A Study of Concepts

by Christopher Peacocke

This book provides a detailed, systematic, and accessible introduction to an original philosophical theory of concepts that Christopher Peacocke has developed in recent years to explain facts about the nature of thought, including its systematic character, its relations to truth and reference, and its normative dimension. Philosophers from Hume, Kant, and Wittgenstein to the recent realists and antirealists have sought to answer the question, What are concepts? This book provides a detailed, systematic, and accessible introduction to an original philosophical theory of concepts that Christopher Peacocke has developed in recent years to explain facts about the nature of thought, including its systematic character, its relations to truth and reference, and its normative dimension.

Particular concepts are also treated within the general framework: perceptual concepts, logical concepts, and the concept of belief are discussed in detail. The general theory is further applied in answering the question of how the ontology of concepts can be of use in classifying mental states, and in discussing the proper relation between philosophical and psychological theories of concepts. Finally, the theory of concepts is used to motivate a nonverificationist theory of the limits of intelligible thought.

Peacocke treats content as broad rather than narrow, and his account is nonreductive and non-Quinean. Yet Peacocke also argues for an interactive relationship between philosophical and psychological theories of concepts, and he plots many connections with work in cognitive psychology.

Genres: Philosophy

286 pages, Paperback

First published October 7, 1992

Reviews

[Alina · Rating 5 out of 5]

Peacocke’s starting question is ambitious: What is a concept, understood in such a way that the concept is the constituent of the content of any of our propositional attitudes (e.g., beliefs and desires) and can be identified with notions like “meaning” and “sense”? It’s meaningful to see what it can look like to offer a formal theory in addressing an issue this complex and sprawling. It’s easy to muse over this question from more speculative angles, like this which start off from ideas of our social conditioning and history.

I am not sure whether Peacocke’s answers are satisfactory. In fact, I’m strongly inclined to think they are not, but I’m still working this out (I’ll mentioned some of my tentative criticism below). First, I’ll present what I take to be the heart of Peacocke’s theory of concepts (which is laid out in chapters 1-5).

In chapter 1 “Individuating concepts” Peacocke introduces what he takes concepts to be and lays out desiderata for any theory of concepts. Concepts, by his lights, at the minimum should be understood as being distinct from one another if and only if there are two propositions, one which contains concept A and the other which contains concept B, and they differ only with regards to whether A or B is included—and it is possible for one proposition to be informative, and the other to not be so. We should also distinguish concepts from properties. Properties are invoked in order to talk about reality and address issues in metaphysics, whereas concepts are not at that level.

Peacocke also uses “concept” differently from how cognitive scientists do. He points out that the functionalist way in which concepts are defined for them cannot shed light upon the issue of how to tell when a subject’s prototype for a concept has changed, or whether the central beliefs associated with a concept has changed. Peacocke’s solidly person-level conception of a concept, in contrast, will shed light on these matters.

The desiderata for a theory of concepts, according to Peacocke, include that the theory should have a general form and so be applicable to any concept; provide constraints for an account of a particular concept to be legitimate; among some others.

We may move onto Peacocke’s theory. The idea is that any concept is individuated by a set of possession conditions, i.e., the conditions that must be satisfied if a subject properly grasps or possesses the concept, or in other words, if a subject has well-formed propositional attitudes towards contents that contain that concept. This amounts to a variety of conceptual role semantics. Peacocke thinks that the various things we can do with a concept define that concept itself. For example, if the condition that a red tomato is perceived by a person holds, that person’s capacity to perceive a red tomato and to form perceptual judgments about this are part of the account of what the concept red consists in.

The only other conceptual semantical theory with which I’m familiar is Brandom’s. Peacocke’s is strikingly different from his. Brandom places emphasis on how our social responsibilities play an essential role to explaining what capacities in using concepts we have, and which are counted as well-formed or not. Peacocke doesn’t think the social realm is essential to the identity of concepts.

In this first chapter where Peacocke lays out the foundation of his theory, he focuses on two examples. First is the observational concept red. Second is the logical concept conjunction. The latter concept is individuated by our capacities to find certain mental transitions primitively compelling (e.g., p, q => p C q). That is, we find these transitions compelling not on any basis of noticing various facts or pieces of evidence or drawing inferences from these. Moreover, the basis upon which this concept is individuated need not involve any further reasons or explanation for why we might think that these transitions are compelling.

The concept red, similarly, is individuated by our capacities to do certain things when we’re under the condition of perceiving something red. Under this condition, we’ll be disposed to refer to that thing demonstratively under the mode of presentation of its being red (i.e., we can make perceptual judgments that use ‘red’ properly). Moreover, when we’re not under that perceptual condition but rather are in thought, in which we take something as having the powers to cause perceptual experiences of the sort specified above, we’ll also be disposed to make judgments that use ‘red’ properly. The mental transitions specified in the naming of these capacities, e.g., to transition from an experience to a perceptual judgment, are, like the case of conjunction above, supposed to be primitively compelling.

