Pragmatism and the Point of Inquiry By Lucas P. Halpin “Philosophy ought to imitate the successful sciences in its methods.” - Pierce This essay is expository, but not exegetical. I’ll present a version of truth-rejecting pragmatism. My goal is to keep it simple, keep it clear, and portray it as a reasonable and attractive scientific view of language. We’ll work our way towards the view via series of violations of the commonsense web of belief about language, motivated by naturalism. Once truth has been rejected, it’s only natural to wonder what the point of inquiry, or science, is. Supposing there are no objective values in science, scientific practice could only be motivated by the satisfaction of subjective values. But, that just doesn’t mesh with the actual practice of science. The purpose of the second half of the essay is to identify “the point” of scientific inquiry. It’s not so different from the ordinary, truth-seeking conception of science as one might think. PART ONE REJECTING TRUTH 1:1 Privileged Access Descartes drew a distinction between what that which can be known with absolute certainty, and that which can be doubted. Let’s suppose we have certainty about x iff we have direct cognitive access to x. As commonsense would have it, we have direct cognitive access to our own thoughts. Supposing one directly apprehends their own thoughts, it is natural to conclude that she, the apprehender and owner of such thoughts, must exist. Hence the famous slogan, “I think, therefore I am.” Direct cognitive access to thought, or content, or meaning, is a component of the commonsense web of belief. Frege and Russell both seemed to share this commitment. Thus, Frege objected to the idea that Mt. Etna, the big rock itself, might be a constituent in our thoughts. We do not have direct cognitive access to Mt. Etna. “Now that part of the thought which corresponds to the name ‘Etna’ cannot be Mount Etna itself; it cannot be the meaning of this name. For each individual piece of frozen, solidified lava which is part of Mount Etna would then also be part of the thought that Etna is higher than Vesuvius. But it seems to me 1 absurd that pieces of lava, even pieces of which I had no knowledge, should be parts of my thought.” (Undated letter to Jourdain, in Frege 1980, p. 79) For this, among other reasons, Frege posited the notion of sense, something distinct from both term and extension to which we do have direct cognitive access. Content, on Frege’s view, is comprised of senses. Russell, in contrast, constructed his thoughts out of terms and extensions only.1 To accommodate the commonsensical privileged access idea, he restricted term reference to extensions of a certain variety only, those to which we have direct cognitive access. All other meaning is derivative, constructed from extensions to which we have direct cognitive access. There is no Fregean intermediary, “sense.” 1:2 Meaning ain’t in the Head/Extension The idea that we have direct cognitive access to content is no longer taken for granted. This is so for broadly naturalist reasons. As the naturalist has it, we have privileged access only to those things which are in our heads.2 I’ll provide no argument for this thesis, except to say that it seems to fit more “naturally” with the rest of our science. Accordingly, we have privileged access to contents only if contents are in the head.3 Suppose we combine this naturalist thesis with the Russellian view that content is comprised as extensions. Then, if extensions are not in head, we don’t have direct cognitive access to content. Given that we do talk about the outside world, extensions are not in the head. At least, you can’t construct an outside the head extension using only inside the head extensions as parts. So, we don’t have direct cognitive access to the contents of your thoughts. The remarks from the preceding paragraphs do not address Frege. Perhaps content is comprised of senses, not extensions, and we do have direct cognitive access to senses. We need to know more about senses. I take language to be comprised of three basic parts, a syntax, reality, and map from syntax to reality. Basic Parts of Language: a) Syntax (e.g., terms) b) Reality (e.g., extensions) 1 See Russell, “Knowledge by Acquaintance and Knowledge by Description.” I’m using ‘privileged access’ loosely here. 3 It’s common to cite Putnam here, “The Meaning of ‘Meaning’.” He argues that psychological states don’t intrinsically determine extension, and so his point is very similar to Quine’s, discussed in the next section. 2 2 c) Map (from syntax to reality) The syntax gets to be “about” reality because it is mapped to it. In looser terms, we care about our syntax because of how it relates us to reality, and somewhere in this relation is the stuff we think of as “meaning.” With this basic construal of the parts of language, let’s suppose that senses have two essential features. First, they are neither terms nor extensions. Second, they determine extension. The map from syntax to reality essentially determines extensions.4 Senses essentially determine reference. Maps are, essentially, neither terms nor extensions. Senses are, essentially, neither terms nor extensions. So, as I will construe senses, senses just are maps. The fit seems reasonable.5 So construed, Frege claims that we have direct cognitive access to maps. Taking senses, rather than extensions, to be meanings we can now construe Quine as objecting to Frege: Meaning ain’t in the head either. “Uncritical semantics is the myth of a museum in which the exhibits are meanings and the words are labels. To switch languages is to change the labels. Now the naturalist’s primary objection to this view is not an objection to meanings on account of their being mental entities. . . The primary objection persists even if we take the labeled exhibits not as mental ideas but as Platonic ideas or even as the denoted concrete objects. Semantics is vitiated by a pernicious mentalism as long as we regard a man’s semantics as somehow determinate in his mind beyond what might be implicit in his overt behavior. It is the very facts about meaning, not the entities meant, that must be construed in terms of behavior.”(Quine, Ontological Relativity, p. 27) Quine does not take the extension of ‘dog’ to be petting and bone throwing, and certainly not ‘dog’ uttering. Rather, the idea is that if there is a semantic map, it is to be found in our behavior. Our behavior is not in our heads. We only have privileged access to what is in our heads. So, we don’t have privileged access to behavior. Assuming meaning = senses = maps = behavior, we do not have privileged access to meaning. I’ll call the view that the semantic map = behavior pragmatism. Pragmatism: Maps are in our behavior. 4 This is not to say that the map intrinsically determines reference, so that it will “point to” the same referent in every world. 