This commentary is part of The Atlas Society's 1999 online "CyberSeminar" entitled " The Continental Origins of Postmodernism ."
Reading through Will Wilkinson’s excellent essay on Rorty, I found myself stalled at the point where Will presents a general argument against any “thoroughgoing pragmatist theory of truth.” Here is the argument.
“Suppose we have two beliefs, P and Q,...and we want to decide which is most advantageous, which is best for us to hold. And suppose Rorty says that P is better for us to hold than Q. Immediately we’ll want to ask, ‘Is P’s being better for us to hold than Q a bona fide fact, i.e., is it really TRUE that P is better for us to hold than Q.’ ...[In consequence, the pragmatist] is saddled with a vicious regress. If it is pragmatically true that P is good for us to believe, then it is good for us to believe that it is good for us to believe that P. And if that is the case, then it is good to believe that it is good to believe that it is good to believe that P. And so on. The regress is vicious because if one begins on the regress it becomes impossible to give any content to claims about a belief’s goodness or practical advantage....[Such a claim] fails to be a claim at all.”
Let us examine this argument, and see what it has to teach us about the relations between pragmatism and the other two traditional theories of truth (the correspondence and coherence theories), about the philosophical motivations behind these theories, and about their respective abilities to withstand regress arguments such as the above. For purposes of this essay, I will continue Will’s lead and try to make my claims applicable to any thoroughgoing pragmatism, not just Rorty. I will however change the locution “is good for us” to the syntactically simpler “is useful.” Also, in what follows it will be convenient to choose some definite but ordinary empirical statement for P. Let us say that P is the sentence “There’s a thorn tree in the garden.” Q can be simply “Not P.”
The first thing that occurred to me about Will Wilkinson’s argument is that, if it’s a question of “deciding which” of P or Q is true, that is, a question of confirming truth, then (my beloved!) correspondence theory of truth is at least as susceptible to a regress, if not more so. For, how will we decide whether P is true? Being good empiricists, I suppose we will check the garden and see. So we can then say, “I saw a thorn tree in the garden” (P’). But how did we decide that P’ is true? Was the light adequate? Did I look carefully enough? Was merely looking enough, or should I have felt it as well? Some fakes can be extremely realistic; perhaps I should have taken a sample from inside the “tree” for analysis. Is there really a root structure? Each of these questions represents the need for a P’’, e.g., “The light was adequate to see the tree in the garden” (P1’’); “I looked carefully enough to see a thorn tree in the garden under normal viewing conditions” (P2’’); etc. But there would appear to be no limit to this sort of double-prime level statements. Therefore it is not possible to construct a conjunction of all the Pi’’ from which P’ follows. Moreover, clearly for each single Pi’’ there will recur a further set of equally unlimited Pij’’’, and so on and on, in a combinatorial explosion.
Who ever imagined that knowledge would not be fallible?
So even on the correspondence theory of truth there is a “verification regress” when it comes to deciding which statements are in fact true--at least when it comes to statements about the empirical world. Indeed, isn’t this the very problem that led people to embrace non-correspondence theories of truth in the first place? Rorty thinks it is the principal problem even today (Objectivity, Relativism, and Truth [ORT] 6; his phrase: “no way of formulating an independent test of accuracy” [emphasis his]). I believe coherence theorists like Hegel have traditionally been interpreted as reasoning, essentially, “if only empirical truth were ‘inside’ the system like mathematical truth is, then logical consistency alone would determine what is true, and there would be no need of ‘an independent test of accuracy.’” Pragmatism works the same way: by bringing truth inside the system of sentences (or ideas, or whatever), pragmatism attempts to eliminate the need of an independent test of accuracy. (I will say more below about how pragmatism brings truth inside the system of sentences.)
But is the verification regress actually a serious problem for the correspondence theory? I don’t see why it should be. In fact we justify statements like P by reference to statements like P’ all the time in everyday life without difficulty. When a greater degree of justification is needed, we commonly appeal to statements like P1’’ and P2’’, adding further such statements to meet specific challenges or concerns as they arise [note 1]. Without launching into an elaborate theory of evidence and justification--which I confess to lacking anyhow--let me just say that I think we can conceive this process as essentially similar to the process whereby evidence is brought to bear in a case before a law court. That is, like the process of evidence and justification in a law court, the process of justifying statements such as P by reference to perceptual judgments like P’, P1’’, and P2’’ is similarly fallible and perhaps even subject to certain biases, but also--similarly--fundamentally unproblematic.
