Animal Cognition: Theory and Evidence
Review of Species of Mind: The Philosophy and Biology of Cognitive Ethology by Colin Allen and Marc Bekoff
William S. Robinson
Dept. of Philosophy & Religious Studies
402 Catt Hall
Iowa State University
Ames, IA 50011-1306
USA
wsrob@iastate.edu
Copyright (c) William Robinson 1999
PSYCHE, 5(26), September 1999
Previously at: http://psyche.cs.monash.edu.au/v5/psyche-5-26-robinson.html
KEYWORDS: cognitive ethology, behaviorism, animal thought, animal consciousness, intentionality.
REVIEW OF: Colin Allen and Marc Bekoff (1997). Species of Mind: The
Philosophy and Biology of Cognitive Ethology. Bradford (MIT Press).
ISBN 0-262-01163-8. xii + 209pp. Price: $US35 hbk.
Species of Mind seeks primarily to establish the credentials of
cognitive ethology as an exciting and above all respectable branch of
scientific inquiry. Difficulties in achieving the aims of cognitive
ethology are not downplayed, but it is argued that they can be overcome
sufficiently to sustain an empirical research program. An important subgoal
is the promotion of an interdisciplinary approach to the required research.
If cognitive ethology were a more mature discipline, a book with the above
goals would hardly be necessary. It is partly because so many of its
questions are unsettled that cognitive ethology is exposed to doubts and
criticisms. One way of dealing with this situation would be to propound a
single approach, method, and set of established results as a model for
cognitive ethology, and argue that if the discipline is developed in the
recommended way, it is respectable science. Allen and Bekoff reject this
path and go out of their way to keep questions open. This way of proceeding
often requires them to forego simple claims about actual advances in favor
of complex, and somewhat weaker, claims that defend the empirical character
or likely future productivity of various approaches to cognitive ethology.
This circumspection is on the whole commendable, although it sometimes
leads to rather indecisive discussions.
An early concern in the book is whether investigation of animal mentality
is compatible with naturalistic science. The authors reject narrowly
reductionistic forms of naturalism, but leave the positive characterization
of naturalism largely open. One point they do wish to make (p. 12) is that
Darwinian continuity (of animal minds with human minds) is a more
naturalistic view than the traditional Cartesian claim of discontinuity.
Another early concern is to distinguish intentionality (in Brentano's sense
of aboutness or representation, not the ordinary sense of purposiveness)
from consciousness. The authors are concerned to show that many questions
about the former can be pursued independently of views about the latter.
Discussion of consciousness is deferred until relatively late in the book
(chapter eight).
Chapter two is "A brief historical account of classical ethology and
cognitive ethology." It provides a useful introduction to these subjects
that will serve to prepare nonethologists for understanding the issues in
the rest of the book.
Chapter three explores the difficulty raised by the fact that the same
bodily motion may constitute different behavior in different contexts, and
the same behavior may be achieved by different bodily motions on different
occasions. Descriptions in terms of bodily motions may thus fail adequately
to characterize behavior; but descriptions in terms of functions may draw
charges of unwarranted mentalistic inflation. It is thus difficult to
satisfy the desire for a way of describing activities of animals that can
be used reliably by different investigators across various contexts and
species. The authors "favor a pluralistic approach according to which it is
an empirical question which schemes for categorizing behavior will turn out
to be empirically most productive" (p. 47).
While the authors emphasize constructive suggestions for the development of
cognitive ethology, their project evidently requires a certain amount of
responding to critics of that enterprise. Chapter four takes on the
fundamental complaints that animal mental states are unobservable and
cannot be reported upon by their possessors. The authors hold that, barring
extreme and implausible interpretations of "private", privacy does not
entail having no effects. Thus, they argue, mental states in animals may be
introducible as part of the best explanation for observable events. If this
kind of introduction is to work, however, ethologists must clearly identify
what needs explaining. Helpful in doing this will be a distinction between
behavior that is relatively stimulus bound, and behavior that is relatively
stimulus free. The authors explain this distinction, and defend its
importance against a number of strategies that seek to dismiss "stimulus
free" behavior as, e.g., merely bound to stimuli of nonobvious complexity,
or as exhibiting chance.
In their pivotal fifth chapter, Allen and Bekoff argue for a thesis that is
considerably stronger than the cautious pluralism of chapter three. Namely,
they hold that eliminativism and behaviorism ignore content, and that this
makes them "less suitable for the explanatory purposes of cognitive
ethology than are approaches that ascribe content bearing states" (p. 69).
