Abstract: The concept of rationality differs between psychology, philosophy, economics and biology. For psychologists and philosophers, the emphasis is on the process by which decisions are made: rational beliefs are arrived at by reasoning and contrasted with beliefs arrived at by emotion, faith, authority or arbitrary choice. Economists emphasise consistency of choice, regardless of the process and the goal. Biologists use a concept that links both previous ideas. Following Darwin’s theory of natural selection, they expect animals to behave as if they had been designed to surpass the fitness of their conspecifics and use optimality to predict behaviour that might achieve this. I introduce the terms PP-rationality, E-rationality and B-rationality to refer to these three different conceptions, and explore the advantages and weaknesses of each of them. The concepts are first discussed and then illustrated with specific examples from research in bird behaviour, including New Caledonian crows’ tool design, hummingbirds’ preferences between flowers and starlings’ choices between walking and flying. I conclude that no single definition of rationality can serve the purposes of the research community but that agreement on meanings and justifications for each stand is both necessary and possible.
“When I use a word,” Humpty Dumpty said, in rather scornful tone, “it means just what I choose it to mean-neither more nor less”
‘The question is,” said Alice, “whether you can make words mean so many different things.”
The main questions that concern the contributors to this volume are:
Are any non-human animals rational?
What are the character and limits of rationality in animals?
Are unobservable processes such as reasoning valid causal accounts of behaviour?
What leads to differences in the kind of rationality exhibited by different species?
These are tough issues in the best of cases, but the real problem, as I see it, is that without a semantic effort we cannot even begin to discuss them: the questions contain words whose meanings cannot be assumed to be shared among those interested in the matter. Even accepting that too much defining inhibits thinking about the real issues, and that (as Humpty Dumpty tells us) definitions are arbitrary, clearly we cannot avoid reflecting on what our central theme, “rationality”, means for different authors. Responding to this need, my modest goal here is to discuss some ways in which this polysemous word is and perhaps should be used.
Guided by their differing goals and acceptability criteria, scholars in various disciplines have reached within-field consensus on workable definitions of rationality, and they produce data, reflections, models, theorems, and so on that provide evidence for the presence or absence of rationality and its boundaries as they understand them. These definitions, however, are at best consensual within particular fields. In my experience, a great deal of time is wasted arguing at cross-purposes while holding different understandings of rationality in mind. To mitigate this difficulty, I start by presenting an admittedly idiosyncratic discussion of various conceptions of rationality. In the case of my own field, biology, I will be forced to make a definition up, as none really exists at the moment.
I do not think that it is advisable (or feasible) to use a one-size-fits-all definition. Notions from different fields highlight such different aspects that to propose one overarching definition would be futile because few would follow it. I shall instead subsume all meanings of rationality into three categories, derived from my perception of the main uses in Philosophy and Psychology (PP-rationality), in Economics (E-rationality) and in Evolutionary Biology (B-rationality). I find all these uses necessary and appropriate for specific aims, but as I describe each of them I shall highlight what appears to me to be their virtues and their vices.
1)2.2. PP Rationality
The Oxford Companion to Philosophy’s entry for “Rationality” is a good starting point:
This is a feature of cognitive agents that they exhibit when they adopt beliefs on the basis of appropriate reasons […] Aristotle maintained that rationality is the key that distinguishes human beings from other animals. […] A stone or a tree is non-rational because it is not capable of carrying out rational assessment. A being who is capable of being rational but who regularly violates the principles of rational assessment is irrational. […] Rational beliefs have also been contrasted with beliefs arrived at through emotion, faith, authority or by an arbitrary choice. (Brown 1995, p. 744)
I suspect that this definition would sound acceptable to most non-philosophers, and also, to some extent to contemporary cognitive psychologists (behaviourists may feel more comfortable with what I call ‘E-rationality’, discussed in the next section). Hence I will use this entry as a working definition of PP rationality. Two features are particularly noteworthy.
First, the emphasis is on process, not on outcome. We can separate rational from non-rational beliefs depending on how they were arrived at, rather than according to their contents or the pattern of behaviour that results from them. There is clearly a difficulty in distinguishing ‘appropriate’ from inappropriate reasons, and the criteria for this distinction are likely to depend on cultural context. For example, to believe that giraffes result from a cross between panthers and camels did count as PP-rational once upon a time because it was based on what were then appropriate reasons. Indeed this belief was held by champions of rationality such as Aristotle and other Greek scholars. This empirically mistaken belief would not qualify as rational today, but no doubt it is rational today to believe in theories that will prove factually wrong as time goes by and science progresses.
