1
Cognitive Ethology and Philosophy of Mind: An Interdisciplinary Approach

If no organic being excepting man had possessed any mental power, or if his powers had been of a wholly different nature from those of the lower animals, then we should never have been able to convince ourselves that our high faculties had been gradually developed. But it can be shewn that there is no fundamental difference of this kind. We must also admit that there is a much wider interval in mental power between one of the lowest fishes, as a lamprey or lancelet, and one of the higher apes, than between an ape and a man; yet this interval is filled up by numberless gradations. --Charles Darwin (1871, p. 445) Only the most benighted of evolutionary gradualists could be sanguine that the apparently radical intellectual discontinuity between us and other creatures will prove to be merely quantitative. --Jerry Fodor (1994 p. 91)

How widely are mental phenomena distributed in nature? Are the capacities for thought and feeling unique traits of humans, or are these and other mental states also found in nonhuman animals? If they are found in nonhuman animals, how similar are the mental states of those animals to the mental states of humans? Are human mental states different in kind, or are there "numberless gradations" filling the interval between humans and lampreys (as Darwin hypothesized)? Answers to questions such as these about continuity between the minds of humans and those of other animals are central for understanding both the evolution and the nature of mental capacities. But many scientists have been reluctant, for various reasons, to take seriously the attribution of mental states to nonhuman animals. Unless this reluctance can be overcome, it will not be possible to assess Darwin´s gradualist hypothesis scientifically.
Many nonscientists find the reluctance of scientists to admit the existence of nonhuman mentality to be contrary to common sense. Common sense does not always, however, provide clear intuitions about the mentality of nonhuman animals. On the one hand, it is normal to talk about companion animals and other familiar animals in mentalistic terms: it is not uncommon to hear of cats who are angry when their owners return after a weekend away, or of dogs who think they are going for a walk. Starting as young children, humans attribute mental characteristics to a variety of objects, both animate and inanimate. But these common-sense, "folk-psychological" judgments are typically taken more seriously when applied to animals, and as such they seem to support the view that mental states are not uniquely human traits. On the other hand, it is also commonly recognized that there are large gaps between the mental abilities of humans and those of other animals; for example, many people talk to their pets but do not expect them to understand much of what is said. The existence of these gaps raises questions about the extent to which other animals really are like us. And it is common sense to many people that the psychological dissimilarities of other animals from humans are sufficient to permit farming, hunting, various types of research, and other practices that would be considered unacceptable if applied to humans. Furthermore, unmitigated appeals to common sense are unlikely to be effective in changing the attitudes of scientists toward animal mentality. Science has a history of overturning the opinions of common sense.
From a psychological or an epistemological point of view, differences between humans and other animals make it difficult for us to imagine or know what the subjective experiences of other organisms might be like. From an evolutionary point of view, there are difficult questions about how such gaps could have arisen by natural selection. The task of bridging these gaps and understanding the bases for mental-state attributions is at the same time a philosophical project and a scientific project. It is a philosophical project because it requires philosophical investigation of mentalistic concepts and of the aims and methods of cognitive ethology; it is a scientific project because impoverished knowledge of animal behavior results in impoverished arguments and faulty conclusions. Cognitive ethology provides a tractable approach to obtaining such knowledge.
Our approach is to assume that at least some organisms (minimally, humans) do have mental lives and that the best way to understand mental-state attributions across species boundaries is within the comparative, evolutionary, and interdisciplinary framework provided by cognitive ethology. A goal of this book is to make that framework as explicit as possible. Where there are shortcomings in our account, we hope to convince our readers that the difficulties are tractable within the interdisciplinary approach. By assuming a realist attitude toward mental states, we choose to set aside the worries of "eliminativists," who believe that talk of minds is hopelessly confused and will be dispensed with in the future. Although ordinary mentalistic talk may be confused, we will work with the assumption that it is not hopelessly confused. Of course, much ordinary mentalistic language is bound up in cultural practices, such as determining innocence, guilt, or responsibility for various actions. For this reason, we also accept that it may be necessary to reconceptualize much of folk psychology if we are ever to have a clear view of the cognitive similarities among different species.
In the course of the book we shall address a variety of arguments that have been put forward to establish the hopelessness, misguidedness, or sheer impossibility of studying the minds of animals. All these arguments--even those given by scientists--have something of a philosophical character, but they differ with respect to the degree of attention they pay to empirical research on animal behavior. Not so long ago, undergraduate philosophy majors were taught that if a question had an empirical component it was not a philosophical question. Many science undergraduates were also taught that if a question could not be directly answered empirically then it was of no concern to scientists. Although these attitudes persist in some quarters, there is now much greater understanding of relationships between theory and evidence, especially the fact that theories are always underdetermined by evidence because there are many theories that are compatible with the same set of evidence (Quine 1953). There is consequently more respect for interdisciplinary efforts such as the one undertaken here. However, there is still much to be learned about the relationships between theory-motivated philosophical arguments and empirical work. In this book, we hope to contribute by example to the understanding of what can be achieved by direct collaboration between scientists and philosophers.
Although Darwin proposed his ideas about mental continuity more than 120 years ago, the dominant present-day view among animal-behavior researchers is that such issues are beyond the pale of respectable scientific research. Related developments in both philosophy and psychology have much to do with this prevailing attitude. In philosophy, empiricism is the a class of views that maintain that all human knowledge must ultimately be derived from sensory experience. In the early twentieth century empiricism reached its zenith in the movement known as logical empiricism (or logical positivism) which emphasized that the meaningfulness of any concept depended on its reducibility to logical constructions from observable, verifiable experiences. In psychology, the behaviorists implemented the positivist program with a strict operationalism that allowed the use of mentalistic terminology only insofar as individual terms could be "defined" strictly using observable relationships between stimuli and responses. These "definitions" were not intended to capture the ordinary meanings of terms, but to replace the ordinary terms with a "more scientific" vocabulary.
Donald Griffin, in a series of books and articles dating back to 1976, has urged that the comparative, evolutionary study of animal behavior cannot be completed if issues of animal mind are ignored, and has urged the development of the field for which he invented the label "cognitive ethology." Griffin´s writings have been attacked as anecdotal and anthropomorphic, as bad science, and as just plain muddled thinking. Some of the critics of his work in particular and of cognitive ethology more generally bring behavioristic presuppositions to their arguments; thus, it is our view that, in defending cognitive ethology, it is not feasible to ignore the behavioristic challenge.
Cognitive ethology, however, requires more than a convincing response to the behaviorists. Behaviorism is no longer the force that it once was, even within comparative psychology. But even those who are sympathetic to cognitive ethology worry that, hard as it is to make progress on difficult issues of mind when the subjects are human, it is many times harder when one is working with nonhuman animals, with whom linguistic communication is either impossible or highly limited. Although human cognitive psychology relies very little on verbal introspective reports by research subjects, the ability of subjects to respond to complex verbal instructions is nonetheless essential to the design of many experiments. Cognitive ethology will be shown viable only if it can be shown that a mentalistic approach to the study of animal behavior is capable of sustaining an empirical research program despite the challenges presented. A major goal of this book is to indicate how such a research program might be sustained.

