Introduction

One quite surprising development in modal logic has been the development of systems in which there are possible worlds in which some sentences are both true and false. This development has been greeted with incredulous stares of such intensity that the stares directed at Lewis's claims become shy glances. This development has been extended to the actual world, and there is a group of logicians who claim that there are some contradictions that are true, as well, of course, as being false. Note that they are not claiming that all contradictions are true, only that some are. These logicians are the dialethic logicians.

In this chapter we will consider some of these impossible worlds claims, and see what sense can be made of them. Impossible worlds appeared when doxastic logic was being discussed. There is also a group of logicians, the relevant logicians, who focused their attention on conditionals. On some relevant accounts of conditionals there are worlds in which sentences are both true and false. Conditionals have been the inspiration for many unusual things in logic, and have led to many strange uses of possible worlds. We begin with epistemic logic, move on to relevant logics, and then turn to dialethic systems of contradiction tolerant logics.

Omniscience

One of the sources of controversy in epistemic S4 is that the system describes a knower who is both logically and deductively omniscient.

We have asserted on several occasions that thoughts are complex mental events. They have components that occur together in certain relationships when the thoughts occur. The thought that today is Monday and tomorrow is Tuesday has as a component the idea that today is Monday. The thought that the sky is blue has as components the idea of the sky and the idea of being blue. Hence thoughts are composed of ideas, and thinking is ideation. The assumption that thoughts are complex played a central role in developing the expression theory of meaning. The recursion clause of the neo-Gricean analysis depends on it, as does the theory's ability to account for subsentential word meaning. In the next chapter, we will use the fact that ideas (or equivalently, concepts) are thoughts or parts of thoughts in order to define “idea.” While widely affirmed, the thesis that thoughts literally have parts has been questioned by many and vehemently denied by some. Since the postulate is foundational to everything else in this book, and central to cognitive psychology, this chapter will be an extended argument for the constituency thesis. We will observe in passing that while the thesis is plausible for thoughts and the objects of belief, it fails for the act of belief and other propositional attitudes.

IDEAS AS THOUGHT-PARTS

The relationship between words and ideas is like that between sentences and thoughts (§13.1). Words express ideas, just as sentences express thoughts.

Grice originated the program of defining speaker meaning in terms of intention, and provided what is still the benchmark analysis. Many variations have been proposed, all fairly close to the original. I shall review a wide variety of facts that show that Grice's basic approach was wrong, and that mine is more promising. While my analysis is a descendent of Grice's, the fundamental conception has been transformed. Speaker meaning is not the intention to communicate. It contrasts with other semantic acts in that it need not be audience-oriented. Speakers can mean something without expecting or hoping that their intentions be recognized.

Most of this chapter will focus on cognitive speaker meaning, comparing Grice's analysis to Theorem 3.5, according to which S meant that p by e provided that S used e as an indication that he occurrently believes p, without covertly simulating an unintentional indication of the belief. In light of the equivalence between meaning that p and expressing the belief that p (Definition 2.1), the discussion will be an implicit comparison of a Gricean analysis of expression with the analysis provided in Chapter 3.

The major extant competitor to the Gricean analysis will be critiqued in §4.4. This approach seeks to define speaker meaning or expression in terms of commitment and truth rather than in terms of intention and indication. Commitment has been defined variously, in terms of lending one's authority to a belief, or of conventional norms requiring belief, or of intended truth or verification conditions.

Chapter 14 introduced the thesis that ideas or concepts are thought-parts, and sought to clarify and establish the underlying assumption that thoughts have constituents. This chapter will formally define “idea,” and set out the basic properties of concepts. We will introduce the distinction between atomic and complex concepts, and analyze what it is to conceive concepts. We will discuss the content of concepts, and the fact that concepts represent objects. And we will distinguish the intentional content of ideas from their extension or objective reference.

FORMAL DEFINITION

The terms “idea” and “concept” are ambiguous in English, and have been used with many different meanings in philosophy and psychology. In particular, they have been used to mean universals (e.g., Husserl), sensory images (Hume), objects of thought (Descartes), contents (Burge), senses (Katz), conceptions or belief systems (many contemporary psychologists), and mental representations generally (Brentano). We will use “idea” and “concept” exclusively for thought-parts, distinguishing them from universals and other objects, images, and conceptions (see Chapter 19) as well as from contents (§15.6) and senses (§21.1). An idea, on my conception, is only one kind of mental representation. Whereas ideas or concepts are wordlike mental representations, images are picturelike, and conceptions are theorylike.

