Idea Transcript
Thinking without Words by José Luis Bermúdez. Oxford, Oxford University Press, 2003. Pp. xiii+225. H/b £25.00. That some animals, early hominids and infants think (or did so once) is increasingly hard to doubt. Our evidence includes everything from the manufacture and use of tools (55, 127), to the ways infants react when shown apparently impossible physical events (81). Bermúdez’ project is to explain the nature of this thinking. How can we know what an infant, ape or early hominid thinks? What determines the contents of their thoughts? Are they thinkers in just the sense that you and I are, or do we humans become fundamentally different types of thinker with our own distinctive cognitive abilities after infancy? Thinking without Words offers answers to these questions, mixing philosophical argument with accounts of empirical research. The notion that there is such a thing as ‘thinking without words’ is doubly controversial. First, some philosophers hold that creatures incapable of communication by language are also incapable of thought. Bermúdez counters this view in Chapter 3, where he argues at length that some non- and pre-linguistic creatures are capable of real thinking—that is, their decision-making involves both structured representations and instrumental reasoning about the consequences of actions (51). Second, the notion of ‘thinking without words’ is controversial because, assuming that nonlinguistic creatures can think, it is not obvious that their thinking differs in kind from ours. And even if their thinking does differ in kind from ours, it is not obvious that this has anything to do with language. However, as I will shortly explain, Bermúdez argues that languageless creatures’ thinking is more limited than ours and that ‘an important class of thoughts … is in principle unavailable to nonlinguistic creatures’ (150). Not only does their thinking exclude inference (111) and practical reasoning (131–2); they are also unable to think thoughts involving logical constants such as NOT and quantifiers such as EVERY (Chapters 8–9). If Bermúdez is right that languageless creatures don’t infer or quantify, what could their thinking be like? Come to think of it, how could they be thinking at all? Bermúdez adopts a nice strategy for answering these questions. Take a theory about some aspect of thinking, a theory which is coherent but doesn’t quite work because it lacks the resources to capture our ordinary ways of thinking. Then see whether that theory might work for a creature whose thinking is more limited than ours. Recycling theories in this way makes a virtue of their limitations, provided the theories are limited in just the ways that the target creatures’ thinking is limited. Consider NOT, a concept which apes are supposed to lack on account of their languageless state (as I’ll soon explain). Some philosophers have held that NOT modifies predicates rather than whole sentences. While that probably isn’t how our concept NOT works, there could be an element of ape thinking—‘protonegation’— that does work this way (142–5). So it is that Bermúdez attempts to characterise nonlinguistic thinking by recycling a failed theory of negation. The same strategy is used to explain what determines the contents of nonlinguistic creatures’ thoughts. Take ‘success semantics’, a theory according to which ‘the content of a belief is given by … the condition that would have to obtain for the various desires associated with it to be satisfied’ (65). While success semantics probably isn’t right for our thoughts (66–8), Bermúdez suggests, it’s the sort of theory we need for nonlinguistic thought. 1
Why is nonlinguistic thinking supposed to be limited and to differ from our thinking? In Chapters 8 and 9, Bermúdez argues like this: 1.
Thinking thoughts containing NOT requires the ability to think ‘about how the truth-value of one thought might be related to the truth-value of another thought’ (178); this is ‘intentional ascent’.
2.
Intentional ascent requires ‘semantic ascent’, which is the ability to think about sentences.
Therefore: 3.
Using NOT requires the ability to communicate by language.