Peacocke thinks we ought to avoid circularity in a theory of concepts. While his picture of possession conditions as individuating concepts might at first seem circular, it is not, because he thinks that any account of the possession conditions of a particular concept can avoid mention of the concept itself in the specification of the antecedent of a conditional. For example, in specifying the possession conditions for ‘red’, we don’t need to say that you must see something as red as the starting condition; instead, we can just say that the property red obtains in your visual field. (I’m not totally sure about the motivations behind this requirement, other than that it goes hand-in-hand with thinking that concepts “bottom out” in the world at some point, which secures that we can indeed know about the world, e.g., solipsism or idealism aren’t true).

There are two more essential parts of Peacocke’s picture worth stressing. They both fall out from the need to see how concepts connect up with how we know the world and form true beliefs about the world. A starting point is to appreciate that concepts on their own do not amount to propositions, and the proposition is the basic unit of making a statement or forming a belief about the world; the proposition is the basic unit that can have the property of being true or false. So the question arises: How do concepts combine as to form propositions?

Peacocke thinks that the possession conditions, which are supposed to exhaustively define or individuate a concept, do not involve any mention of the principles behind such combination. Rather, he thinks, possession conditions yield which mental transitions we’ll find primitively compelling, and as these transitions include those towards propositional attitudes, which already have complete propositional contents, we should think that the ways concepts get combined into such complete contents are determined by the possession conditions themselves in some fashion which is difficult to make explicit. (I have big confusion with this point, which I’ll return to below).

Second, Peacocke specifies that any account of the possession conditions of a given concept needs to be interdependent with an account of how the beliefs we’re disposed to forming, given these conditions conditions, are true. Call this a “determination theory”: it’s an account of how the semantic value of the concept is determined by appeal to its possession conditions and the world. Peacocke can use the requirement that an account of possession conditions comes with a determination theory to predict which apparent concepts are actually concepts at the end of the day. Arthur Prior proposed “tonk” as a concept (I’m not very familiar with this proposal, but did encounter it in a class on logic a while ago). Peacocke thinks that “tonk” cannot be an actual concept because it is impossible to say which sentences out of the contradictory ones that utilize it are true. (I suspect Peaocke’s treatment of Prior here could have significance for the issues I’m interested in, so let me dogear this for myself).

This lets us gloss Peacocke’s account in a new way. For a person to meet the possession conditions for a given concept can be understood in terms of how that person knows what it is for something (e.g., a referent) to be the concept’s semantic value. Then, we should understand the semantic value of a complete propositional content to be something built up off of the semantic values of its constituent concepts. He shows us how this is supposed to work out for the example of conjunction. The semantic value of ‘and’ is something that guarantees that transitions on its basis are always truth-preserving. So, this semantic value can be understood in terms of a function that takes in the variables of the semantic values of the individual propositions p and q upon which conjunction operates; and it outputs the truth value of the complete proposition jointly p and q. In this example, we can see that no external, additional principles of combination need to be raised to explain how combination of constituent concepts happen. We first had individually p and q, and then get jointly p and q. (I worry about this point… seems like non-logical cases aren’t analogous. I’ll return to this later).

That’s chapter 1. I’m running out of space, so will be more sketchy in summing up the other chapters. In chapter 2 “Structure and system,” Peacocke addresses the issue of combination mentioned above in more detail. He centers his account on the fact that it is built into the nature of concepts that they are general. This can be made sense in terms of: If you can entertain the proposition Fa and also cognize a distinct individual b, and b can be properly predicated by F, then you have the capacity to form propositional attitudes containing the content Fb. This is one important fact. The other important fact, in order for Peacocke to defend his conclusion, is that possessing a concept is for you to know what it is for things out there to be its semantic value.

From these two premises, Peacocke gets that it’s presupposed by their joint truth that in possessing a concept, you also know the semantic significance of combining the concept F with an appropriate object-level sense. The general nature of concepts and the dispositions to form judgments baked into the possession of a concept, in other words, gets us that we can generatively combine concepts into new propositions on the basis of having these possession conditions alone.

In chapter 3 “Perceptual concepts” Peacocke offers an account of the possession conditions for observational concepts like ‘red’ and ‘square’ and shows how these can be appreciated as essentially figuring into perceptual experience in general. In chapter 4 “The metaphysics of concepts” Peacocke squares his account with concerns of naturalism.

In chapter 5 “Concepts and norms in a natural world” Peacocke explains how his account addresses the normative nature of concepts. The interesting part of this chapter for me was Peacocke’s treatment of Kripke’s Wittgensteinian position on the indeterminacy of meaning. Peacocke defends his account against the potential objection Kripke’s position could leverage against him, that he hasn’t sufficiently accounted for the normative dimension of concepts (I was confused here whether Kripke would think that the indeterminacy of meaning is an essential element of the normativity of concepts… I’d like to return to that literature at some point). Peacocke emphasizes that we’re used a concept correctly only if we find certain instances of patterns of reasoning or mental transitions primitively compelled for the particular reason that these patterns are named in the possession condition of the concept, and these patterns of mental transition are truth-preserving. So meaning isn’t indeterminate for Peacocke. The semantic value of any concept is just whatever makes these patterns of mental transition truth-preserving.