Even when “idea” and “concept” both denote thought-parts, they have somewhat different patterns of usage in standard English. The terms are equally natural when discussing expression or containment. They are interchangeable when discussing the contents and objects of thought-parts, and what they represent.

Chapter 2 began with the fundamental distinction between speaker meaning and word meaning. Even though “banana” means a kind of fruit, Alfred Kahn once used it to mean a decline in the gross national product in order to make fun of politicians who wanted him to avoid the terms “recession” and “depression” because of their negative associations. We have defined what it is for a speaker to mean or express something in terms of the speaker's intentions. We must now begin analyzing what it is for a word or other expression to mean something or to express an idea. What makes it true, for example, that “vixen” means “female fox” in English, but does not mean anything in French? Why does “jument” mean “female horse” in French today, whereas earlier it meant “pack horse”?

It seems self-evident that such facts about English and French are determined in some way by facts about the individuals who have actually spoken English and French. It would be as hard to maintain that an abstract system such as English arose and evolved independent of the actions of concrete human beings as it would be to maintain that the species Canis familiaris arose and evolved independent of concrete dogs.

One of the most influential objections to ideational theories charges that ideation does not have the independence needed to explain what meaning is in terms of ideation. The objection is that ideas are not conceptually, ontologically, or epistemologically prior to word meanings. Hence the postulation of ideas to explain meanings does not advance our understanding of meaning, because it is ad hoc, regressive, or circular. To explain meaning in terms of ideas is to explain meaning in terms of itself, or something that can only be known on the basis of itself.

Having defined what a speaker means in terms of what the speaker expresses, we will now define speaker expression in terms of intention and evidential meaning. The basic idea is that to express a thought or other mental state is to provide an indication that it is occurring to us. We will distinguish speaker expression from closely related concepts, particularly that of evidential expression. Competing analyses of speaker expression will be discussed in Chapter 4. Word expression will be defined in Part II.

SPEAKER, WORD, AND EVIDENTIAL SENSES

As we noted in Chapter 2, expression resembles meaning in having speaker, word, and evidential senses, related in roughly the same ways.

1. The look on S's face expresses fear. (Evidential Expression)

2. The word “fear” expresses the idea of fear. (Word Expression)

3. By saying “I'm afraid,” S is expressing fear. (Speaker Expression)

The first and most obvious difference concerns the subject to which the predicate “express” applies. In the speaker sense, the term applies to a person or other animate object, and does so in virtue of the person's saying or doing something. In the other two senses, the term “express” applies to inanimate objects like words or facial expressions, and not in virtue of their doing or saying anything themselves. Second, “x expressed fear” entails “x expressed himself” in the speaker sense, but words and facial expressions cannot express themselves. Third, “x expresses fear” entails “x is an expression” only in the nonspeaker senses.

Chapter 12 emphasized the differences between thought and belief, observing that a person can believe something without currently thinking it, and think it without currently believing it. In doing so, however, we presupposed a major point of similarity, namely, that belief and thought can have the same relational objects. Thinking the thought that it will rain and believing that it will rain are different mental states with the same relational object. It is conventional in English and philosophy to refer to the objects of belief as propositions (or statements). To believe that it will rain is to believe the proposition that it will rain. Not all thoughts are propositions, however. We can think “Will it rain?” but this thought cannot be an object of belief or disbelief, and cannot have a truth value. We will define propositions as “declarative” thoughts, and will briefly examine the contrasting category of nonpropositional thoughts. After setting out our standard method of referring to propositions, we will defend the corollary synonymy criterion of propositional identity against the Mates objection and others.

Thoughts, propositions, and beliefs are all closely related to sentences. The fact that thoughts are expressed by sentences, and that speakers use sentences to express thoughts, played a major role in the theory of meaning presented in Parts I and II. We exploit secondary conventions of ideo-reflexive reference (§7.6) in order to use sentences to refer to the thoughts they express.