5 Instead, we might construe the syntax as having two layers, terms and senses. The sense-syntax is primarily mapped to reality, the term-syntax derivatively. So construed, the senses ≠ maps. Either version might help with Frege’s puzzle, accounting for the difference in cognitive significance between ‘a = a’ and ‘a = b‘. However, it is difficult to see how the two-syntax version helps with direct cognitive access to content, where content is distinct from bare syntax. A syntax is just a syntax, not typically regarded as content. So, privileged access to, minimally, the map, will still be required to save privileged access to content. Furthermore, the sense-syntax will have to be intrinsically mapped. 3 The pragmatist view violates the commonsense web of belief about meaning. As commonsense would have it, you can know just by thinking about it what your terms mean. Furthermore, according to commonsense, we utter the sentence (how/when/where we do) because the sentence means what it does. According to the pragmatist, the sentence means what it does because we utter it (how/when/where we do). Again, this constitutes a rather dramatic rejection of our commonsense web of belief. 1:3 Science and the Science of Science “The search for the pattern of inquiry is, accordingly, not one instituted in the dark or at large. It is checked and controlled by knowledge of the kinds of inquiry that have and have not worked; methods which, as was pointed out earlier, can be so compared as to yield reasoned or rational conclusions.” (Dewey, The Pattern of Inquiry) We might distinguish between backwards looking pragmatist views, which focus on acts of utterance (a type of behavior) and their causes/correlates alone, calling these ‘use,’ ‘indication,’ or ‘co-variational’ theories of meaning, and views which also look forwards, to dog petting and bone throwing, reserving the name ‘pragmatism’ for the latter. Here, I’ll lump forwards and backwards looking views together. We can bring out a salient commonality between pragmatic theories that focus solely on linguistic behavior (or brain behavior, it makes no difference), and more inclusive pragmatic theories (petting and bone throwing), via a comparison to scientific theories generically. Let’s take nomic necessity as primitive. We’ll treat that as a free lunch in the present essay. Then, science is in the business of describing, among other things, the laws, i.e., the nomic necessities. Loosely put, we look at actual instances of the phenomenon in question, and take these as evidence for the existence of general laws. And the crucial bit for the present essay: one counter instance to a law is sufficient to disconfirm the theory.6 Taking English to be our subject matter, we would then be looking at observable features of English, describing them, and taking these as evidence for the laws of English. It’s not clear what stuff comprises the observable features of English. This will depend not only on what we can “observe,” but also what we take to be English. To make life simple, let’s take our theories of two of the three parts of language, syntax and reality, for granted. The linguists and physicists have done their job, and are waiting on the semanticists to fill in the remaining component, the map. In short, they are waiting for a description of 6 In so far as I can tell, so long as there are different probabilities, just as there are different laws, the present meta-theory can be recast to accommodate the preference for statistical theories. 4 the law like connections, running from the syntax, through behavior, to reality. Also, to keep things simple, let’s focus on a backwards looking pragmatist view, and focus on utterances (rather than neural activity or something along those lines). So construed, the observable features of English would be utterances (acts) and the conditions in which they are uttered (reality). Loosely put, our job as scientists of language is to look at actual instances of the phenomenon in question, utterances, and abstract our way towards general laws connecting these acts with bits of syntax and bits of reality, and thereby reveal the law-like semantic map. And crucially, one counter instance to a semantic law is sufficient to disconfirm the theory. So construed, the job of a theory of meaning is to find the laws behind what people say/do. To say an utterance is false would then be to say that it violates a law of meaning. Since laws are inviolable, all that shows on the pragmatist view is that your theory of meaning is “false.” A linguistic item can’t really be false. Finally, there’s no point to a notion of truth that doesn’t discriminate. This is a familiar issue to linguists. When the topic of syntax arises in your introductory linguistics class, the first thing your professor will tell you is that she is not doing what your grade school grammar teacher was doing. She is not going to prescribe or proscribe grammar. Her theory won’t tell you to say “whom,” and it won’t tell you not to say “ain’t.” The theory is describing grammar. Its job is to explain how ‘ain’t’ fits in, how “whom” used to fit in, etc. The radical pragmatist is making a similar point. The semantics which involves the normative terms ‘true’ and ‘false’ is prescriptive. It tells you what you should and shouldn’t say. The paradigm successful sciences are descriptive. They don’t involve two sets of laws for each subject: one set of inviolable laws governing the exception-less behavior of objects, and a second set of violable laws, violation of which results objective disvaluing. Just one set. Since the pragmatist is mimicking the paradigm sciences, she says nothing about violable laws. Putnam locates the impediment to the science of meaning in the traditional notion of “meaning” itself. “Analysis of the deep structure of linguistic forms gives us an incomparably more powerful description of the syntax of natural languages than we have ever had before. But the dimensions of language associated with the word “meaning” is, in spite of the usual spate of heroic if misguided attempts, as much in the dark as it ever was. [[P]] In this essay, I want to explore why this should be so. In my opinion, the reason that so-called semantics is in so much worse condition than syntactic theory is that the prescientific concept on which semantics is based – the prescientific concept of meaning – is itself in much worse shape than the prescientific concept of syntax.” (Russell, The Meaning of ‘Meaning’) 5 From the perspective of radical pragmatism, the notion of meaning is as flawed as the notions which it serves. On her view, you aren’t going to fix the science of language by fiddling around with your notion of meaning while keeping the objective normative semantic notions “true” and “false.” The latter have to go. This essay focuses on radical pragmatism. Radical Pragmatism: A pragmatist that rejects the normative notions, true and false. 1:4 A Case Study: Davidson In this section, I’ll present a spoof on Davidson, faux Davidson, to generate contrast. I’ll start with one of my favorite passages. “If you see a ketch sailing by and your companion says, ‘Look at that handsome yawl,’ you may be faced with a problem of interpretation. One natural possibility is that your friend has mistaken a ketch for a yawl, and has formed a false belief. But if his vision is good and his line of sight favorable it is even more plausible that he does not use the word ‘yawl’ quite as you do, and has made no mistake at all about the position of the jigger on the passing yacht. We do this sort of off the cuff interpretation all the time, deciding in favor of reinterpretation of words in order to preserve a reasonable theory of belief.”