Moreover, they hold that folk psychology is a "prototheory" of behavior
which contains terms for contentful states and generalizations of a
causal-explanatory kind which "may be suitably refined and incorporated
into a fully scientific theory of mind and behavior applicable to both
humans and nonhumans" (p. 66).
Significant attractions of these theses for cognitive ethology are that
descriptions in contentful (or, intentional) terms can be easily tied to
evolutionary explanations (Millikan, 1984, is highly relevant background
here), that they may permit generalizations across species, and that they
are less difficult to find than either strictly behavioral or "syntactic"
(in the sense of Stich, 1983) descriptions. But, of course, many will be
suspicious that these benefits are illusory. Many discussions in the
remainder of the book implicitly or explicitly respond to various aspects
of this suspicion, particularly those raised by scientists. The focus of
chapter five is on objections coming from philosophers, most prominently
Stich (1983) and Dennett (1969, 1996). These writers have stressed the
difficulty of determinately specifying particular content ascriptions. It
is, for example, tempting to say that Fido, wagging his tail before his
bowl, is expecting food. But food is material with the function of
nourishment, and it is exceedingly doubtful that Fido has any such concept
as "function" or "nourishment"; thus it seems that "food" cannot be an
accurate term for what it is that Fido is expecting. This kind of
difficulty will infect any casual attribution of intentional content to
animal mental states.
Allen and Bekoff face this argument head on, making, among others, the
following points. (1) For cognitive ethology to be a science, it need not
achieve a level of precision that would support predictions of particular
actions of individual animals. (2) The fact that accurate content
ascription is difficult does not show that it is in-principle impossible.
Fido may not have the concept of food, or any concept that corresponds to a
single English word; but "there is still no reason to think that we cannot
manipulate English so as to explain what the dog's concepts are" (p. 82).
(3) Many of the difficulties put forward against content ascriptions for
animals apply to adult humans (who make mistakes and have various lacunae
in their knowledge) and to children (who may ask for food, for example,
before having a concept of nourishment). But in these cases, we do not draw
the conclusion that content ascriptions must be useless, or that they
cannot lead to reasonable predictions.
Allen and Bekoff's discussion will not satisfy those who (like the present
reviewer) have doubts that folk-psychological concepts are terms of a
causal-explanatory theory. However, the authors' position that they are
entitled to forego any attempt to allay these doubts is quite
reasonable. If they can show that content ascription to animals is no worse
off--or even not too much worse off--than content ascription to humans,
they will have gone as far as can be expected of them to meet a fundamental
philosophical objection to ascribing (intentional) content to states of
animals. There remains, however, a large gap that must be filled by
empirical work. For even if it is possible in principle for the resources
of English to provide adequate descriptions of animals' concepts, it might
not be practically possible for ethologists ever to find such a
description. The view that (useful) intentional descriptions will be less
difficult to find than behavioral descriptions might still be illusory,
because it might be that any conceptual description that comes out of field
work might be so inadequate as to prevent scientific advance. Whether this
problem will defeat the establishment of cognitive ethology as a science
is, however, an empirical matter. Thus, the authors quite properly turn, in
their next two chapters, to empirical studies.
Chapter six centers on a summary presentation of results from Bekoff's
(1995) study of play bows in canids. (A play bow is a distinctive posture
in which the rear stands, the forelegs "kneel" and the head is held low to
the ground.) The apparent point of the summary is to invoke empirical
support for the view that play bows can be legitimately assigned a
meaning--namely, "I still want to play". The method of the study was, in
brief, to film and analyze play episodes. Play bows were identified, as
were four types of actions that occur outside of play, and that could be
interpreted (by us or by the subjects) as aggressive. (For brevity, I will,
unlike the authors, call these "possibly aggressive" actions.) A key
question was whether play bows occurred randomly during play episodes, or
whether they were especially likely to occur in connection with the
possibly aggressive actions (thus helping to avoid misinterpretation that
would lead to fighting instead of play). Unfortunately, the authors give
such a cautious report that the answer to this question is not clear. The
table that is reproduced from the paper lists the possibly aggressive
actions, and shows mostly low percentages of these as being preceded or
followed by play bows. Further, the results are reported very weakly: "bows
may serve to provide information about other actions that follow or precede
them" (p. 103). However, some relevant facts that are clear in the paper
are not reported in the book's summary, to wit: of the play bows observed
during the reported episodes, 88.8% occurred just before or just after one
of the possibly aggressive actions in dogs, with 94.3% and 97.7% being the
corresponding figures for infant wolves and infant coyotes, respectively.