What is supposed to start the verification regress for the correspondence theory? What would make us suppose that P’ is inadequate without, say, P1’’? It is our knowledge that we can more easily make perceptual mistakes in poor lighting, so that, if it is especially important to be right about P’, we ought to bring in P1’’ as well. But what is the nature of the “inadequacy” attributed to P’ if P1’’ cannot be produced? It is like one side of a court case missing evidence that would be expected to be available and whose lack is therefore damaging. It is not that of a deductive conclusion lacking a necessary premise. Therefore, P’ lacking P1’’ is not like a promissory note that can’t be redeemed. Quite the contrary, it is critical to see that P’ has epistemic weight all by itself. To see something is to have evidence of it, and indeed in the scenario I have described P’ is the main evidence for P. The Pi’’ statements (as well as the triple-primes, quadruple-primes, and so on) are important only as supplements to P’ (but also as potential falsifiers, if for example closer inspection reveals the “tree” to be a fake). Notice in this connection that P’ continues to support P even if P1’’ is lacking--even if the contradictory of P1’’ is true. One sees what one sees, after all, even in very poor light.
In short, since the Pi’’ are not required in order that P’ constitute evidence for P but only in order to investigate further the status of P’, and since such further investigation is not automatically necessary but only becomes necessary when specific problems arise about P’ or when it becomes critically important to be right about P’, there is nothing vicious about the verification process. But isn’t this just to admit that perceptual judgments like P’ are fallible? Yes, but why is that a problem? Who ever imagined that knowledge would not be fallible? Plenty of philosophers! Plato, for starters. I believe that the requirement that knowledge be indefeasible is a key error which has brought many a philosophical program to grief, and which is at the root of anti-correspondence theories of truth.
Well, so if the verification regress is not vicious for the correspondence theory, what about Will Wilkinson’s regress about deciding the usefulness of P in pragmatism? Do we have to decide whether it is useful that P is useful, useful that it is useful that P is useful, and so on? Is there really a problem here? Since “usefulness” without further explanation is a tad vague, I propose to begin by asking the same question of the coherence theory instead. As I have already indicated, coherentism and pragmatism, qua forms of anti-correspondence, are essentially similar strategies motivated by essentially the same perceived problem. So we can ask, upon being told that P is logically coherent with all our other beliefs, whether the coherence of P itself coheres with all our other beliefs. But offhand there seems no logical problem about it. (The problem would be if we tried to say that P’s coherence was not coherent.) More fundamentally, it is hard to see why we should have to ask about the coherence of P’s coherence to start with. For the statement that P is coherent is perfectly intelligible. The standard of coherence is referred to specifiable rules of logic, applicable to any statement in the system. This is just what I meant by saying that coherentism avoids the supposed need for an “independent test of accuracy” by bringing truth inside.
I can’t see that the situation is any different when we change the standard from coherence to usefulness, other than that usefulness is slipperier. But usefulness presumably refers to being instrumental in achieving goals. Some appropriate set of goals rather than just any, no doubt, but the point is that once again the criterion is determinable inside the system. Therefore the statement that P is useful is in principle decidable and does have content. So, although the question whether P’s usefulness is useful is intelligible, there is no particular reason to ask it and a regress of such questions would not be vicious.
For that matter, if the question were about the correspondence of P (does P’s correspondence to the facts itself correspond to the facts?), the question would be askable though rather pointless, in just the same way. In fact, on all three theories it appears to me that the truth of P actually guarantees the truth of the statement that P is true as well as all subsequent recursions of this reflexive statement. Which brings me to say, what has probably been apparent to most readers for some time, that there are really two distinct kinds of regress involved here. The first is the verification regress, which arose when Will Wilkinson said we wanted to decide whether P is true (i.e., useful, coherent, or corresponding). The second is a reflexive regress on the assertion that P is true, which arose because that is the regress Will actually develops. We saw that the verification regress affects the correspondence theory because, on the correspondence theory, truth is a relation between a given sentence and the facts and consequently not itself a part of the system of sentences. We found that the view that this is a problem results from the (false) premise that knowledge must be indefeasible; nevertheless it has historically been regarded as an insuperable difficulty, leading to attempts to bring truth inside the system of sentences, two important of which attempts have been coherentism and pragmatism. As for the reflexive regress, we found that it is not the least vicious for any of the three theories under review, since for all three the sentence “P is true” can be explicated in the given theory’s own terms and particularly without recourse to the question whether it is true that P is true.