We argued in the previous chapter that ideational theories of meaning are neither regressive nor circular. Ideas are logically and epistemologically independent of words. And the intentional content of ideas is a radically different sort of property than the meaning of words. Many have charged that if this is true, then ideational theories of meaning are incomplete. The objection is that ideational theories cannot provide a satisfactory explanation of what it is for a word to have a meaning without explaining what it is to have an idea, or what it is for an idea to be of something. Ideational theories of meaning are also thought to be incomplete because they ignore language-world relations, that is, reference. We will conclude our defense of the expression theory by responding to these objections.

UNDEFINED TERMS

We have defined meaning as the expression of ideas and other mental states. We have defined the term “expression” for both speakers and words. Speaker expression was defined in terms of intention and evidential meaning. Word expression was defined in terms of speaker expression and convention. We have defined ideas in terms of thoughts. An idea, we have said, is a thought or a cognitive thought-part (Definition 15.1). We think a thought, or conceive an idea, when it occurs to us. We spent several chapters clarifying the term “thought” to make sure that it was clearly understood, and that we know what it denotes. We all know introspectively what it is to think.

This chapter will briefly introduce the concept of speaker reference, distinguishing it from the markedly different notion of word reference. When S said “The president is tall,” he may have been referring to Al Gore even though the words he uttered referred to George Bush. This difference between word reference and speaker reference may have occurred because of a false belief about who is president, a false belief about what “president” means, or a verbal slip. Alternatively, the divergence may have been intentional, as when S is using a code, speaking loosely or figuratively, acquiescing in someone else's unconventional usage, or conveying his feelings about who should be president.

A constraint on any theory of word reference is satisfaction of what I call the “disquotation formula”: if it refers at all, “Φ” refers to Φ. “Detroit” (the name) refers to Detroit (the city), “Napoleon” refers to Napoleon, and so on. Putnam and Horwich have suggested that a theory of word reference would consist of nothing but such tautologies. A Davidsonian might say that a theory of word reference is a recursive theory entailing an instance of the disquotation formula as a theorem for each referring term in the language. Most would like a theory explaining why “Φ” refers to Φ. No such theory is remotely possible for speaker reference.

The term “meaning” expresses a close-knit family of concepts. In order to properly identify our subject, and to prevent equivocation, the concepts must be carefully identified and differentiated. In this chapter, we will distinguish speaker meaning from other types, and then discriminate several senses in which a speaker can mean something, defining them all in terms of expression, along with the related concept of speaker implication. Of particular importance for later developments will be the type of speaker meaning involving the expression of thoughts or ideas, in a sense of thought distinct from belief. Competing definitions of speaker meaning will be discussed in Chapter 4, after expression has been defined in Chapter 3. The reader might find it helpful along the way to examine Figure 2.1 at the end of this chapter, which presents all of the different senses of meaning we shall discuss in a classification tree.

SPEAKER, WORD, AND EVIDENTIAL SENSES

A glance through any dictionary will show that the most important and commonly used terms generally have a large set of meanings, some closely related and others quite distinct. “Means” is no exception. Since this is a book on meaning, it will be essential for us to distinguish the senses of “means” even more finely and more sharply than is profitable in a dictionary. The three most important senses for us are represented in the following sentences.

1. Boulders mean glacial activity. (Evidential Meaning)

2. “Boulder” means “large rounded stone block.” (Word Meaning)

3. By “boulder,” S means “kilo of cocaine.” (Speaker Meaning)

In §13.1, we said that thoughts are like sentences in being attitudinally neutral, abstract, generatively constructed representations with truth values and sentencelike constituent structures. The analogy between sentences and thoughts has led Fodor (1975) and others to speak of mental representation as a “language,” and to speak of the “language of thought.”