Naturalism about the Mind

In contributing to the development of cognitive ethology, we believe ourselves to be contributing to the philosophical project of naturalizing the mind. Naturalism is a broad philosophical stance that denies either the existence of supernatural or nonphysical entities or their relevance for understanding any given phenomenon. Another major strand in philosophical naturalism (see Quine 1953) is that the methods of science and philosophy are intertwined. When naturalism is concerned with what kinds of entities exist, it is part of the philosophical subdiscipline of ontology--the theory of existence. ("Ontology" should not be confused with "ontogeny," the development of individual organisms.) When naturalism is concerned with how a phenomenon can be understood, then it is part of the philosophical subdiscipline of epistemology--the theory of knowledge.
The phrase "naturalizing the mind" has a variety of meanings to different authors. We take the central idea behind naturalism about the mind to be an ontological view to the effect that mental phenomena are in some sense a part of the physical world. Naturalism in this sense is opposed to the Cartesian view that mind is substantially distinct from the rest of nature, and it is generally opposed to theories of mind that link mental properties with notions such as the possession of an immaterial soul or spirit. (See Shapiro 1997 and the discussion below for a disagreement with this characterization of naturalism.)
Of what consequence is this issue about naturalism for cognitive scientists in general, and cognitive ethologists specifically? One can do good work on memory or planning without worrying about the philosophical underpinnings of these concepts, just as one may do excellent experimental work in particle physics that is not affected by one´s opinion about the relative merits of the "Copenhagen" and "many-worlds" interpretations of quantum mechanics. But whether or not a particular scientist is personally attracted to such issues, the discipline as a whole is affected by them. Theories are accepted both for their ability to account for observations and for their coherence with the rest of science. In this vein, the notion of a mental representation is puzzling because mental representations have features that are hard to account for in ordinary causal terms. Consider planning, which requires the ability to represent future events that have not yet occurred and may never occur. Clearly such events cannot be causes of their representations. So how is it possible to understand representation of future goals from within a causal perspective? This is one aspect of the problem of naturalizing representation, which if not resolved leaves cognitive scientists with an unpaid debt no matter how empirically successful their theories are (von Eckardt 1993, especially p. 197).
In order to assess the prospects for naturalism about the mind, it is necessary to get clear about what would satisfy the desire for a naturalistic theory.