Second, PP-rationality is understood not in terms of observable behaviours but of entities such as thoughts and beliefs. To judge whether behaviour is PP-rational one needs to establish if it is caused by beliefs that have emerged from a reasoning process. To assess the PP-rationality of non-humans, we would have to devise means to expose not just our subjects’ beliefs and the processes by which they were arrived at, but also to find a basis for judging whether these processes include ‘appropriate’ reasons in the sense discussed in the previous paragraph. This makes it very hard to assess whether, for example, a lion is rational, irrational or non-rational.
The adherence of cognitive psychologists to this second point is typified by Oaksford and Chater ( 1998), who point out that, although stomachs may be well adapted to perform their function (digestion), “they have no beliefs, desires or knowledge, and hence the question of their rationality does not arise” (p. 5). This exclusion of stomachs places them in the same rationality bracket as stones or trees, and seems reasonable within this framework, but it raises the question of which definition of rationality is at issue when questions about rationality are raised about the non-human world.
These two features would appear to place PP-rationality in a wholly unsuitable position to address our brief. Our focus is on non-human animals, whose thoughts, desires and beliefs are inaccessible in practice and possibly also in principle, and certainly are not the stuff of normal animal research. In dealing with non-verbal subjects, biologists find the notion of using such entities as causes of behaviour problematic, even though the use of some of them are now (after the cognitive revolution) widely accepted. Some kinds of behaviour are best explained by reference to ‘concepts’ and ‘representations’ that are observable only indirectly. For instance, if an animal is exposed repeatedly to an interval between two events such as a flash and a food reward, it will later show the same interval between the flash and performing a food-related action. Since the animal produces the interval, it is fair to say that the interval is represented in the animal and causes its behaviour. However, the fact that a representation causes behaviour does not imply that the subject has used reasoning.
The difficulties with PP-rationality are not limited to research with non-human animals. Many processes that give rise to the beliefs held by human subjects are in fact inaccessible to the holders of these beliefs, making it very hard to determine whether a belief has been arrived at on the basis of appropriate reasons. The hundred or so possibilities that chess masters are aware of examining before each actual move are a small subset of the available legal moves (de Groot 1965; Simon and Schaeffer 1992). It is likely that this subset is determined by unconscious processes that delve into the 50000 or so positions chess masters remember, and that choices are often made under the irrational influence of emotional or aesthetic factors without the player being aware of their influence or of their access to her full knowledge base of chess positions. Thus, even if the whole process ends in the belief that a given move is best, and if the player feels that she has arrived at this conclusion by reasoning, the elements that entered into her reasoning process may have been influenced by the kinds of mechanism that the present definition would explicitly exclude from rationality. If, say, the player has acquired a Pavlovian aversion to a given position because she saw it while she had a toothache, then she will play so as to avoid it, and in doing so, she will be influenced irrationally by her knowledge base, though this influence and the active parts of her knowledge base may be unconscious.
I am aware that my concerns apply not just to assessments of rationality but to many other aspects of animal experience including welfare, pain, goal-directed behaviour, theory of mind, and so on, as well as to some aspects of human experience. Nevertheless, I think that within our present focus (rationality in non-humans), PP-rationality is particularly hard to assess. The combined weight of these problems would lead me to exclude PP-rationality from my own research, were it not for my (perhaps PP-irrational) desire to “understand” my avian subjects and my belief that some progress can be made through painstaking experimentation. I shall return to this issue in the last of my empirical examples.
The focus of PP-rationality as I have described it is the rationality of beliefs or of the agents that hold them, or--in the language of cognitive psychologists--the rationality of information processing, rather than the rationality of actions. Yet psychologists, along with economists, are often concerned with the latter. To the extent that action is understood as essentially caused by certain mental processes and beliefs, my preceding comments about PP-rationality apply in similar ways. However, some notions of rationality concern themselves not with the mental processes that lead to beliefs or to behaviour, but with the resulting patterns of behaviour itself, and to this I now turn.