I believe that the language of thought idea is valuable for two reasons: it takes mental representations seriously, and it stresses their syntactic character. It emphasizes that mental representations have constituent structure. As a metaphor, the notion is unobjectionable, even apt. By providing a useful model of mental representation, the language of thought hypothesis serves to demystify the notion of thought, and to suggest ways to study and describe it. But when treated literally, as a serious hypothesis about what mental representations are, the view is misguided. Partly as a consequence, perhaps, there is a considerable lack of clarity in Fodor and others as to how thoughts are supposed to be related to the hypothesized formulas in the language of thought. Are its formulas the thoughts themselves, or “vehicles” that thinkers use in thinking? In this chapter, I will briefly survey the principal forms of the language of thought hypothesis that stretch the sentential analogy too far, and show why they are mistaken.

We will not be concerned with three things that might well be called “language of thought” hypotheses.

I argued in Parts I and II that the venerable formula “meaning consists in the expression of ideas” is true for the vast majority of expressions when the term “idea” denotes thoughts and their parts. I have devoted Part III to clarifying the notion of thought and ideation invoked in the expression theory, defending the assumption that thoughts have parts, and showing that occurrent thought plays an essential and distinctive role in the explanation of human behavior. I believe that the definition of “idea” and “concept” as denoting thoughts and their cognitive parts (Definition 15.1) is an accurate analysis of one of the standard meanings of these terms rather than an unconventional stipulation. Nonetheless, ever since the “new way of ideas” emerged with Hobbes, Descartes, and Locke, the terms “idea” and “concept” have been officially applied by philosophers, psychologists, and linguists to almost everything but thought-parts, including objects of thought, contents, images, words, inner speech, and conceptions or belief systems. Some of this usage seems to have been motivated by reductive theorizing, some by simple blindness. Nearly everyone who has used the terms “concept” and “idea” has done so inconsistently and unrigorously, defining them differently on different occasions or applying them when their official definition does not apply.

A man is thinking of the sky, on my analysis, provided a certain thought-part is occurring to him, namely, the idea of the sky. In general: S is thinking of Φ iff the idea of Φ is occurring to S (Definition 12.2).

One of the most venerable doctrines in the history of philosophy, linguistics, and psychology is the thesis that words are conventional signs of mental states, principally thoughts and ideas, and that meaning consists in their expression. This expression theory of meaning, as I call it, is firmly entrenched in our commonsense understanding of the world. But familiarity has bred complacency as well as contempt. Development of the doctrine was limited through the nineteenth century, and the twentieth century brought denunciation of the expression theory from generations of scholars. Behavioristic theories of meaning have now faded from view. But referential theories still dominate the field, despite insurmountable problems. This work is an extended effort to clarify, deepen, and defend the expression theory, thereby systematizing what is known about meaning and expression. The best way to do this, I believe, is to carry out the Gricean program, explaining what it is for words to have meaning in terms of speaker meaning and what it is for a speaker to mean something in terms of intention. To succeed in this project, we must develop the theory of thought as a fundamental mental phenomenon distinct from belief and desire, identifying ideas with parts of thoughts. This work, then, is a philosophical treatise on the foundations of semantics.

MEANING AS THE EXPRESSION OF THOUGHT

Like many other central philosophical and scientific ideas, the expression theory was first set out by Plato (427–347 b.c.) and Aristotle (384–322 b.c.).

Chapter 15 explored the basic properties of ideas or concepts, and Chapter 16 defined what it is to possess them. This chapter looks at how we come to possess concepts. We will identify four sources: observation (abstraction), communication, constructive thought, and abstractive thought. Psychological investigation of concept formation has focused on a process called “concept learning.” Since the process involves the formation and testing of hypotheses containing the concept to be acquired, it cannot be a way of acquiring concepts. This fact has been used to argue that all concepts are innate. But the argument assumes groundlessly that concept learning is the only method by which concepts could possibly be acquired. The proper conclusion to draw is that since concepts are thought-parts rather than beliefs, the acquisition of concepts is not a belief-forming operation. We will explore the possibility that some concept-formation processes, involving observation and abstraction, are basic psychophysical processes. I believe that the concentration on concept-learning experiments can be traced back to the historical confusion between the acquisition of concepts and the learning of word meanings.

ACQUISITION

We are said to acquire or form a concept when we change from nonpossession to possession.

Definition: S acquired (formed) concept C at t iff S possessed C at t but not before.

Acquisition coincides with an initial occurrence marking the beginning of a period of possession, during which the concept may or may not reoccur.