(Davidson, “On the very Idea of a Conceptual Scheme,” page 196 in Truth and Interpretation) The passage above nicely illustrates how one might go in for a theory according to which everything everyone says is true. You just construe what they say as having whatever meaning, or map, it must have to be true. But this not what faux Davidson does. He’s not a radical pragmatist. Faux Davidson attempts to distinguish between the true and the false. Consider an instance of a disquotational T-schema: ‘S’ is true iff S. T-schema (disquotational): ‘S’ is true iff S. T-Schema Instance: ‘It’s raining’ is true iff it’s raining. Tarski took it for granted that disquotation correctly produces the truth conditions for a sentence. Thus, a theory of truth, on Tarski’s view, can be “tested” against instances of the schema. However, if the theory didn’t use disquotation, there would be a non-trivial matching problem, of matching ‘S’s with Ps in instances of the non-disquotational schema below. T-schmea (non-disquatational): ‘S’ is true iff P. 6 Faux Davidson’s idea is that the non-trivial matching problem found in non-disquotational schema is the problem of meaning. Rather than disquoting to get S, Davidson advises us to look and listen for where people utter sentences (hold them to be true), and use this as empirical evidence guiding our choice of P. Here’s Davidson. Focus on the differences between (T), (E), and (GE). “(T) ‘Es regnet’ is true-in-German when spoken by x at time t if and only if it is raining near x at time t. On the other hand, we have the evidence, in the form: (E) Kurt belongs to the German speech community and Kurt holds true ‘Es Regnet’ on Saturday at noon and it is raining near Kurt on Saturday at noon. We should, I think, consider (E) as evidence that (T) is true. Since (T) is a universally quantified conditional, the first step would be to gather more evidence to support the claim that: (GE) [for every x and for every t](if x belongs to the German speech community then (x holds true ‘Es regnet’ at t if and only if it is raining near x at t)).” (Davidson, Radical Interpretation, p. 135) Let’s make a couple of gratuitous assumptions. First, holds true = utters. Second, truth conditions = ought-to-be-uttered conditions. On faux Davidson’s view, we start by looking at specific utterances, noting the conditions in which they occur. We record these as data and then we generalize to get laws governing human utterances, similar to (GE) above, but with ‘utters’ in place of ‘holds true’. (GE) constitutes a claim about what is the case (utterance conditions). (T) constitutes a claim about what ought to be the case (ought-to-be-uttered conditions). We derive the ought from the is with a little help from demoncracy. The normative laws of language are just what they need to be to render as much of what is uttered ought-to-be-uttered as possible. Democracy has intrinsic normative powers. The most important feature of faux Davidson’s theory, for present purposes, is the double layering, one set of inviolable laws of the normal scientific variety, and a second set of laws, serving a higher purpose, but constantly being violated. The radical pragmatist responds by simply noting the dissimilarity between this and paradigm science. Part One Conclusion Russell and Frege, following Descartes, thought we had privileged cognitive access to content, not just syntax. Given what I take to be broadly construed, naturalist tendencies, Putnam and Quine (and, I take it, the classical pragmatists) rejected the commonsense idea that we have privileged cognitive access to 7 content, thereby doing considerable damage to our commonsense web of belief. Pragmatists generally, locating maps in behavior, went still further in rejecting the idea that we say it because of what it means. Rather, it means it because we say it. Thereby, they did considerably more damage to our commonsense web of belief. Philosophers such as Rorty (Quine?)7 complain that theories of meaning that retain a notion of truth haven’t gone far enough. Contrary to commonsense, there is no objective, normative truth either. “What really needs debate between the pragmatist and the intuitive realist is not whether we have intuitions to the effect that “truth is more than assertability”. . . Of course we have such intuitions. How could we escape having them? We have been educated within an intellectual tradition built around such claims . . . The pragmatist is urging that we do our best to stop having such intuitions“ (Rorty, intro to Consequences, XXiX) PART TWO THE POINT OF INQUIRY What is one to do without truth? Criticism which Fine offers on behalf of pragmatism’s foes will strike a chord with many. “suppose it suffices for universal relativism that it is true enough provided only that some believe it (whether one or many), then how could it be a doctrine that the relativist can advocate to non-believers (for whom it is already false!) in order to persuade them to change their minds? Indeed, what would be the point of having people change their minds? And if relativism is a doctrine that cannot be sould on its merits, then what merit does it really have?”(Arthur Fine, Relativism, Pragmatism, and Science, p. 52 in New Pragmatists, Cheryl Misak) Set aside variance/relativism. Why have someone change their mind about anything? Pragmatism needs is a point off in the distance, something for theories to move towards, something in virtue of which progress constitutes progress, objectively valuable or not.8 Science does move. Why does it do so? In the remainder of this essay, I’ll explain the point of theory change, according to radical pragmatism. As 7 For what it’s worth: Putnam has recently complained (Sardonic Comments) that Quine was not a verificationist, he was naturalist, and stimulus conditions are not verification conditions. In loose terms, whether or not stimulus conditions are, or might as well be, verification conditions depends on how truth is treated. 8 For a recent essay that takes the Davidson strain in Rorty more seriously than the present essay, see “Drawing Battle Lines and Choosing Bedfellows,” Sharyn Clough. 8 was noted in the introduction, it’s not so different from the project of seeking a theory of correspondence as one might think. 