In addition to NOT, this argument is supposed to work for any truth-functional connective, quantifier, tense operator or modal operator. The first premise, [1], is merely stated. Although quite widely held, it is not obvious how this premise could be true. Is Bermúdez’ position that the ability to think about the truth values of thoughts is always, or even just sometimes, actually exercised in thinking thoughts containing NOT? If it is, then we need to know how this ability is exercised. As he notes, thinking THE CAT IS NOT HERE doesn’t involve thinking about the truth values of thoughts in anything like the sense in which it involves thinking about cats (178). If, on the other hand, Bermúdez’ position is that thinking thoughts containing NOT never involves exercising an ability to think about the truth values of thoughts, then he needs to explain how merely having this ability could be a necessary condition for thinking thoughts containing NOT. Either way, it is unclear how, if at all, an ability to think about the truth values of thoughts is involved in thinking NOT-thoughts. The second premise, [2], is argued for. The argument starts by endorsing the view that thinking about thoughts requires thinking about their vehicles. Then, since intentional ascent involves ‘conscious cognitive access to the target thoughts’, it follows that the vehicles of these thoughts must be accessible to consciousness (159). Finally, Bermúdez argues that ‘public language sentences are the only possible … vehicles for thoughts that are to be the objects of reflexive thinking’, and he does this by eliminating rival candidates (160). One problem with this argument is that since, as Bermúdez says, ‘we have little idea of what the vehicle of nonlinguistic thought might be’ (192), he can’t have eliminated this particular vehicle or have shown that a thought’s having this vehicle would not permit it to be an object of reflexive thinking. However, there is some discussion of what the vehicles of nonlinguistic thought might be in Chapter 3, and this does suggest that such vehicles are unlikely to be objects of reflexive thinking. Another problem with arguing by elimination is that it sheds no light on why the conclusion is true. If we really can think about a thought only when we have somehow expressed it in a sentence that we use to communicate with, I’d like to know why. An argument from elimination leaves it mysterious why there should be a link between sentences we use for communication and thoughts we can think about. In short, then, much more needs to be said to explicate and defend the two premises of Bermúdez’ main argument relating thought and language. The project of understanding early hominid, infant and animal thought is surely interesting and important, but I don’t yet see what their lack of language has to do with it. A second way of relating thought and language is implicit in Thinking 2
without Words. Bermúdez remarks that ‘[w]e have no theoretical framework for understanding the content and nature of nonlinguistic thought’ (vii). Does this mean that we do have a theoretical framework for understanding language-users’ thoughts? In Chapter 2, Bermúdez gives the impression that we do, and that Gotlob Frege and Michael Dummett have together succeeded in providing it. The key to their theory is the claim that to grasp a thought is to understand a sentence which expresses it. This theory doesn’t work for the languageless, Bermúdez claims, because ‘no account has yet been given of what it might be to grasp a thought independently of understanding a sentence’ (19). I read Bermúdez as implying that thinking without words is especially problematic because we have an adequate conception of thinking which applies only to language-users. But do we? Let’s concede that grasping a thought is understanding a sentence which expresses it. Now we need to know what constitutes understanding a sentence. And of course no one has got very far with explaining this—or, at least, no one has got further with this than they have with explaining what it is to grasp a thought. Why not? Probably because understanding a sentence involves grasping the thought it expresses. An animal’s having a language would make it easier to understand the nature of its thinking only if we knew what it is to understand a sentence; but the problem of understanding sentences turns out to be no easier than the problem of grasping thoughts. In this respect, language-using animals’ thinking is no better understood than languageless animals’. The arguments relating thought and language are quite a small part of this book. The main part, Chapters 3–7, is a systematic attempt to describe a kind of thought and an associated kind of thinking, and to relate this theoretical description to a wide variety of experimental research. Even if Bermúdez hasn’t fully explained what this kind of thinking has to do with language, the more important question is surely whether he’s right about how any of his subjects—animals, infants and early hominids—think. On this point the book offers many valuable insights. I’m about to recommend it to you as a tremendous resource which takes in an enormous range of experimental work and is full of ideas about the nature of thought. But first I want to raise another objection. Bermúdez describes his book as a first step towards providing ‘a conceptual foundation for the various disciplines that are committed to giving psychological explanations of the behavior of nonlinguistic creatures’ (193). Talk of foundations is misleading because Bermúdez’ position is in part an error theory. You wouldn’t know it from reading this book, but an important cluster of theories in developmental psychology are deeply committed to the hypothesis that infants think and reason in just the ways that adults do. Take Renée Baillargeon’s research on infants’ understanding of objects. The presentation of her research in Thinking without Words is selective; Bermúdez says only that Baillargeon has shown that infants are ‘sensitive’ to certain higher-order principles about the ways objects behave (54, 78). But Baillargeon, like other researchers, is not content to describe infant cognition using uninformative notions like ‘sensitivity’. She aims for a deeper understanding of what happens when infants perceive objects. Her experiments are designed to show that ‘infants, like older learners, formulate rules or hypotheses about physical events and revise and elaborate these hypotheses in light of additional input’ (Aguiar and Baillargeon, ‘Developments in Young Infants’ Reasoning About Occluded Objects’, Cognitive Psychology, 2002, 45:267–336, p. 329). According to Bermúdez, however, infants are incapable of testing or revising 3
hypotheses; at best they are capable of ‘protoinference’, which is not how older learners reason. So, far from constituting a foundation for Baillargeon’s psychological explanations, Bermúdez’ views are incompatible with them. This incompatibility should be made explicit and its consequences explored in detail, as the conflict with a central strand of theorising about infant cognition is part of what makes this book interesting. Bermúdez engages with scientists’ views about thought more directly in other work (for example, in The Paradox of Self-Consciousness, Cambridge, MA: MIT Press, 1998, pp. 62–76); a deeper level of engagement would have enhanced this book. Thinking without Words is an accessible and fascinating book; it develops a theory of how animals, early hominids and infants think, which is supported by reference to many different areas of research as well as by philosophical arguments. Anyone who cares about any kind of thinking should enjoy reading it. STEPHEN BUTTERFILL St John’s College Cambridge CB2 1TP UK
4