not restricted to songbirds. Other non-songbirds such as doves and pigeons can learn to discriminate between different sounds64. There are songbirds species where both males and females sing63,65, and where both sexes can learn songs equally well65. Furthermore, females in some songbird species can be made to sing by administering testosterone to adults47,66, or oestrogen to hatchlings67. Females have been shown to sing songs they have learned from an adult conspecific, usually their male partner65–67. Furthermore, in some songbird species where the females do not sing, they do learn the characteristics of tutor songs, for which they show a preference in subsequent tests68,69. The strength of zebra finch female song preferences http://tics.trends.com
431
measured in operant tests is not different from those of their male counterparts70. Taken together, the evidence suggests that non-songbirds and female songbirds are capable of learning and storing auditory information in a similar fashion to songbird males. Causal links
As we have seen, the comparisons of songbird males with non-songbirds or females have yielded ambiguous results. A problem with such analyses is that there are many differences between males and females, and between songbirds and non-songbirds, apart from the fact that one sings and the other does not. Thus, it is difficult to draw conclusions as to whether brain nuclei that occur in songbirds and not in non-songbirds50, or that are larger in males than in females49, are involved in song learning or in some other behaviour that differs between these groups of birds. It is possible that differences between these groups that are in fact related to song learning occur in other regions of the brain that do not show between group differences in occurrence, size or cytoarchitecture. This is not to say that the neuroecological analyses of bird song have been in vain. It is likely (although not certain) that the brain regions that have been found in songbirds and not in non-songbirds (or that are larger in songbird males than in females) have something to do with vocal
432
Opinion
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
Questions for future research • Are there any examples of ‘cognitive modules’ – or is adaptive specialization limited to perceptual and motor mechanisms? • What is the neural substrate for spatial learning and memory in foodstoring birds? Is the same brain region involved in spatial learning and memory in non-storers? Furthermore, if hippocampal enlargement in food-storing birds is associated neither with memory nor with the motivation or capacity to store food, why does it occur? • Does auditory learning in female songbirds and in non-songbirds involve the same brain regions as those used by male songbirds? What is the neural substrate for the memory of the tutor song in male songbirds? • Are the areas involved in memory storage for a given input determined solely by the modality of the input, or is storage of some types of input of a given modality independent of storage of other types? For example, are areas that store spatial information about food sources distinct from areas that store spatial information about territorial boundaries? Do the areas that store learned songs differ from those that store other learned auditory input? • Is the capacity and longevity of memory for input of a given modality determined solely by such ‘low-level’ factors as salience, or are ‘higher’ factors also involved – factors associated with the nature of the central processing engaged by the input?
Acknowledgements We are grateful to Tom Smulders, Katharina Riebel and Gabriel Beckers for discussion, and to the anonymous referees for their constructive comments.
behaviour. But of course there is more to vocal behaviour than learning and the memorization of songs. Vocal behaviour also involves perception and production of songs, with all the complicated sub-processes that that entails. We have seen that the seasonal changes in size of the ‘song control nuclei’ are unlikely to be related to song learning, as was thought previously57, but might be involved in some aspects of song production. Brenowitz and colleagues have found that these seasonal changes are not related to the learning of new songs, song repertoire or song output58,59. They suggest that the seasonal changes might be related to song stereotypy: songs being more stereotyped (with less song-to-song variability) during the spring breeding season58. Similarly, it is not clear what aspect of song is related to (season invariant) within-species variation in volume of ‘song control nuclei’. Ward et al. have found significant positive correlations between number of song elements copied and HVC volume and neuron number71. These authors