But it would be going too far to say that there is nothing at all in the argument Will Wilkinson presents. I think the core of Will’s point lies in his question, “is P’s being better for us to hold than Q a bona fide fact?” This question suggests that pragmatist truth is not really inside the system of sentences after all. For it is a matter of fact whether P is useful for us to hold--a matter of fact outside the system of sentences. Therefore the problem that supposedly plagues the correspondence theory is not really avoided. However, the question is not whether there is a fact of the matter about the usefulness of P. The question is whether the pragmatist can articulate his theory of truth and adhere to it without recourse to any extralinguistic facts about usefulness. I think the answer is that he can. Usefulness is a matter of promoting the achievement of goals. Goals are stateable in language, as are outcomes and whether outcomes measure up to goals. The usefulness of propositions such as P can be assessed by measuring whether their adoption leads to achieving goals (such as predicting future observations). The scheme is similar to instrumentalism in the philosophy of science.
The comparison with instrumentalism, however, raises just what I myself see as the key difficulty for pragmatism. Instrumentalism claims that the theoretical objects referred to by scientific theories (atoms, forces, charges, and the like) should not be taken literally but rather as convenient fictions; scientific theories do not make discoveries about unobservable nature but instead only enable us to make reliable predictions about future observations. (Once again, notice that the motivation for instrumentalism is to avoid the need for an “independent test of accuracy” concerning the unobservables.) I think that the pragmatist is driven ineluctably to the same conclusion--that words and sentences do not refer to extralinguistic entities--but this time for all entities, not just unobservables. For suppose that P is taken to refer to a factual tree being in a factual garden. Once this is admitted, it is inevitable that the question arises whether the putative facts to which P refers obtain in reality. For why do we refer to extralinguistic facts at all if we are not interested in them? Therefore, to admit that sentences refer to extralinguistic facts is to admit that the pragmatist (or coherence, for that matter) theory of truth is inadequate. It follows that to insist that the pragmatist (or coherence) theory of truth is adequate ultimately requires one to deny that sentences refer to extralinguistic facts [note 2].
Of course, we all know by now that this consequence, which the uninitiated might regard as a reductio of pragmatism, is not only accepted but insisted upon by Rorty. “[T]here is no sense in which any of these descriptions [e.g., scientific, poetic, or political] is an accurate representation of the way the world is in itself. These philosophers [whom Rorty recommends] regard the very idea of such a representation as pointless” (Contingency, Irony, and Solidarity [CIS] 4). Again, “The pragmatist...does not think that his views correspond to the nature of things” (ORT 23). Derrida is even more explicit: “Il n’y a pas de hors-texte [There is no outside-the-text]” (Of Grammatology 158). Yet Derrida acknowledges that the “logocentric” idea that there is an “outside” to which language refers is so deeply embedded in “phonocentric” language that we are presently incapable of escaping it. This is the root of the deep incoherence of “phonocentric” language: it presents itself as referring to an antecedent world outside itself, although it cannot so refer.
I see Rorty as very much a Derridean, though one who still tries to make positive claims.
I see Rorty as very much a Derridean, though one who still tries to make positive claims (i.e., that are more than merely emotional or rhetorical). For example, “A liberal society is one which is content to call ‘true’...whatever the outcome of undistorted communication happens to be, whatever view wins in a free and open encounter” (CIS 67). This sounds very much like Rorty thinks he can say what it is to be “free and open,” that is, that he has access to a standard of being free and open that is not just a matter of whatever view “won” in some previous “encounter” (which need not have been itself “free and open” at all and probably wasn’t, having taken place prior to the “winning” of the “free and open” standard). In the clinches, of course, he would deny having any such access. For instance, “it just happened that rule in Europe passed into the hands of people who pitied the humiliated and dreamed of human equality...Socialization...goes all the way down, and who gets to do the socializing is often a matter of who manages to kill whom first” (CIS 184-5, emphasis his). Again, note these remarks about socialization: are they supposed to represent Rorty’s view of how the world is, or only a view to which our own socialization has compelled us? The answer has to be the latter, but that’s not the way the sentence sounds. I think it is a point we can take from both Derrida and Rorty that, so long as we’re doing philosophy, or any other form of “logocentric” inquiry, our sentences have to appear representational. This being the case, the up-to-the-minute intellectual should begin doing poetry vel sim. instead of philosophy. Derrida, and Rorty too, in spite of his inconsistencies, both seem to embrace this point. Derrida apparently puts it into action (see Rorty’s rhapsody on Derrida’s The Post Card in CIS chapter 6). Rorty doesn’t. But maybe he’ll come around.
NOTE 1 In what follows I will be referring to perceptual judgments (e.g., P’) without reference to the acts of perception that justify them. Not only would introducing a theory of perception involve an enormous amount of complexity, I do not believe a proper theory of perception would resolve the “verification regress,” or even address it, since on such a theory perceptual judgments are not indefeasible by other claims. Readers should assume that any perceptual claim such as P’, P1’’, etc., has been properly supported by a percept.