Three Forms of Reductionism

Opinions vary on what must be done to establish naturalism with respect to mental phenomena. We ultimately reject the suggestion that it means showing how mentalistic terms or predicates (the language of mentalistic psychology) can be reduced to nonmental predicates. According to the strongest versions of reductionism, the successful reduction of a mentalistic term, M, requires a definitional analysis of M in nonmentalistic terms. This would be achieved if one could complete the schema "x has mental property M if and only if x is F" in such a way that F provides a definition of M. Under a very strong conception of definition, the defined term (the definiendum) and the defining phrase (the definiens) must be completely synonymous. But this very strong requirement is extremely unlikely to be met by any suggested definition. For example, it is extremely unlikely that one could find a statement about neural states with exactly the same meaning as a statement about mental states that it is intended to analyze. For this reason, most naturalists do not believe that it is necessary to provide this kind of analysis to support naturalism.
A weaker version of reductionism requires the nondefinitional analysis of mentalistic terms or predicates. That is, one must specify necessary and sufficient conditions for the application of a mentalistic term M in nonmentalistic, naturalistic terms, but the analysandum (the expression being analyzed) need not be equivalent to the analysans (the expression providing the analysis). Thus, it is still required that the schema "x has mental property M if and only if x is F" be completed, but the stronger requirement of synonymy between M and F is dropped. For example, it has often been suggested that possession of a particular brain or neural state is both necessary and sufficient for possession of a particular mental state. The standard philosophical example, almost certainly wrong, is that x is in pain if and only if x´s C fibers are firing. If successful, this analysis would provide identity conditions for pain--the mental state of being in pain would be identified with (or reduced to) the neural state of C fibers´ firing--even though, for linguistic and historical reasons, no one would suppose that "John is in pain" and "John´s C fibers are firing" had the same meanings. Unfortunately for those who favor this version of reductionism, it has been remarkably difficult to provide successful analyses (Stich 1992; Tye 1992). Consequently, many naturalists have favored weakening the requirements for naturalism still further.
One obvious move is to drop the joint requirement for necessary and sufficient conditions in favor of an approach that allows naturalists to get by with specifying, in naturalistically acceptable terms, merely sufficient or merely necessary conditions for the application of mentalistic terms. If successful in doing this, a naturalist would have managed to establish connections that, although weaker than equivalence or identity, would still support the naturalistic claim that mental phenomena are not inherently supernatural. For example, if C-fiber stimulation turned out to be sufficient (in the context of a functioning nervous system) although not necessary for pain, then it would be known that at least some cases of pain do not depend on anything supernatural. In such a case, the notion of pain would not have been completely reduced (other cases of pain might occur without C fibers´ firing), but the partial reduction would nonetheless provide some support to naturalism about pain.