References 1 Tinbergen, N. (1963) On aims and methods in ethology. Z. Tierpsychol. 20, 410–433 2 Bolhuis, J.J. (1999) The development of animal behaviour. From Lorenz to neural nets. Naturwissenschaften 86, 101–111 3 Hogan, J.A. (1994) The concept of cause in the study of behavior. In Causal Mechanisms of Behavioural Development (Hogan, J.A. and Bolhuis, J.J., eds), pp. 3–15, Cambridge University Press 4 Kamil, A.C. (1988) A synthetic approach to the study of animal intelligence. In Comparative Perspectives on Modern Psychology (Leger, D.W., ed.), pp. 230–257, University of Nebraska Press http://tics.trends.com
suggest that ‘naturally occurring variation in neuron number constrains how much song material can be copied or reproduced’. In zebra finch males, MacDougall-Shackleton et al. did not find correlations between the volumes of HVC, lateral part of the magnocellular nucleus of the neostriatum (lMAN), Area X or RA and a number of song parameters72. The only significant correlation was a negative one between lMAN volume and element repertoire size and the number of elements per phrase. On the basis of an evaluation of the literature, Gahr et al. conclude that ‘the songbird model therefore does not support the notion that sex differences in area size and neuron number explain sex differences in a behavior that occurs in both sexes’63. On the basis of an extensive review of the literature on sex differences in the size of the ‘song control nuclei’ in songbirds, MacDougall-Shackleton and Ball suggest that these differences are related to differences in song production49. Concluding remarks: the pros and cons of neuroecology
Our critique of the neuroecological approach should not detract from the undoubted merit of much of the work that has been carried out under this conceptual umbrella. For example, the comparative analysis of relative hippocampal size in food-storing birds and related species has provided us with important information regarding the evolution of brain and behaviour. In addition, the analysis of the ‘song control nuclei’ in songbirds has given us important insights into the neuroanatomy and neurophysiology of the songbird forebrain, and the circuits that play a role in vocalization. It is possible that these brain regions are involved in some way in the storage of vocal information, possibly of the BOS, in singing or in the processing of vocal feedback during song learning. Nevertheless, we maintain that evolutionary or functional considerations cannot explain the neural mechanisms of behaviour in general, and of learning and memory in particular. In addition, our review of the literature demonstrates that, although an application of such functional considerations to the mechanisms of learning and memory might give clues for research, it is often misleading and might provide us with the wrong answers to these important questions.
5 Shettleworth, S.J. (1998) Cognition, Evolution, and Behavior, Oxford University Press 6 Real, L.A. (1993) Toward a cognitive ecology. Trends Ecol. Evol. 8, 413–417 7 Healy, S. and Braithwaite, V. (2000) Cognitive ecology: a field of substance? Trends Ecol. Evol. 15, 22–26 8 Sherry, D.F. and Schacter, D. L. (1987) The evolution of multiple memory systems. Psychol. Rev. 94, 439–454 9 Cosmides, L. and Tooby, J. (1995) From function to structure: the role of evolutionary biology and computational theories in cognitive neuroscience. In The Cognitive Neurosciences (Gazzaniga, M., ed.), pp. 1199–1210, MIT Press
10 Fodor, J.A. (1983) The Modularity of Mind: An Essay on Faculty Psychology, MIT Press 11 Shettleworth, S.J. (1995) Comparative studies of memory in food storing birds: from the field to the Skinner box. In Behavioral Brain Research in Naturalistic and Semi-Naturalistic Settings (Alleva, E. et al., eds), pp. 159–194, Kluwer Academic 12 Clayton, N.S. and Krebs, J.R. (1995) Memory in food-storing birds: from behaviour to brain. Curr. Opin. Neurobiol. 5, 149–154 13 DeVoogd, T.J. and Székely, T. (1998) Causes of avian song: using neurobiology to integrate proximate and ultimate levels of analysis. In Animal Cognition in Nature (Balda, R.P. et al., eds), pp. 337–380, Academic Press