NOTE 2 I am very much aware that this argument is, shall we say, enthymematic. But this essay is already overlong so I won’t try to expand it. The basic argument is that you can’t very well say that language is a representational medium and still say that truth has nothing to do with success or failure in the representational endeavor. Notice by the way that therefore rejection of the correspondence theory implies denial of the representational function of language. The result is assorted doctrines claiming in effect that language is a world by itself, where “meaning is use” (Wittgenstein) or percepts are determined by abstractions, not the other way around (Kuhn). If we combine this with Rorty’s often repeated point that the “language” and “sentences” of contemporary philosophy are just the twentieth century’s version of early modern philosophy’s “mind” and “ideas,” we see that twentieth century attempts to make language a world by itself and the only world to which we have cognitive access are just forms of nineteenth century idealism in new garb. This is one of the main points I have gotten from our current reading of Rorty.
David Potts wrote:
Pyrrhonism was the most thoroughgoing school of skepticism in the ancient world. In a nutshell, the pyrrhonist held nothing--not even that he held nothing. Pyrrhonism distinguished between appearances and judgments about the appearances. For example, there may appear to be an apple, which might appear to be red, to taste sweet, to satisfy hunger, etc. It might even appear, with some thought, that apples are nutritious. But none of the corresponding judgments--that there is an apple, that it is red or sweet, that it does allay hunger, that it is nutritious--is assented to by the pyrrhonist.
The reason is that for any judgment a contradictory judgment can be proposed with an equally strong claim to our assent. Since the competing claims are equally strong no choice between them is epistemically justified. So the pyrrhonist doesn’t choose. About anything. The works of Sextus Empiricus (a pyrrhonist) consist mostly of arguments, one after another, against representative beliefs from all walks of life.
The pyrrhonists found that their policy of believing nothing brought them “ataraxia”--tranquility. That is, rather than struggle to discover the truth about things in order to eliminate their troubles, they found that by relinquishing that very struggle, inner peace followed as it were serendipitously. Or so it appeared. They therefore proposed to live strictly by appearances, including not only sense impressions but the thoughts which naturally flow from them, as well as natural impulses (e.g., hunger), conventionally transmitted attitudes (e.g., that piety is good), and social practices (e.g., Sextus himself was a physician in his “day job”).
Note finally that pyrrhonism, like virtually all ancient philosophies, was not an “academic” theory but was conceived as a way of life and had popular adherents. That is, there evidently were people who attempted to practice pyrrhonism (as for that matter epicureanism, stoicism, platonism, etc.), just as today we can say that Phil Jackson and others practice Buddhism.
I hope the parallels between pyrrhonism and Rorty are clear. Rorty calls himself a “pragmatist,” but I believe traditional pragmatists thought that the pragmatist standard of truth provided objectivity at least in the sense that everybody could be brought to agree about what is most useful. That is, they thought that what is most useful is something we had to discover, not something we make, not just whatever we happen to agree on.
For Rorty, statements do not represent facts at all. Language is better conceived as a tool, like a lever, than as a mirror of antecedent reality. Therefore all the old quarrels and oppositions in metaphysics and epistemology--subject/object, mind/matter, knowledge/opinion, fact/value, absolute/relative, etc.--simply dissolve. In fact, we can stop doing metaphysics and epistemology altogether (ORT 22). We can simply “relax and enjoy” (ORT 44) a life in which we have learned not to worry about whether our statements correspond to reality. This amounts to recommending that we “live by appearances,” whether Rorty would express himself that way or not.
Eyal Mozes wrote:
David Potts questions the validity of Will Wilkinson’s argument, that a pragmatist theory of truth leads to a vicious infinite regress. However, I think David misses an important point regarding Will’s argument.
I agree with David’s analysis of the “verification regress”; but I think David misses the point regarding what he calls the “reflexive regress.” David writes:
“The question is whether the pragmatist can articulate his theory of truth and adhere to it without recourse to any extralinguistic facts about usefulness. I think the answer is that he can. Usefulness is a matter of promoting the achievement of goals. Goals are stateable in language, as are outcomes and whether outcomes measure up to goals. The usefulness of propositions such as P can be assessed by measuring whether their adoption leads to achieving goals (such as predicting future observations).”