Emergent Properties and Reductionism

So much for the cold touch of technical philosophy. Much of the broader scientific debate about the status of mental properties is conducted in terms of the notion of an emergent property. One is told that consciousness and other mental properties "emerge" in an unpredictable way from the actions of neurons. Such claims are sometimes taken to have antireductionistic consequences. But is emergence an antireductionistic notion? With respect to ontology, we believe that the correct answer to this question is negative. An emergent property is, very roughly, a property that belongs to an aggregate of entities but is not a linear sum of the properties of the parts. This means that the mass of a large object would not count as an emergent property of the object, as the total mass is simply the sum of the masses of all its parts.
Uncontroversial examples of emergent properties are harder to find. The solvent property of water is one candidate, for neither hydrogen atoms nor oxygen atoms in isolation possess this property and neither do they possess scaled-down versions of the property. Solvent action seems to emerge from a nonlinear combination of the properties of hydrogen and oxygen. One can, of course, produce a solvent using nothing more than hydrogen atoms and oxygen atoms in the ratio 2:1, so from an ontological standpoint nothing more is required. In this sense, chemical combination of these atoms is sufficient for producing a solvent. Thus, a naturalistic stance toward the solvent property of water is justified.
The nonlinear relationship between the properties of hydrogen and oxygen and the properties of water (if typical for emergent properties) might be taken to entail that the doctrine of emergent properties supports an epistemological version of antireductionism. If one could not infer the properties of the brain from the properties of its neurons and glial cells, then, it would seem that one could not fully understand a reductionistic explanation of the brain´s properties in terms of the properties of its cellular parts. Connections between reduction and understanding often lead philosophers of science to characterize reductionism as having an epistemological component, because epistemology is concerned with questions of knowledge and understanding. We prefer to avoid this characterization of reductionism, however, for it depends on rather vague claims about what one could or could not understand on the basis of given information. Who is the "one" whose understanding is at issue here? Trying to take a God´s-eye view is fruitless. From the point of view of a mere mortal, what one may or may not understand can change. Though it was puzzling in the past, scientists now have a pretty good understanding of how the polarity of a water molecule accounts for water´s ability to act as a solvent, and of how that polarity is due to the electron structure of a water molecule´s component atoms. Thus, what seems irreducible today may not seem irreducible tomorrow, and it would be shaky to pin one´s views about the understandability of mental properties on the limitations of current theories.
The common thread in all the varieties of ontological reductionism is the idea that it should be possible to show how mental phenomena arise from or are constituted by suitable arrangements of the basic materials that make up the rest of the natural world. Because science purports to tell us about the natural world, naturalistically inclined philosophers commonly turn to science as the source for necessary or sufficient conditions relevant to mental phenomena; computer science, cognitive psychology, the neurosciences, evolutionary biology, and quantum mechanics have all been enlisted for support. Stich (1992) and Tye (1992) both suggest that such appeals to science have been unsuccessful in supporting reductionism as a viable approach to naturalism about mental phenomena. We do not wish to go into the details of their arguments, but their general point is easy to state: Despite a long history of attempts to provide reductionistic accounts of mental phenomena, there is not even one case of a widely accepted partial reduction--for example, of a neurally specific sufficient condition for a specific mental property. Although we think ontological reduction is a worthy goal, in view of this miserable history it seems that naturalists would be well advised to consider alternative approaches.