Opinion 14 Smulders, T. V. and DeVoogd, T. J. (2000) The avian hippocampal formation and memory for hoarded food: spatial learning out in the real world. In Brain, Perception, Memory. Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 127–148, Oxford University Press 15 Shettleworth, S.J. and Hampton, R.R. (1998) Adaptive specializations of spatial cognition in food-storing birds? Approaches to testing a comparative hypothesis. In Animal Cognition in Nature (Balda, R.P. et al., eds), pp. 65–98, Academic Press 16 Bolhuis, J.J. (2000) Introduction to Part two. In Brain, Perception, Memory. Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 87–91, Oxford University Press 17 Ewert, J.P. (1997) Neural correlates of key stimulus and releasing mechanism: a case study and two concepts. Trends Neurosci. 20, 332–339 18 Macphail, E.M. and Bolhuis, J.J. (2001) The evolution of intelligence: adaptive specialisations versus general process. Biol. Rev. 76, 341–364 19 Horn, G. (1998) Visual imprinting and the neural mechanisms of recognition memory. Trends Neurosci. 21, 300–305 20 Brennan, P.A. and Keverne, E.B. (2000) Neural mechanisms of olfactory recognition memory. In Brain, Perception, Memory. Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 93–112, Oxford University Press 21 Sherry, D.F. et al. (1981). Memory for the location of stored food in marsh tits. Anim. Behav. 29, 1260–1266 22 Balda, R.P. and Kamil, A.C. (1992) Long-term spatial memory in Clark’s Nutcracker, Nucifraga columbiana. Anim. Behav. 44, 761–769 23 Brodin, A. (1994) The disappearance of caches that have been stored by naturally foraging willow tits. Anim. Behav. 47, 730–732 24 Hitchcock, C.L. and Sherry, D.F. (1990) Long-term memory for cache sites in the black-capped chickadee. Anim. Behav. 40, 701–712 25 Healy, S.D. and Krebs, J.R. (1996) Food storing and the hippocampus in Paridae. Brain Behav. Evol. 47, 195–199 26 Sherry, D.F. and Vaccarino, A.L. (1989) Hippocampus and memory for food caches in Black-Capped Chickadees. Behav. Neurosci. 103, 308–318 27 Sherry, D.F. et al. (1989) The hippocampal complex of food-storing birds. Brain Behav. Evol. 34, 308–317 28 Sherry, D. F. (1998) The ecology and neurobiology of spatial memory. In Cognitive Ecology: The Evolutionary Ecology of Information Processing and Decision Making (Dukas, R., ed.), pp. 261–296, Chicago University Press 29 Krebs, J.R. et al. (1989) Hippocampal specialization of food-storing birds. Proc. Natl. Acad. Sci. U. S. A. 86, 1388–1392 30 Healy, S.D. and Krebs, J.R. (1992) Food storing and the hippocampus in corvids: amount and volume are correlated. Proc. R. Soc. London Ser. B 248, 241–245 31 Colombo, M., and Broadbent, N. (2000). Is the avian hippocampus a functional homologue of the mammalian hippocampus? Neurosci. Biobehav. Rev. 24, 465–484 32 Brown, M.W. (2000) Neuronal correlates of recognition memory. In Brain, Perception, Memory. Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 185–208, Oxford University Press http://tics.trends.com
TRENDS in Cognitive Sciences Vol.5 No.10 October 2001
33 Buckley, M.J. and Gaffan, D. (2000) The hippocampus, perirhinal cortex and memory in the monkey. In Brain, Perception, Memory. Advances in Cognitive Neuroscience (Bolhuis, J.J., ed.), pp. 279–298, Oxford University Press 34 Bolhuis, J.J. et al. (1994) Retrograde amnesia and memory reactivation in rats with ibotenate lesions to the hippocampus or subiculum. Q. J. Exp. Psychol. 47B, 129–150 35 Basil, J.A. et al. (1996) Differences in hippocampal volume among food storing corvids. Brain Behav. Evol. 47, 156–164 36 Volman, S.F. et al. (1997) Relative hippocampal volume in relation to food-storing behavior in four species of woodpeckers. Brain Behav. Evol. 49, 110–120 37 Clayton, N.S. et al. (1997). Seasonal changes of hippocampus volume in parasitic cowbirds. Behav. Processes 41, 237–243 38 Cassel, J.C. et al. (1998). Fimbria-fornix vs selective hippocampal lesions in rats: effects on locomotor activity and spatial learning and memory. Neurobiol. Learn. Mem. 69, 22–45 39 Hilton, S.C. and Krebs, J.K. (1990). Spatial memory of four species of Parus: performance in an open-field analogue of a radial maze. Q. J. Exp. Psychol. 43B, 345–368 40 Kamil, A.C. et al. (1994). Performance of 4 seedcaching corvid species in the radial-arm maze analog. J. Comp. Psychol. 108, 385–393 41 Balda, R.P. et al. (1997). Species differences in spatial memory performance on a threedimensional task. Ethology 103, 47–55 42 Gould-Beierle, K.L. (2000). A comparison of four corvid species in a working and reference memory task using a radial maze. J. Comp. Psychol. 114, 347–356 43 Hampton, R.R. and Shettleworth, S.J. (1996). Hippocampal lesions impair memory for location but not color in passerine birds. Behav. Neurosci. 