A side speculation: While it’s been a long time since I’ve read Kripke on this, I get the sense that Peacocke’s response isn’t satisfactory. He hasn’t addressed the possibility that at a certain level of our making sense of something, there could be indeterminacy regarding which possession condition for which concept is applied, perhaps for the reason that the world is under-determined regarding which concept is more appropriate and the mind hasn’t actually applied a particular one yet. I think some elements of Peacocke’s position sounds alright when it’s circumscribed to the domain of accounting for concepts understood as essentially involving language use and our forming judgment—but this would have to be understood as occurring only at a “late stage” in a process of our sense-making, and if this speculation is right, then the occurrences at earlier stages would likely have implications for how we need to think about the properties of that late stage, some of which might overturn parts of Peacocke’s theory.

I was also interested when he brought up, finally, an example that did not consist in solely observational or logical concepts. He raises the case of the concept “just” or “justice.” I didn’t totally follow his discussion here, which focuses on whether these evaluative-normative concepts supervene on descriptive facts or not. I think this is connected up with the issue of whether there can be genuine indeterminacy regarding which concepts are applied in a given case. I suspect that a more careful treatment of examples like these could reveal holes in Peacocke’s account…

Peacocke also shows that his account is compatible with that we can be under illusions of meaning. It may seem that we mean something, when we in fact don’t, and aren’t meaning anything at all or mean something other than what we apparently do. Peacocke provides some conditions which might lead to such illusions: there may be no underlying patterns of mental transition that are truth-preserving, behind why you find a certain transition compelling. If there are such patterns, these patterns you implicitly rely upon across different instances of the judgments employing the apparently same concept may in fact fail to be mutually consistent. Or, across different instances, you might take your judgment to be answerable to different underlying patterns of truth-preserving mental transition.

Peacocke thinks that Kripke failed to have this solution to meaning skepticism because he dismissed the idea that a principle, specifying a pattern of truth-preserving mental transitions, could be appealed to in accounting for the meaning of something.

A side speculation: I suspect we might be under such conditions most of the time, whenever we’re awake and inevitably making sense of things. This would be because most of what we’re interested in concerns the significance of things; and significance is partly dependent upon what we’re interested in, which is subject to our will and often fluctuates. Then, the apparently same concept, e.g., x is “just,” as used across different judgments may abide by an apparent truth-preserving principle at that moment of judgment. But it would only be a genuinely so if it were the case that you always had the same interests, or were always in the same mood or affective state, across your lifetime. Because we change our interests and affects over time, that principle, as individuated upon that moment of judgment, would not be truth-preserving relative to you. This goes hand-in-hand with the general issue that it’s not clear what it means for a sentence involving evaluative concepts to be true, that there is an apparent fact-value gap. I’d like to think more about this.

In chapter 6 “The concept of belief: self knowledge and referential coherence,” Peacocke specifies at least two clauses are needed in the possession condition of the concept belief, and he defends these. The clauses involve that your mental state represents a certain content as holding, and you accept that content.

Moreover, the case of second-order judgments or self-ascriptions of belief is uniquely normally knowledge, unlike judgments involving other concepts. I’m not sure I followed this point. When I judge that in a glass there’s H2O, this could fail to be knowledge, even if I have mastered the possession condition of H2O. This is because my judgment could depend upon a false premise. In contrast, when we judge that we have belief, we can’t go wrong in this way; this is because we can’t have illusions of belief, according to Peacocke. The appearance of certain content and of our endorsement or acceptance of it just makes up that very content and force of acceptance.

I think this point could be undermined by consideration of the intuitive distinction between occurrent judgment and belief as something that holds in a long-term way. This point about ‘infallibility’ when it comes to knowing our beliefs seems to hold for only occurrent judgment, as one mental phenomenon that gets called ‘belief’ among various others.

Now some bigger picture reflections. My major concern with Peacocke’s theory is under-developed, but here are my starting intuitions. I’ve been puzzled over why Peacocke’s point that facts of combination of concepts should be able to be exhaustively explained by appeal to possession conditions alone is so important to him. I think it might be because if it turns out that the purported possession conditions for a concept do not contain within themselves the explanatory resources for making sense of how we achieve propositional content, then concepts can no longer be tightly connected to semantic value—and moreover, to the intrinsically personal level of sense and meaning. Then, his theory could be at risk of accounting for some potential mental phenomenon which is subpersonal and fails to be the concept, as the phenomenon that is often used in philosophy.

If how we arrive at our beliefs is not deterministic on the basis of a set of principles, i.e., possession conditions, and rather to explain this we must appeal to hard-to-analytically-understand conditions like our creativity or agency, it appears that then we’re forced into thinking that an approach to concepts like Peacocke’s could possibly work only for concepts understood as something top-down individuated by explicit judgments we’ve already formed, as defined in turn in terms of their roles in theoretical reasoning. Much of our mental content that gets associated with belief, however, cannot be defined in this way; for much of our lives we’re not reflecting upon what’s going on with ourselves, but rather are “absorbed” in experience and trust that reality holds in the ways in which we sense it to be. So if all of this followed, Peacocke account wouldn’t get at concepts themselves, per say, but only at one phenomenon that can get called “concept”…