The preponderance of these bows occurred just before or just after the most
"aggressive" of the actions--74%, 79% and 92% for dogs, infant wolves and
infant coyotes, respectively.
The cited study thus does support the view that play bows during play
episodes have some kind of signalling function that helps avoid
degeneration of play into fighting. But this result leaves open the
question of just how to characterize the "message" in play bows. For this
purpose, one needs to tie the observational evidence to theories of
intentionality. Toward this end, the authors provide interesting
discussions of views of Rosenberg (1990), Dennett (1987), and especially
Millikan (1984). There are, however, some difficulties. The authors refer
to Dennett's "implausible assumption of perfect rationality" (p. 95). But,
as the authors are clearly aware (p. 93), Dennett does not assume that any
organism is (or even could be) perfectly rational; nor does he hold that
lack of perfect rationality makes the intentional stance useless. It is
thus unclear what assumption the authors wish to attribute to Dennett, and
what their complaint about this assumption amounts to. (These questions
surface again in Chapter 9.) After discussing Millikan's views, the authors
conclude that "it is probably more correct to view [play bows] as
intentional signals, a limiting case of intentional icons" (p. 108).
Millikan sharply distinguishes intentional icons from beliefs--the latter
but not the former involve "identification", and have participation in
inference as one of their proper functions. The authors are aware of these
differences; puzzlingly, however, they do refer (p. 109) to animals'
beliefs about other animals' intentional states, and they offer a curious
attempt (p. 97) to meld Millikan's approach with a thesis of Rosenberg's
concerning higher-order beliefs and desires involved in pretense. Readers
who are suspicious of unwarranted intentionalistic inflation of behavioral
descriptions will likely not be satisfied with these parts of the
discussion. In fairness, however, the authors do not claim to offer a
definitive conclusion about the correct analysis of the intentionality of
animals' behavior, and they make a strong case that attempts to relate
experimental work to theories of intentionality can be fruitful.
Chapter seven draws upon a large number of studies concerned with aspects
of antipredatory behavior. Animals have a repertoire of different responses
to predators, which may be called forth by different kinds of predators;
and the correct and timely classification of predators depends upon
vigilance behaviors. All these aspects of antipredatory behavior can be
studied empirically. They involve complexities for which it is plausible to
suppose that appeals to cognitive abilities are necessary, both to obtain
the best explanations and to suggest hypotheses for fruitful further study.
Some of the experimental results cited in this chapter suggest that the
problem of which concepts to attribute to animals can be solved
sufficiently to permit some solid scientific advance. The authors are,
however, acutely honest about the difficulties of such attribution. They
give an extended discussion of studies of effects of group size on the
vigilance behavior of certain birds that raises (among other points) the
question of what counts as a group, or what counts as being in a group of a
certain size, from the point of view of the animals involved. Skeptics
about our ability to find adequate descriptions for animals' concepts will
find many points for their case here. These points, however, have emerged
from a research program that is, at least, empirical; and the authors argue
for the viability and fruitfulness of continuing this kind of research.
Allen and Bekoff regard the questions we have reviewed so far as pursuable
(empirically and theoretically) independently of resolving the question of
animal consciousness, and they consistently put off the latter question
until chapter eight. In this chapter, they disclaim ability to answer
Nagel's famous question as to what it is like to be a bat (or other
nonhuman). They argue, however, that there can be an empirical inquiry into
whether there is anything at all it is like to be a certain
organism. Pivotal in their argument is the concept of an organism's ability
to remain sensitive to an input, while discounting it, i.e., while avoiding
responding to its normal meaning. An example of this ability is the
separation of perception and judgment that we make when we see the lines in
the Mueller-Lyer illusion as unequal in length, but judge them to be equal
and treat them accordingly. The argument that is based on this ability
seems to be as follows. There can be behavioral evidence that a (nonhuman,
nonverbal) organism can be both subject to an illusion and able to discount
it in responding. "If one takes seriously the idea that an organism can
discriminate its appearance states from its judgments (beliefs) about the
environment, then one is committed to the distinction between the way
things appear to the organism and its beliefs about them. In our view,
attributing conscious, subjective experiences may provide the best
explanation for the ability of some organisms to make this distinction"
(pp. 152-153).