2:1 Radical Pragmatism and Subjective Values The radical pragmatist does not assign objective, normative, truth values to sentences. As the radical pragmatist has it, science involves only the subjective valuing of sentences (sentences of idiolects). Here’s a passage from Rorty. “So the pragmatist sees no need to worry about whether Plato or Kant was right in thinking that something nonspatio-temporal made moral judgments true, nor about whether the absence of such a thing means that such judgments are “merely expressions of emotion” or “merely conventional” or “merely subjective.” [[P]] This insouciance brings down the scorn of both kinds of Philosophers upon the pragmatist. The Platonist sees the pragmatist as merely a fuzzy-minded sort of positivist. The positivist sees him as lending aid and comfort to Platonism by leveling down the distinction between Objective Truth—the sort of true sentence attained by “the scientific method”—and sentences which lack the precious “correspondence to reality” which only that method can induce. Both join in thinking the pragmatist is not really a Philosopher. The pragmatist tries to defend himself by saying that one can be a philosopher precisely by being anti-Philosophical, that the best way to make things hang together is to step back from the issues between the Platonists and the positivists, and thereby give up the presuppositions of Philosophy.” - Rorty9 The positivists drew a distinction between sentences with genuine (empirical, verifiable), truth evaluable, content and those without. Typically, ethical sentences were regarded as falling out the without-side. Perhaps expressions of moral judgment are merely expressions of one’s own valuing, not truth evaluable claims with objective truth values. Perhaps radical pragmatists should think of uses of the terms ‘true’ and ‘false’ in the way positivists thought of uses of the terms ‘good’ and ‘bad’. In that case, scientific evaluations and ethical evaluations would not be different in kind. The radical pragmatist won’t see herself as downgrading science. Rather, there are just no distinctions between “grades” of the sort the positivists, or the realists/platonists, wanted. Here’s Rorty again, on James this time. “The question is precisely whether “the true” is more than what William James defined it as: “the name of whatever proves itself to be good in the way of belief, and good, to, for definite assignable reasons.”[James, “Pragmatism and the Meaning of Truth”] On James view, “true” resembles “good” or 9 Rorty, 1982, xvi-xvii. 9 “rational” in being a normative notion, a compliment paid to sentences that seem to be paying their way and that fit in with other sentences which are doing so. To think that Truth is “out there” is, on their view, on all fours with the Platonic view that The Good is “out there.”(Rorty, Intro to Consequences of Pragmatism, p. XXV) In the present essay, I will commit the radical pragmatist to this much: linguistic items generally are either valued or not valued, they never have objective values. However, I will not give utterance conditions for English meta-theoretical sentences involving the word ‘true’. This essay is a meta-theory (sometimes meta-meta-theory), and the term true simply won’t be incorporated. Still, we do need to think more about valuing. However, to make sense of the role of valuing in the radical pragmatist theory, we need to take a detour through idiolects. 2:2 Pragmatism and Idiolects I take it that the commitment to naturalism motivates the radical pragmatist to describe the semantics of language in terms of utterance conditions, rather than truth conditions. It also motivates her to appeal to idiolects. I’ll draw this commitment out by addressing an issue in faux Davidson’s view. According to faux Davidson, the theory which construes the most utterances as ought-to-have-been uttered is the correct theory of ought-to-have-been uttered conditions. But isn’t the theory according to which utterance conditions = ought-to-be-uttered conditions the theory which maximizes ought-to-beuttered-hood? After all, in that case, everything that is uttered ought-to-have-been uttered. And doesn’t that render all sentences true, and hence undercut the true/false distinction? There answer is, “Yes, but.” But, only if there is such a theory. There can’t be, on faux Davidson’s view. The issue then is why? Why aren’t utterance conditions = ought-to-be-uttered conditions? Sweeping some complexities under the rug, the short answer is simply that theories of utterance conditions and theories of ought-to-be-uttered conditions have different domains, or cover different ground. A theory of ought-to-be uttered conditions, if it’s to do the sort of work desired of it, has to cover reasonably sized groups of individuals. That’s how you get objective, normative conflicts between scientists: Tolemy was false and Copernicus was true, rather than: Tolemy was Tolemy-true and Copernicus was Copernicus-true. There can be no general theory of utterance conditions that covers these same reasonably sized groups of people simply because there is too much variation in the linguistic 10 behavior of individual speakers. No general10 recursive theory of utterance conditions is going to capture all of this behavior. This is not to say that the connection between utterance conditions and utterances isn’t causal, or that we cannot describe them (though we will come back to this). This is just to say that there is significant variation across individuals, and within individuals across time. In the event that you attribute different laws to different individuals, you have trivially postulated different utterance-idiolects. One set of laws for the deer, another for the frogs, one set for the Holmes, another for the Watson. Since utteranceidiolects are the only lects there are on the radical pragmatist’s view, she understands natural language semantics in terms of utterance idiolects. This is another violation of our commonsense web of belief about language. As commonsense would have it, we all (all English speakers) speak the same language and that is what facilitates objective, normative, sceintific conflict, i.e., genuine competition for “The Truth.” That is also what facilitates communication. None of that can be right on the radical pragmatist view. Chomsky suggests that some aliens, looking at us through a telescope perhaps, will regard us all as using the same syntax. It might be that they view us all as having the same map as well. We are all human beings after all. So construed, there would be no idiolects, just one shared language. However, we aren’t looking at each other through a telescope. We are looking at each other through a microscope. We can see all the pores and wrinkles, and these things matter to us. We want our scientific theories to capture the details and hence the differences. To the extent that one set of semantic laws can’t capture all the idiosyncrasies of human linguistic behavior, we will need multiple sets of laws that cover less semantic ground. 2:3 The Nature and Place of Valuing “I stated my conclusion that Mr. Russell’s interpretation of my view in terms of satisfaction of personal desire, of success in activities performed in order to satisfy desires, etc., was due to failure to note the importance in my theory of the existence of indeterminate of problematic situations as not only the source of but as the control of, inquiry.”(Dewey, Propositions, Warranted Assertability, and Truth, p. 182) “The problematic nature of situations is definitely stated to have its source and prototype in the condition of imbalance or disequilibrium that recurs rhythmically in the interactivity of organism and environment;—a condition exemplified in hunger, not as a “feeling” but as a form of organic behavior 10 Theory whose variables range over all participants in some scientific “activity.” 