Alternative Approaches to Naturalism

Tye (1992, p. 436) states his own version of naturalism as follows: "Mental states participate in causal interactions which fall under scientific laws, and are either ultimately constituted or ultimately realized by microphysical phenomena." This view has two components. The second part--concern for the ontological issue of what mental states are "constituted" or "realized" by--is a feature (shared by most reductionistic approaches) that, in accordance with Shapiro (1997), we will label ontological naturalism. Shapiro also labels such views, somewhat disparagingly, as "Lego naturalism," because of their concern with the nonmentalistic building blocks of mental phenomena. Tye´s version of ontological naturalism differs from others by being less optimistic about the likelihood of providing analyses of mental terms in terms of either necessary or sufficient conditions. The first part of Tye´s view--insistence on the place of mental states in law-governed causal interactions--has much in common with Shapiro´s (1997) methodological naturalism, according to which naturalism about mental states requires only that there be a productive, systematic, empirically tractable theory that includes mental-state predicates within its theoretical vocabulary.
Shapiro´s view lacks the ontological component of Tye´s. For Shapiro, it doesn´t matter what mental states are made of so long as they can be studied by methods that are acceptable to scientists. Indeed, whereas Tye´s view combines ontological and methodological components, Shapiro explicitly rejects an ontological component to his naturalism; he thinks that the ontological approach is unable to make the distinction that, on his view, really matters to naturalism. Shapiro believes that naturalists are not really concerned with whether mental phenomena are natural or supernatural. Rather, he writes (1997, p. 11), "we should expect of a naturalistic thesis that it allows us to distinguish scientific kinds and properties from nonscientific kinds and properties."
We agree with Shapiro that methodological unity between the study of mental phenomena and other natural sciences would provide support for naturalism. We disagree, however, with Shapiro´s rejection of ontological concerns. On our view, both the ontological and the methodological strand of naturalism provide useful characterizations of criteria that might be satisfied by a naturalistic theory of mind. In our view it is premature to decide now what is likely to be the best approach to naturalizing the mind. Neither ontological nor methodological approaches to naturalism require reductionism (although they are compatible with it), and the fact that there are alternative approaches to naturalizing the mind is valuable. There are various ways in which mental phenomena might be assimilated to other natural phenomena, and naturalism gains support whether that assimilation is reductionistic, ontological, or methodological, or if it takes some other form that we have yet failed to consider.
For scientists, most philosophical attempts to characterize naturalism do not provide adequate advice on how to achieve the goal of a naturalistic theory of mind (other than "Keep doing whatever you are doing"). In general, knowing what would count as satisfying some goal is not the same thing as knowing how to achieve it--one might well know that one would be rich if one had a million dollars yet have no clue how to make that much money. Likewise, many philosophical views about naturalism purport to specify what scientists are or should be aiming at but do not give much help with the practicalities of taking aim. It is fair to respond that most philosophical discussions do not aim to provide practical suggestions for scientific research. But the naturalist´s cause could be advanced if such suggestions were forthcoming. It is not entirely surprising that suggestions are infrequent. Philosophers often do not know enough about the relevant sciences to be able to make practical suggestions and the task is difficult even for those who are relatively knowledgeable. For example, after visiting the Kenyan research site of Dorothy Cheney and Robert Seyfarth (Cheney and Seyfarth 1990), Daniel Dennett (1987) admitted that his earlier methodological suggestions for cognitive ethology (1983) did not take account of the complexities and difficulties presented by ethological fieldwork. Similarly when one of us (C.A.) was given the opportunity to conduct field studies of bird behavior (described in chapter 7) with the other (M.B.), the sheer practical difficulties of implementing certain research ideas meant that many ideas were shelved. Conversely, scientists often do not know enough about the intricacies of philosophical theories of mind to be able to identify the ways in which their work is relevant to those theories. In view of the difficulties for any individual of becoming completely conversant in two fields, a collaborative, interdisciplinary approach seems necessary.
By pushing cognitive ethology as an approach to naturalizing the mind, we do not suggest that it is the only possible approach, or even that it is the only approach that is likely to succeed (and perhaps we should not suggest that it is likely to succeed at all). Rather, we intend this book as an extended investigation of the prospects for understanding naturalism about mind within an ethological framework. In particular, we shall attempt to show that achieving a thorough understanding of the evolution and the phylogenetic distribution of mental phenomena might facilitate the assimilation of mental phenomena to other natural phenomena. Although we shall often be somewhat critical of those who insist that only in the laboratory can one conduct properly controlled experiments, this is not to say that we reject the relevance of laboratory investigations of animal behavior; it is only to say that they alone cannot provide a complete picture of the comparative and evolutionary aspects of animal cognition.