But what does it mean to say “I have achieved my goals,” or “the outcomes have measured up to the goals,” or “I have successfully predicted future observations”? Does it mean that the goals have been achieved in fact, in reality? That the outcomes have in fact, in reality, measured up to the goal? That my predictions of observations have in fact, in reality, turned out correct? A consistent pragmatist would have to answer: “no; when I say that acting on belief P will lead to achieving your goals, I simply mean that if you act on belief P, then it will be *useful* for you to believe that you have achieved your goals, or that the outcomes have measured up to the goal, or that you have successfully predicted future observations. I.e., it means that after you act on belief P, then acting on the belief that you have achieved your goals would help you to achieve your goals, or that acting on the belief that you have successfully predicted future observations would lead to successfully predicting future observations.” And of course, the same question can again be applied to the next stage: do you mean that after acting on belief P, and then acting on the belief that I have achieved my goals, then my goals will be achieved in fact, in reality? And again, the consistent pragmatist would have to answer no, continuing the infinite regress.
To put this symbolically: let Q stand for the proposition “your goals are being achieved,” or “the outcomes measure up to the goals,” or “you are successfully predicting future observations.” Then on a pragmatist theory of truth, any claim P can be expanded into: “if you act on the belief that P, then Q will become true”; which can be further expanded into: “if you act on the belief that P, then if you act on the belief that Q, then Q will become true”; further expanded into: “if you act on the belief that P, then if you act on the belief that Q, then if you act on the belief that Q, then Q will become true”; and so on, leading to the vicious infinite regress Will has pointed out.
David Potts compares the pragmatist theory of truth to instrumentalism in science. While there are serious problems with instrumentalism, it is not subject to the same vicious regress, because it is limited to non-observables. E.g., an instrumentalist will deny that the statement “electrons exist” refers to anything in reality, saying that the statement means that if we act on the belief that electrons exist, that will help us to, e.g., build electrical refrigerators that successfully cool their contents. But the instrumentalist can still consistently affirm that the statement "this refrigerator cools its contents” is a factual statement, subject to a correspondence theory of truth, because the refrigerator and its contents are observable entities. This prevents the infinite regression, a solution that is not available to the pragmatist.
In sum, I think the problems with the pragmatist theory of truth are more serious than David Potts claims. I agree with Will Wilkinson that this theory leads to a vicious infinite regress, and cannot be stated consistently while still having an intelligible meaning.
David Potts wrote:
I am very happy to see the always-incisive Eyal Mozes weigh in with a comment. I personally think there has been too little discussion and too little debate among our group, and Eyal raises an interesting and important point. But I do not agree with his analysis.
Eyal says, in response to my claim that pragmatism holds an “internal” standard of truth, as opposed to the “external” standard of the correspondence theory: “But what does it mean to say ‘I have achieved my goals’...? Does it mean that the goals have been achieved in fact, in reality?... A consistent pragmatist would have to answer: ‘no; when I say that acting on belief P will lead to achieving your goals, I simply mean that if you act on belief P, then it will be *useful* for you to believe that you have achieved your goals...”
Eyal is perfectly right to say that there is a reflexive regress if all you can say about “P is useful” is that it is itself a useful thing to believe. But my claim is that the pragmatist can explicate usefulness in terms of goal achievement, where goal achievement is understood as something ultimately subjective such as feeling satisfied or successfully predicting observations. In the language I used in my essay, goal achievement is something the pragmatist cashes out as something “inside the system of sentences.” And I said furthermore that to bring truth inside the system of sentences and thus “solve” the dreaded verification regress is what induces the pragmatist to say that truth is usefulness to begin with.
Therefore the pragmatist can indeed say quite a bit more about the usefulness of P than that it is itself useful.
But, Eyal Mozes asks, is this more, this goal achievement, something “in fact, in reality”? This phrase “in fact, in reality” is unfortunately ambiguous. If “reality” refers to something extralinguistic, then no, the goal achievement in question is not asserted to be “in reality.” It is rather, as I have said, inside the system of sentences. On the other hand, if “reality” refers to the system of sentences itself, then yes, goal achievement is “in reality”--at least, it is if the internal conditions are “really” satisfied. For example, one might find that certain beliefs lead to what seem like good consequences only to find that they conflict with other important goals or that they cause problems in the long run. All this is an internal matter, however, and so not a threat to the pragmatist standard of truth.
In short, the facts the pragmatist wishes not to rely on in articulating a standard of truth are only those outside the system of sentences. But it is not an embarrassment to the pragmatist to talk about the “facts” about the sentences themselves.
Back to Will Wilkinson, "Review of Richard Rorty's 'Solidarity or Objectivity?' and 'The Contingency of Language'"
> Return to the parent page for this 1999 online CyberSeminar, "The Continental Origins of Postmodernism."