11 such as is manifested, for example, in bodily restlessness and bodily acts of search for food.” (Dewey, Propositions, Warranted Assertability, and Truth, p. 183) In this section, I will offer a partial account of valuing. The theory is not analysis; we are already waist deep in violations of the commonsense meta-theory. It’s not reductive. It’s not exegetical either, though I make no claim to originality. In giving an account of valuing, I shall subject myself to one constraint and to one constraint only. Valuing will be what it needs to be in order to account for nature and goings-on of language. Let’s start with a definition. Inquiry: Inquiry = theory change. (The agent has inquired iff her theory changes. A stipulation, not an analysis.) On the present view, there are two types of inquiry, theory change within a fixed language and theory change which involves language change. Just as we need to postulate different idiolects for different individuals, we will need to posit different idiolects for the same individual at different times. 11 In these cases, the relevant individuals have also changed their theories, and so, inquired. That’s not the standard use of the term ‘inquire’. But it’s the use I am adopting here, a bit of language change. The radical pragmatist does not need an additional theory component to explain the first sort of inquiry, theory change within a fixed language. Her theory of utterance conditions already gives her that. Suppose a given agent has a fixed idiolect. Let utterance conditions = stimulus conditions. If he (the native) utters S, that’s because the stimulus condition for S obtains. If he changes his utterances, that’s 11 For what it is worth: If you cut your idiolects as finely as belief/utterance change itself, then you’ll lose the distinction between language change and theory change. All instances of theory change will be instances of language change. The other way to get rid of the language/theory change distinction is to give up on maps (semantic laws) entirely. In evaluating these options, as a naturalist, one should consider whether one would treat laws in other areas of nature in the same way: give them up entirely, are cast the laws as changing with every object that changes. Also, Quine, Rorty, and Davidson, seemed to think that the language change/theory change distinction stands or falls with the analytic/synthetic distinction. Here’s a few interesting passages. “Carnap, Lewis, and others and others take a pragmatic stand on the question of choosing between language forms, scientific frameworks; but their pragmatism leaves off at the imagined boundary between the analytic and the synthetic. In repudiating such a boundary I espouse a more thorough pragmatism. Each man is given a scientific heritage plus a continuing barrage of sensory stimulation; and the considerations which guide him in warping his scientific heritage to fit his continuing sensory promptings are, where rational, pragmatic.”(Quine, Two Dogmas) “Quine’s suggestion that the difference between a priori and empirical truth is merely that between the relatively difficult to give up and the relatively easy brings in its train the notion that there is no clear distinction to be drawn between questions of meaning and questions of fact. This in turn leaves us (as Quine pointed out in criticizing Carnap) with no distinctions about alternative “theories” and alternative “frameworks” “(Rorty, World Well Lost, p. 5) “To give up on the analytic synthetic distinction as basic to the understanding of language is to give up on the idea that we can clearly distinguish between theory and language. Meaning, as we might loosely use the word, is contaminated by theory.”(Davidson, On the Very Idea of a Conceptual Scheme) 12 because the stimulus conditions have changed, i.e., reality has changed: “It’s raining,” “Now it’s not raining.” Of course, more complexity is wanted. But at the level of abstraction and idealization we are currently working at, we don’t need another piece for our theoretical puzzle. Thus, we don’t need valuing. We don’t need it to describe the map, the closest thing in the present account to “sense.” We don’t need it to describe the utterance condition, the closest thing in the present account to “referent.” Still, there is one thing we don’t have. We don’t have a theory of language change. Suppose an agent goes about speaking a certain idiolect for a time. However, after a certain point, time t, the theory that you were using to describe and predict this agent’s behavior stops working. To account for this, you posit different laws, and your new theory succeeds in describing and predicting the agent’s behavior after time t, but not before. Furthermore, no theory is adequate to the agent’s behavior both before time t and after time t. In light of this, you posit language change, i.e., a change in the semantic laws. The agent was speaking one idiolect, but is now speaking another. Nothing in our theory is capable of predicting language change, the second kind of inquiry. Here’s a simple example. Let utterance conditions = stimulus conditions. Now suppose the stimulus an agent receives remains constant, but she changes her theory anyway. In other terms, she is inquiring, but her stimulus remains constant. In that case, we will have to suppose that her utterance conditions for sentences have changed. She now has a new idiolect. Why would she do that? Thinking of language change as a type of motion, what determines whether there is motion or not? This will be the role of valuing. Valuing: Valuing is the impetus to language change. We need a bit more. Let’s pause for a brief lesson from Socrates and Hume. Socrates asks Euthyphro if the pious is loved by the Gods because it is pious, or if it is pious because it is loved by the God’s. Hume had his answer. It is pious because it is loved by the Gods. Hume had his list too. I can’t remember exactly what was on Hume’s list, the “be attitudes” plus humor, I think. Anyhow, Socrates might have asked a similar question about truth. Is the truth loved by scientists because it is truth, or is it truth because it is loved by scientists. The radical pragmatist has her answer. It is true because it is loved by scientists. But the radical pragmatist also needs a descriptive list. What do scientists love? We aren’t using ‘truth’ in our meta-theory. In slightly different terms then, not only do we need an impetus to motion, we need a direction, something that scientific language is evolving towards. Valued: Valued = the direction of language change. 13 2:4 Systematicity, for the Love of Science “Inquiry is the controlled or directed transformation of an indeterminate situation into one that is so determinate in its constituent distinctions and relations as to convert the elements of the original into a unified whole.” (Dewey, The Pattern of Inquiry, p. 1) “The biological antecedent conditions of an unsettled [indeterminate] situation are involved in that state of imbalance in organic-environmental interactions which has already been described.”