Cognitive Ethology and Naturalism

The idea that mental phenomena are found in nonhuman organisms is not essential to a naturalist position with regard to mind; one might, for example, identify mental properties with a level of computational complexity not found in nonhumans. Nevertheless, skepticism about animal minds is one of a number of pieces of the Cartesian legacy whose defeat would go some way toward vindicating naturalism. In this context, the Darwinian idea of mental continuity between the species provides a framework for constructing a naturalistic view of mind.
Using Darwin as his guide, Griffin has argued persistently for the claim that ethological observations of animals support attributions to them of thought, consciousness, and other mental states. Griffin does not, however, provide a clear account of the grounds for attributing mental states on the basis of behavioral observations other than in his metaphor that communication provides a window on animal minds. For this he has been criticized both by those who are sympathetic toward cognitive ethology (including Allen and Hauser (1993), Jamieson and Bekoff (1993), and Bekoff and Allen (1996)) and by those who are not sympathetic (e.g., Heyes (1987a,b)). A number of unsympathetic critics, including Premack (1988) and Heyes and Dickinson (1990), have suggested that ethology, which relies heavily on observing animals in their natural habitats, simply cannot support the kinds of mental-state attributions that interest cognitive ethologists, and that progress on issues of comparative cognition can be made only under laboratory conditions. We favor the more pluralistic view that both laboratory work and fieldwork are important to the study of comparative cognition, but fieldwork is essential to the proper interpretation of the results of laboratory experimentation. In chapter 9 we will illustrate this point with respect to the argument of Heyes and Dickinson. Here we will provide a more general argument for the centrality of ethology in studies of comparative cognition.
Natural selection generally acts on the functional properties (in the sense of Cummins 1975) of organismic traits; the material properties of an organism or its traits are important only insofar as they affect the functional capacities of those traits. Nervous systems of animals are no exceptions--the functions for which they are selected include the control of behavior. Nervous systems are also, at least in humans, the organ of the mind. Thus, in trying to understand the evolution of mentality it is reasonable to consider the evolution of nervous systems. The branch of ethology specifically concerned with relationships between behavior and neurobiology is known as neuroethology. If selection acts on traits in virtue of their functional characteristics, then it acts on nervous systems in virtue of the behavioral functions they support. Thus, in order to understand nervous-system function from an evolutionary perspective, it is essential to understand the functional aspects of behavior. This is one of the tasks classically taken to define ethology; therefore, ethology is a cornerstone of any attempt to understand the evolution of mentality. Ethologists study animal behavior from a variety of perspectives, including the examination of relationships between behavioral phenotypes and selective pressures. Thus, ethologists favor observations and experiments on animals under conditions that are as close as possible to the environments in which selection occurs--that is, they favor fieldwork.