A problem with this argument is that what perception provides to an
organism can be given a purely informational description. Information can
be regarded as being processed in various ways, and in a complex
information processor it can be discounted under certain circumstances.
There are programs for nonmonotonic reasoning that could be described as
"discounting" certain information, or no longer "taking it seriously",
without merely erasing it, or making it forever irretrievable. But such
programs are not convincing demonstrations of computer consciousness. Thus,
we need more than what the authors give us, if we are to conclude that
there is consciousness in an information-discounting organism, and not
merely unconscious, but appropriately complex, processing. In fact, to get
to a positive conclusion about consciousness from the authors' premises, we
must antecedently accept that "appearance" in its full-blooded
"consciousness" sense can legitimately be applied to what sense organs
provide to animals. But the legitimacy of attributing full- blooded
conscious appearances to animals is a way of describing exactly what an
argument for animal consciousness is supposed to show. (By "full-blooded
consciousness sense" of "appearance", I mean the sense in which, e.g., ties
can appear blue (whether veridically or not) when looked at by a normal
observer; there is a "what it is like" to have them appear that color. This
is to be contrasted with the "bloodless" sense, in which one may say that
it appears that Candidate X is going to lose. Here, there is no quality
that the incipient losing presents to appearance; there is only a judgment
that information in one's possession inconclusively supports the statement
that X will lose.)
A further difficulty with the argument for animal consciousness is that
even if conscious, subjective experiences (appearances in the full-blooded
sense) were found to be always present in humans when they
discounted perceptual information, it would by no means be clear how those
experiences explain the ability to discount information; thus,
invoking the idea of inference to the best explanation to help us support
animal consciousness is suspect. The difficulties of the authors' argument
for animal consciousness, however, should not obscure the significance of
the first premise of that argument. If we presuppose (or find some
other argument to show) that consciousness is an effect of all neural
activities that are complex enough to do a certain job (e.g., discount
information without merely becoming insensitive to it), then it is
important that we can establish empirically that that job is accomplished
in nonverbal animals.
Chapter nine is largely a defense of the value of field work, against the
view that only laboratory studies can provide a proper approach to animal
mentality. Much of the discussion takes the form of a detailed reply to
Heyes and Dickinson (1990, 1995). Among the key points made are the
following. (1) Field work is not all passive observation--subjects in the
field can be induced to respond to experimentally contrived situations. (2)
Not all intentional attributions need to take the form of beliefs; i.e.,
internal states may be needed that represent properties without being
connected to inferences in the manner characteristic of beliefs. (3)
Hypotheses about intentional states, like scientific hypotheses in general,
cannot be tested in isolation; thus simple behavioral laboratory indicators
are inadequate as criteria for intentional states. And (4) interpretation
of laboratory controls and results can go wrong if broader considerations,
available only from field work, are not taken into account.
Despite the length of this review, there are many topics that have had to
be left without mention. Species of Mind is a book that is rich in
both its ethology and its philosophical ideas. As a philosopher, I found
the ethology quite accessible; and I judge that most of the philosophy will
be found accessible by ethologists. This is a book that is intended to open
discussion rather than to close it, and it succeeds admirably in this aim.
Those just coming to the subject of cognitive ethology and those who are
already involved its debates will profit from this book, or pass it by at
their considerable cost.
References
Bekoff, M. (1995). Play signals as punctuation: The structure of social
play in canids. Behaviour, 132, 419-429.
Dennett, D.C. (1969). Content and consciousness. Routledge and Kegan
Paul.
Dennett, D.C. (1987). The intentional stance. Bradford: MIT Press.
Dennett, D.C. (1996). Kinds of minds. Basic Books.
Heyes, C. & Dickinson, A. (1990). The intentionality of animal action.
Mind and Language, 5, 87-104.
Heyes, C. & Dickinson, A. (1995). Folk psychology won't go away:
Response to Allen and Bekoff. Mind and Language, 10, 329-332.
Millikan, R.G. (1984). Language, thought, and other biological
categories. Bradford: MIT Press.
Rosenberg, A. (1990). Is there an evolutionary biology of play? In M.
Bekoff & D. Jamieson (eds.), Interpretation and explanation in the study
of animal behavior, vol. 1. Westview.
Stich, S. (1983). From folk psychology to cognitive science. Bradford:
MIT Press.