(Dewey, The Pattern of Inquiry, p. 2) We need an item for the list, something that scientists love, something for inquiry to gravitate towards, a bit more precisely, a direction for language change. We’ll get at things, once again, in a roundabout way via a discussion of faux Davidson. Idiolects, trivially, constitute a sort of “variance.” However, it is commonly thought that pragmatism is incompatible with variance, or at least certain forms of it. According to Davidson 12, Rorty and others, if there were an interestingly variant language, we would not be able to give a semantic theory for the language, i.e., we wouldn’t be able to state the relevant laws. Not being in a position to describe the laws, we’d have no reason to regard it as a language, rather than chaotic noise. Here’s a passage from Rorty, echoing Davidson’s idea. “But … if we can never give a translation [describe the laws], why should we think that we are faced with language users at all? It is, of course, possible to imagine humanoid organisms making sounds of great variety at one another in various circumstances with what appear to be various effects upon the interlocutors’ behavior. But suppose that repeated attempts to systematically correlate these sounds with the organisms’ environment and behavior fail. What should we say? One suggestion might be that the analytical hypothesis we are using in our tentative translation schemes use concepts that we do not share with the natives – because the natives “carve up the world” differently, or have different “qualitative spaces” or something of the sort. But could there be a way of deciding between this suggestion and the possibility that the organisms’ sounds are just sounds?”(Rorty, World Well Lost, p. 6) If the problematic variance concerns objectively normative aspects of languages, for example, incommensurable truth conditions, we don’t need to discuss that. The radical pragmatist rejects objective normative aspects of language. If the claim is that we cannot describe variant syntaxes, it is plainly “false.” If the claim is that we cannot describe variant behavior (maps), it is plainly false. If the claim is that there is only one reality, it is plainly irrelevant. The idea has always been that variance is in us, the syntax and the map. There are a variety of variances. Some are more interesting than others. All are 12 On the Very Idea of a Conceptual Scheme. 14 interesting enough, once you’ve done away with the traditional, starry notions of “truth” and “meaning.” We won’t canvass them here. However, we can take this issue as a clue. I’ll use the term ‘systematicity’ as short for ‘systematic correlation with the environment’. And we shall conceive of this loosely. ‘Systematicity’ has a life of its own in the semantic literature. We are not taking that on here. I like the connotations the term has: an elegant and repetitive pattern. But, I’m not prepared to commit the radical pragmatist to anything specific here. Getting to the point. . . The radical pragmatist needs to deny that “systematicity” is a necessary feature of idiolects (language as such), where idiolects are those things we actually have. In that case, she is not required to identify another’s idiolect as systematic, in order to apply the term ‘language’, or the term ‘idiolect’, to it. This is a familiar sort of maneuver for pragmatists who sympathize with Wittgenstein. In response to faux Davidsion, this will give her some leeway in what she describes as a language. More crucially for the purpose of the present essay, the move purchases her an item for the list. Systematicity, accordingly, is something scientists love. They simply gravitate towards it. Of course, the radical pragmatist will want to say something about the mechanisms via which this “gravitation” occurs. But that is detail work to be done later. Scientific languages evolve towards systematicity. That is a (the?) point off in the distance. Movement towards that point constitutes “scientific progress.” Thesis: Systematicity is not a necessary feature of scientific language, or a means to an end. It is an end. Something science moves towards, i.e., it is an item for the list of things that scientists value. Thing Valued: Systematicity The big question at this point is: what is systematicity? Is it something like compositionality? I do not know the answer. A mathematician, logician or linguist (maybe a biologist) would be more equipped to give a good one than I. In the next section (2:5) I will make a few relevant points, run through a case, and simply note that the actual connections between actual syntaxes/maps and utterances conditions, do appear to be bit messy. (You just have to look at abandoned scientific languages to see it.) In the following section (2:6), I’ll present a simple theory of language change. 15 2:5 Schema, a State of Imbalance Aside from the natural connotations the term carries, I don’t have an account of systematicity. Still, I’ll do what I can to suggest that language is often very unsystematic. It’s common, while theorizing about language, to engage in a fair amount of idealizing, abstraction, and flat out toy making. That has its time and place. But now we are looking for a point such that, as the language moves (changes) towards that point, this constitutes progress. Naturally, the language is moving away from something as well, a lack of systematicity. It may be that the ugly warts the language is moving away from are to be found in the details, i.e., the things I’ve been ignoring. For example, I have been operating under the assumption that utterance conditions = stimulus conditions, where stimulus conditions are things that stimulate the sensory apparatus. But, utterance conditions ≠ stimulus conditions. Sentences are often uttered (or psychologically endorsed) for a variety of reasons that have nothing directly to do with sensory stimulation. It may be that these other utterance conditions are a potential source of the lack of systematicity in a language. For example, suppose the radical pragmatist wants, as I think she should, a descriptive, non-normative theory of logic/entailment. She might then distinguish between two types of utterance conditions. An utterance is stimulated, if it is caused by non-linguistic reality impugning on the sensory apparatus. An utterance is provoked if it is caused by the utterance of another sentence. Then, an utterance, U-1, may be stimulated by reality impugning on the sensory apparatus. Utterance U-2 may be provoked by the utterance of U-1. And, U-2 may also have a stimulus condition, in particular, a stimulus condition that does not obtain. So, this is one potential source of lack of “systematicity.” Let’s look at some cases. I’ll suppose, for the sake of illustration, that we describe systematicity in another’s language by giving a recursive theory designed to generate instances of U-schema. So construed, a theory of systematicity must recursively assign utterance conditions to sentences. Assumption: The relationship between a syntax/map and reality (utterance/stimulus conditions) is systematic iff it can be described recursively13 in the meta-theory. For the fragment of one’s own idiolect constituting her own theory, the connection between utterances and reality (utterance conditions) will always come out looking very clean, if not systematic, in her metatheory. This is because the U-schema instances will look disquotational. 