110, 831–835 44 Hampton, R.R. and Shettleworth, S.J. (1996). Hippocampus and memory in a food-storing and in a nonstoring bird species. Behav. Neurosci. 110, 946–964 45 Nottebohm, F. et al. (1990) Song learning in birds: the relation between perception and production. Philos. Trans. R. Soc. London Ser. B 329, 115–124 46 Nottebohm, F. (2000) The anatomy and timing of vocal learning in birds. In The Design of Animal Communication (Hauser, M.D. and Konishi, M., eds), pp. 63–110, MIT Press 47 DeVoogd, T.J. (1994) The neural basis for the acquisition and production of bird song. In Causal Mechanisms of Behavioural Development. (Hogan, J.A. and Bolhuis, J.J., eds), pp. 49–81, Cambridge University Press 48 Solis, M.M. et al. (2000) Song selectivity and sensorimotor signals in vocal learning and production. Proc. Natl. Acad. Sci. U. S. A. 97, 11836–11842 49 MacDougall-Shackleton, S.A. and Ball, G.F. (1999) Comparative studies of sex differences in the song-control system of songbirds. Trends Neurosci. 22, 432–436 50 Gahr, M. (1997) Hormones make songs sexually attractive: hormone-dependent neural changes in the vocal control system of songbirds. Zoology 100, 260–272 51 Margoliash, D. (1986). Preference for autogenous song by auditory neurons in a song system nucleus of the white-crowned sparrow. J. Neurosci. 6, 1643–1661
433
52 Doupe, A.J. and Konishi, M. (1991). Songselective auditory circuits in the vocal control system of the zebra finch. Proc. Natl Acad. Sci. U. S. A. 88, 11339–11343 53 Volman, S.F. (1993) Development of neural selectivity for birdsong during vocal learning. J. Neurosci. 13, 4737–4747 54 Jarvis, E.D. and Nottebohm, F. (1997) Motordriven gene expression. Proc. Natl. Acad. Sci. U. S. A. 94, 4097–4102 55 Bolhuis, J.J. et al. (2000) Localized neuronal activation in the zebra finch brain is related to the strength of song learning. Proc. Natl. Acad. Sci. U. S. A. 97, 2282–2285 56 Mello, C. V. et al. (1992) Song presentation induces gene expression in the songbird forebrain. Proc. Natl. Acad. Sci. U. S. A. 89, 6818–6822 57 Nottebohm, F. (1981) A brain for all seasons: cyclical anatomical changes in song control nuclei of the canary brain. Science 214, 1368–1370 58 Tramontin, A.D. and Brenowitz, E. (2000) Seasonal plasticity in the adult brain. Trends Neurosci. 23, 251–258 59 Brenowitz, E.A. et al. (1991) Seasonal changes in avian song nuclei without seasonal changes in song repertoire. J. Neurosci. 11, 1367–1374 60 Brenowitz, E.A. et al. (1995) Brain space for learned song in birds develops independently of song learning. J. Neurosci. 15, 6281–6286 61 Kirn, J.R. et al. (1989) Song-related brain regions in the red-winged blackbird are affected by sex and season but not repertoire size. J. Neurobiol. 20, 139–163 62 Deviche, P. and Gulledge, C.C. (2000) Vocal control region sizes of an adult female songbird change seasonally in the absence of detectable circulating testosterone concentrations. J. Neurobiol. 42, 202–211 63 Gahr, M. et al. (1998) Sex difference in the size of the neural song control regions in a dueting songbird with similar song repertoire size of males and females. J. Neurosci. 18, 1124–1131 64 Cynx, J. (1995) Similarities in absolute and relative pitch perception in songbirds (starling and zebra finch) and a nonsongbird (pigeon). J. Comp. Psychol. 109, 261–267 65 Baptista, L.F. et al. (1993) Singing and its functions in female white-crowned sparrows. Anim. Behav. 46, 511–524 66 Chilton, G. and Lein, M.R. (1996) Songs and sexual responses of female white-crowned sparrows (Zonotrichia leucophrys) from a mixeddialect population. Behaviour 133, 173–198 67 Simpson, H.B. and Vicario, D.S. (1991) Early estrogen-treatment alone causes female zebra finches to produce learned, male-like vocalizations. J. Neurobiol. 22, 755–776 68 Brenowitz, E. A. (1986) Altered perception of species-specific song by female birds after lesions of a forebrain nucleus. Science 251, 303–305 69 Riebel, K. (2000) Early exposure leads to repeatable preferences for male song in female zebra finches. Proc. R. Soc. London Ser. B. 267, 2553–2558 70 Riebel, K. et al. (2001) Female and male zebra finch siblings do not differ in their adult preferences for the father’s song. Ethology (Suppl. 36), 251 71 Ward, B.C. et al. (1998) Individual variation in neuron number predicts differences in the propensity for avian vocal imitation. Proc. Natl. Acad. Sci. U. S. A. 95, 1277–1282 72 MacDougall-Shackleton, S.A. et al. (1998) Neural correlates of singing behavior in male zebra finches (Taeniopygia guttata). J. Neurobiol. 36, 421–430