Theories of Mind: A Pluralistic Approach

No one knows whether mentalistic terms provide the right vocabulary for cognitive ethology, or whether evolutionary accounts of behavior provide the key to a naturalistic account of mentality. Potentially, philosophers have as much to learn from ethologists as vice versa. Philosophical theories of mind, insofar as they are empirically tractable, can provide suggestions for empirical investigation. In return, ethological research into cognition provides data points for the refinement of philosophical theories--for example, with respect to the importance or unimportance of language for mentality. Much of the rest of this book will explore specific examples of this two-way interaction.
Our approach is pluralistic. Owing to the complexities involved, the more points of departure available the greater will be the chances of success. Thus, we are less concerned with producing a complete theory of mind than with showing how different theories of mind have different consequences for making the scientific study of animal mind empirically tractable.
Two aspects of mentality have been the major targets of contemporary philosophical theorizing. Both are important to cognitive ethologists. One is consciousness, particularly in the sense that some mental states feel like something to their possessors. A puzzle that worries many philosophers is how the machinations of neurons could add up to such feelings. Some philosophers, including Nagel (1974), put this in terms of wonder at how it could be that events in the cortex could give rise to the subjective quality of experience. Others, such as Jackson (1986), put the point in terms of the existence of knowledge that is not available from a purely physical (or neurological) description. Clearly, anyone who wonders about the experiences of nonhuman animals must hope to find some philosophical clarity on the notion of consciousness.
The other major target of contemporary philosophers of mind is the intentionality of mental states. Owing to the nineteenth-century psychologist Franz Brentano (1874), the use of the term "intentional" and its cognates by philosophers has a special sense that differs from the ordinary sense of the term. In ordinary language the term "intentional" can be used as a synonym for "purposeful," but in the philosophy of mind it has come to have a broader but more technical meaning. Although there is disagreement about exactly how the term should be defined, the general idea it is intended to capture is that mental states have semantic or representational content, sometimes described by saying that they represent (or are "directed toward," or "about") other states of affairs.
Intentional notions, in this sense, appear in several topics that are of concern to cognitive ethologists. For instance, an ethologist who considers the possibility of planning in nonhuman animals is wondering about an intentional notion, for a plan involves the representation of future actions. Many of the terms that appear in folk psychology are also intentional in Brentano´s sense. For example, beliefs and desires are intentional, for to have a belief one must have a belief about something and to have a desire one must have a desire for something. Intentionality, in Brentano´s sense, is not limited to the so-called propositional attitudes, such as belief and desire, found within folk psychology. Mental representation, information, and other notions widely used by cognitive scientists are also intentional in the sense that one cannot have information without its being information about something and one cannot have a representation without its being a representation of something.
Why is intentionality puzzling? One reason is that it appears to fall outside the usual causal order of the world. Consider an organism that is capable of formulating a plan. The plan represents future actions, yet these are actions that have not yet occurred and can therefore not be causes of that plan or of its content. Thus, although it might be tempting to think that a belief is about a leopard because it was caused by a leopard, this cannot be a completely general account of intentionality.
Within recent philosophy of mind it has been common to follow a divide-and-conquer strategy of treating consciousness and intentionality independently. Not all philosophers who agree that it is appropriate to consider intentionality independent of consciousness. Searle (1992) is among those who disagree, on the grounds that genuine intentionality requires the subject´s awareness of the contents of his or her own mental states. We shall address this concern in chapter 8.
During the last quarter of the twentieth century, philosophers have generally moved away from attempts to provide language-based criteria for understanding mental phenomena and toward approaches that are more grounded in scientific practice. A number of these approaches will be covered in later chapters. For the moment, we wish to illustrate some consequences of this trend by briefly introducing the work of Ruth Millikan (1984), which is of special interest to ethologists because it attempts to provide an evolutionary analysis of intentionality that is divorced from the notion of consciousness. According to Millikan, intentionality is a property derived from the biological or "proper" functions of those things that possess it. Millikan uses bee dances as examples of what she calls "intentional icons." According to her account, bee dances are about the location of nectar because (presumably) the ancestors of current bees were able to pass on the behavioral trait of dancing to current bees as a consequence of the selective advantage afforded to those who exploited correlations between features of dances and the locations of nectar.
Millikan´s account is historical in the sense that a thing´s intentionality depends not on its present characteristics but on its being the product of a selective process that allows us to say what the thing is for. For example, her theory leads her to say (1984, p. 93) that an exact duplicate of a human being produced by a random process (e.g., an extremely unlikely quantum accident), although it might be conscious or have other mental states, would not have any intentional states (beliefs, desires, etc.), since (initially at least) the creature´s brain states would lack the right selectional history to explain their existence. It is also a consequence of Millikan´s theory that certain features of the behavior of plants exhibit intentionality. These consequences of her theory might at first seem to make it of questionable relevance to an investigation of the nature of minds. We shall return to this issue; here we only want to point out the extent to which the notion of intentionality has been divorced from the notion of consciousness. This is not to the liking of some philosophers of mind--especially John Searle (1992)--who believe that the separation of issues of consciousness from issues of intentionality is symptomatic of what is wrong with current philosophy of mind.