13 The term ‘recursively’ is really functioning as a place holder here. It maybe that something else is wanted here. 16 Schema Instance: ‘Snow is white’ is uttered iff snow is white. Schema Instance: ‘E = MC2’ is uttered iff E = MC2 But what about fragments of your language that are not part of our theory? Consider the schema instance below. Schema Instance: ‘Everest is in Texas’ is uttered iff [ ] Supposing the relationship between utterances and utterance conditions is causal, we can simply put ‘Everest is in Texas’ in the bracket on the right of the schema. We wouldn’t say ‘Everest is in Texas’ unless Everest is in Texas, we think to ourselves. But, suppose we encounter “a native” who actually utters, ‘Everest is in Texas’. We cannot put ‘Everest is Texas’ in the slot on the right of the schema above in that case. Since Everest is not in Texas, those utterances are most definitely not caused by Everest’s presence in the state of Texas. What is one to do? The short answer is that you have to figure out what actually causes instances of ‘Everest is in Texas’ for the relevant “native.” If that seems overly charitable, keep in mind that we are currently looking for the point of inquiry. If we are to claim that the native’s language systematically related to reality, then we have to assign the sentence an utterance condition recursively. Let’s consider another case. Schema Instance: ‘Phlogiston is being released in the process of combustion’ is uttered iff [ ]. What are we to put here? Again, whatever causes the utterance goes there. Now consider the instance below. Schema Instance: ‘Phlogiston was never released in the process of combustion’ is uttered iff [ ]. This pair is interesting. It seems that the utterance conditions, the stimulus conditions let’s say, of both sentences actually obtained, for speakers of the phlogiston idiolect. It’s important to be clear about this. In a descriptive, non-normative theory of utterance conditions, you are held strictly accountable to what is uttered. You don’t get to vilify acts on the ground that they are “false.” They were both uttered. Let’s suppose further that the sentences are logically incompatible. Perhaps we can introduce don’t-utterit conditions. Then, ‘Phlogiston is being released in the process combustion’ would be a provocation condition for not uttering ‘Phlogiston was never released in the process of combustion’. Nevertheless, the stimulus condition for the latter sentence also obtained. So construed, the utterance condition and the don’t-utter-it condition both obtain (contingently, as a result of how non-linguistic reality turns out to be). Furthermore, it may be that the stimulus condition for the first instance is a proper part of the stimulus 17 condition for the second, “incompatible,” instance. That is a little strange. I’m inclined to find it undesirable as a feature of language, a little messy. Consider also the following schema instance. Schema Instance: ‘[ ]’ is uttered iff oxygen is released being released in the process of combustion. What is the “phlogiston native” to put here? I’ll dispense with the project of giving examples that, just maybe, point to a lack of some unspecified form of systematicity. 2:6 Describing and Predicting Language Change “Most important was an idea that was picked up and developed by James and above all by Dewey: the recognition that evolution, at the level of species, and learning, at the level of individuals, share a common selectional structure.”(Brandom, Intro to Perspectives on Pragmatism, p. 5) In this section, I’ll present a very simple theory of language change. I think it is wrong in letter, but “right in spirit,” if that makes any sense. It is very simple and as such, will constitute a nice toy theory. I’ll work with the assumption from the previous section, and a proviso to that assumption. Assumption: The relationship between a syntax/map and reality is systematic iff it can be described recursively in the meta-theory. Proviso: Since a theory of reality is part of the meta-theory, the theory of systematicity is only as good as the theory of reality. Faux Davidson used a principle of democracy/charity to jump the chasm between the natural and the normative. Democracy determines and privileges the normative semantic map which applies across utterance idiolects, i.e., determines truth conditions. Instead, I will appeal to democracy to determine how systematic the native’s language is. But, it’s democracy within an idiolect, not across idiolects. In short, I will pick the recursive meta-theory that assigns utterances to as many sentences of “the native’s” language as possible.14 With respect to some “native” sentences of some languages, we may not be able 14 There is really no need to privilege any particular map, so long as it is naturally construed. So, you might look at backwards maps, forwards maps, sideways maps, diagonal maps, etc. The more you look at, the better your 18 to assign them utterance conditions recursively. In that case, we accuse “the native” of a lack of systematicity. In doing so, we are not jumping the chasm from the natural to the objectively normative. We do our best to find out whether their language is, in fact, systematically connected to reality. If it appears to us that it’s not, then we say that it’s not. That’s it. If the natives are scientists, then they also value the items on the list of things that scientists value. Systematicity is on that list. So, we have grounds for shared, though subject, evaluation. Seen in the light of the present naturalist theory, all that this means is that we are headed in the same direction, towards sytematicity. Now suppose that the natives most democratic and charitable U-theory of our language fails to assign Uconditions to some of our sentences, those involving ‘oxygen’ perhaps. In that case, we accuse them of having a lack of “expressive power.” There is an aspect of reality which their map does not relate them to at all. The paragraphs above are meta-meta-theoretical. They describe what meta-language descriptions of systematicity and lack of expressive power look like. We still need to describe how object languages change. Language change does not, on the present view, occur via stipulations in a meta-language (see last two paragraphs of this section). Language change occurs via random mutation and selection towards systematicity.15 Here are a couple of rules for predicting language change. Predicting Theory/Language Change: 1) If your democratic recursive U-theory for their language16 results in gaps on the right hand side of a schema instance, their theory is unsystematic and you predict language change towards systematicity (via mutation and selection). 2) If their democratic recursive U-theory for your language results in gaps on the right hand side of a schema instance, their language is expressively impoverished and you predict change towards increasing expressive power. It should be noted that these predictions will rely on theses about the relevant agent’s intelligence. Here’s a quick definition. theory of “their language,” or language swarm, will be. Then you simply note which connections are systematic and which are not. That is your science, and it involves no privilege. 15 Mutation and selection are the closest correlates to context of discovery (mutation) and context of justification (selection). 