Empirical Approaches to Intentionality

Different notions of intentionality have the potential to cause confusion about the role of intentional terms in the description and explanation of animal behavior. But rather than see disagreement about the correct account of intentionality as a problem for cognitive science, we see it as an opportunity for developing an empirical account of intentionality. This attitude may seem to present a problem. We are a long way from being able to give an uncontroversial definition of intentionality, but many behavioral scientists believe it is not possible to study a phenomenon without a rigorous (preferably operational) definition of that phenomenon. That this idea is false should be obvious from the early investigations of the chemical natures of gold, carbon, and other elements. Before gold´s atomic structure was understood, overt properties such as density, hardness, color, and reactivity were used to determine whether a given specimen was indeed gold. It would have been premature to define gold in terms of those properties, since, like carbon, gold could have turned out to occur in more than one form. A precise definition of "gold" formulated before comparative work had been done on numerous putative examples of gold would have begged certain questions about the nature of gold, since by definition things that shared the overt properties would have been gold and things that lacked the properties would not have been gold. Rough characterizations in terms of overt properties provide an initial classificatory scheme which is then revised by careful comparative work (such as the work that led to revisions in the concept of gold to include ideas about atomic structure within an empirically productive theoretical framework). (See Kripke 1972 for a general account of scientific terms on which these considerations are based; see Crick 1994 for an application to the notion of consciousness.)
The motivation for a comparative approach to the study of intentionality should now be clear. The empirical utility of a notion of intentionality will depend on whether it can be fit into an appropriate theoretical framework. This cannot be decided a priori by philosophers any more than philosophers could have decided whether gold, carbon, etc. was a better classification scheme than earth, air, fire, water. Cognitive scientists would be ill-advised to look to philosophers for a crisp and empirically rigorous definition of intentionality (even if some philosophers promise to provide it). Philosophical conceptions of intentionality distinguish a certain class of phenomena from others. Given a particular classification scheme based on a particular philosophical conception, further investigation may show whether there is a scientifically useful theoretical basis for including all the phenomena initially characterized in this way. Phenomena initially included may come to be dropped from the categorization scheme, and some phenomena initially omitted may be usefully included. Or the phenomena picked out by the philosophical categories may turn out to be so heterogeneous that no useful theory can be built around them. From this perspective, the variety of philosophical views about intentionality is a good thing insofar as they suggest different bases for comparative studies.
The question whether to treat intentionality as a property of minds, as a property of sentences, as an aspect of biological function, or in some other way takes on a different significance from this perspective. Brentano, Millikan, and other authors of theories of intentionality provide criteria for distinguishing some phenomena (intentional ones) from others (non-intentional ones). Some of these criteria are more easily applied than others. The resulting categorization schemes may not agree in all cases, but they may provide equally useful starting points for more detailed comparisons of the phenomena. The results of comparative work may lead to refinements in the notion of intentionality, or to its abandonment, but this cannot be predicted reliably when the empirical work has not been done. The main point here, though, is that choosing to start from a particular categorization scheme does not commit one to accepting that it is the correct scheme. Indeed, investigation of conflicting categorization schemes might even hasten convergence on a more useful scheme.

Ethology, Intentionality, and Consciousness

Griffin has placed the issue of animal consciousness firmly in the center of cognitive ethology. Somewhat ironically, those philosophers who have turned their attention to cognitive ethology have had rather little to say about consciousness, and rather more to say about intentionality in Brentano´s sense and about the associated phenomena of representation and meaning. Because "intentionality" in this sense is a technical term within philosophy, it has received comparatively little attention from ethologists, although there are notable exceptions, including Cheney and Seyfarth (1990) and Beer (1991) (also see Bekoff and Allen 1992). It is also the case that although not explicitly discussed in these terms, much of the current research on deception and self-recognition is about intentionality (see Byrne 1995).
We believe that to elucidate the relationship between consciousness and intentionality is the most difficult task facing all existing attempts to naturalize the mind. It is a task that we do not expect to accomplish fully within these pages. However, we will try to indicate how a comparative, ethological approach may have something to contribute to the project.
Many recent philosophical discussions of consciousness have stressed the heterogeneity of the notion. For some, including Wilkes (1984, 1995), this has called into question its scientific utility; for others, including Dennett (1991), Flanagan (1992) and Nelkin (1993), it suggests that empirical progress may be possible if one is careful to treat different aspects of consciousness independently. Most recent discussions of intentionality specifically dissociate it from consciousness, as we have already indicated with respect to Millikan´s account.
According to Searle (1980, p. 454), intentionality requires "some awareness of the causal relation between the symbol and the referent." This appeal to awareness requires further elucidation. Searle´s claim that this awareness probably arises from the biochemical properties of nervous systems does not help to elucidate the idea. He draws an analogy to the biochemistry of digestion, but because digestion is identified as the mechanical and chemical breakdown of ingested substances it is clear why biochemistry is important to digestion. It is less clear why it should be important to consciousness, which is not identified in chemical terms. We shall argue (in chapter 8) that a comparative approach with cognitive ethology as a major component has the potential to demystify the notion of awareness to which Searle appeals. We believe that progress is possible by paying attention to the cognitive ability to detect perceptual errors.

Concluding Remarks

Our aim in this book is to promote an interdisciplinary approach to theories of mind. The perspective should be broadly naturalistic, although there are various senses of naturalism that might prove satisfactory and deserve independent investigation. We see Darwinian continuity as one plausible route to a naturalistic theory of mental phenomena and ethology, and cognitive ethology as essential to the pursuit of this route. Our project, then, is to explore how evolutionary accounts of mental phenomena can inform and be informed by philosophical accounts. Wherever possible, we will concentrate on available data or suggest studies that will promote understanding of animal minds.