16 We may want to replace ‘language’ with ‘theory’ in these rules. So construed, you wouldn’t accuse them of a lack of systematicity if they do not endorse the sentence which your recursive theory does not assign an utterance condition. 19 Intelligence: Intelligence = rate of language change (towards systematicity). Theses about intelligence will tell you, roughly, what the rate of change is expected to look like, how far it can go, etc. They should not be taken for free, as assumptions.17 Rather, they will be supported to the extent that the theory containing them makes successful predictions about change. Your evidence18 that your meta-theory of their language is correct comes in three kinds. First, since your theory is part of your meta-theory, your evidence for your theory is evidence for your meta-theory. Second, your meta-theory describing the laws (utterance conditions) of the natives’ language is to be judged by their actual utterances. Third, your theory of their degree of systematicity is to be judged, together with your theory of their intelligence, by how well it describes/predicts changes to the native’s language. An analogy may help here. Suppose you have a theory of frog behavior. Your theory of frog behavior will be judged by your ability to predict frog behavior. This is analogous to your theory of utterance conditions for the native’s language. If you go on to make the additional claim that the frog is not interacting systematically with its environment, then you predict evolutionary change in frogs. This will require a thesis about the “evolutionary intelligence” of frog DNA, analogous to the claim about the intelligence, or rate of language change, of the “native.” How fast do mutations occur? How fast are these selected for, etc?19 Finally, I’ll issue a warning on the limitations of the theory. A consequence of the fact that you use your first order theory of non-linguistic reality in your meta-theory of language change is that your meta-theory of language change can be better than your theory of reality. Hence, you do not use your meta-theory to predict instances of changes to your own theories. You use it to predict changes to those that are worse, according to your theory. Hence, you don’t increase your understanding of non-linguistic reality by studying your language. “It follows that in making manifest the large features of our language, we make manifest the large features of reality. One way of pursuing metaphysics is therefore to study the general structure of our languages.”(Davidson, The Method of Truth in Metaphysics) 17 On the present view, you should take for free only what is taken for free in the paradigm sciences. So, you take it for free that there is an external world, for example. 18 The radical pragmatist shouldn’t use normative epistemic terms. Reads better though. 19 Ruth Millikan has emphasized the naturalist view of language as a biological object. See Biosemantics, the Oxford Handbook in the Philosophy of Mind, ed. McLuaghlin (available on-line). There are salient differences and similarities between her account and this pragmatic account. 20 It’s also not possible, according to the present view, to adjudicate your object language disputes with “the natives” by moving into the meta-language and presenting with specifically meta-theoretical evidence to the effect that their lack of systematicity. One agent’s lack of systematicity is another agent’s lack of expressive power. So, while scientific values are values shared by scientists, this does not mean that agreement without language change is possible. You can present them with the evidence (stimulus conditions, to avoid normative epistemic terminology) for you first order theory. Perhaps you’ll leave them speechless and they’ll evolve. But this is first order business, not meta-theory. 2:7 What’s So Special About Us? The reader may have noticed that I’ve made language possession very easy. A syntax, some behavior (a map), reality, with no normative component, and with “systematicity” between the syntax/map and reality optional. That reflects a choice about how to use the term ‘language’. I am choosing to use it in such a way that it does not capture what is distinctive about us. While having language is not special or distinctive, I do think our language has two features that are special and distinctive. I’ll start with a brief discussion of a feature of language that is not unique to us. Science, as I have construed it, involves the emergence of order from chaos. The emergence of order from chaos within a system that has causal and random components is fascinating. But it is not unique to us. One sees this in ant colonies for example. So, on this view, not even science is unique to us. To be clear, this is a choice about how to use the term, as a technical term, within the present meta-theory. It is not a claim involving the standard the use of the term. Again, we are up to our waist in violations of the commonsense metatheory, and violations of the a priori entailments between terms occurring in that theory. While the chaos-to-order phenomenon occurs regularly I nature (that’s my impression anyway), I do not think it is, in the typical case, well-conceived of as an end. Rather, it is means to an end, finding food in the animal planet video I saw on ants. The present suggestion is that with respect to the phenomenon of language use in science, the emergence of order is an end in itself.20 Scientists aren’t like ants, where the behavior evolves towards systematicity in response to food stimulus, ignoring most other stimulus (e.g. fire, according to animal planet). Scientific language evolves towards systematicity in response to any stimulus, food, fire, wind, rain, magnifying glasses, etc. The language of science wants to mesh with reality into a giant web of systematicity. This would be the closest thing to the traditional issue of expressive capacity, within the present account. So, we might say, roughly, that the expressive capacity 20 This is so even if we adopt a forward looking map. 21 of our language is unique, thereby linking the present answer to a more traditional and familiar answer to the question of what makes us unique. The second issue is the rate of adaptability. What makes us special is intelligence, i.e., the rate of language change. We, like all creatures, are a swarm of languages. But, the rate at which one of our languages, the spoken language (or signed, or brailed), adapts is unlike anything else seen in nature.21 It is more plastic and adaptable than any possessed by our biological peers. In essence, natural selection has selected something very special for us. She has selected for one of the languages of which we are comprised to be hyper-selectable. Conclusion The purpose of this essay was to establish the point of inquiry according to pragmatism. According to the present theory, much scientific inquiry, i.e. theory change, is language change. It proceeds in accordance with scientific values. That is to say, within science, unsystematic language evolves towards systematic language. That is an end to be achieved, not a means to an end, nor a necessary feature of language as such. 21 This is just a guess. I don’t actually know. 22