Philosophy of Language: Wittgenstein Professor Boedeker

handout on On Certainty

The Tractatus

In the Tractatus, Wittgenstein distinguished between two different kinds of propositions:

  1. propositions with sense, which are true in some possible situations, and false in others.
  2. propositions without sense, i.e., senseless propositions, which are either
  3. tautologies, which are true for every possible combination of truth-value assignments to their base propositions.
  4. contradictions: the negations of tautologies.

It only makes sense to say that someone knows (1) propositions with sense, and not (2) senseless tautologies (4.461, 5.1362).

Furthermore, only tautologies are certain (4.464, 5.525). This is not because of a speaker’s subjective, psychological state, but rather because tautologies cannot be false; they are necessarily true.

Whether a proposition has or lacks sense depends on the form of its base propositions, i.e., the ranges of their possible truth-value assignments. (See the handout on the Tractatus on sense, reference, and form.)

The form of a proposition’s base propositions is shown in the rows of its truth-table, by each row representing a possible combination of truth-value assignments to its bases, and by all the rows representing just the possible combinations of truth-value assignments to its bases.

On Certainty

In On Certainty, Wittgenstein distinguishes between two different kinds of utterances, i.e., assertions of a proposition by a person in a context:

  1. empirical hypotheses
  2. frame propositions that hold fast for me; also called “hinge”, or “axis”, propositions. (Virtually all of the propositions canvassed by G.E. Moore in his essays “A Defense of Common Sense” and “Proof of an External World” are examples of such propositions: “Here is one hand”, “I have a body”, “I have never been far from the surface of the earth”, “The earth is at least 100 years old”, etc.)

As in the Tractatus, it only makes sense to say that someone knows (1) empirical hypotheses, not (2) frame propositions.

Unlike the parallel distinction in the Tractatus, however, this is not a distinction among kinds of propositions. In fact, the very same proposition might be an empirical hypothesis as uttered by one person in one context; whereas it might be a frame proposition when uttered by a person (either the same or different) in another context. One reason for this is that Wittgenstein in On Certainty counts as frame propositions some of what the Tractatus would count as propositions with sense. (In fact, all of the propositions canvassed by G.E. Moore in his essays are propositions with sense in the terms of the Tractatus.)

In On Certainty, Wittgenstein expands his view of language beyond the merely semantic dimension (in which the meaning of an expression is just its contribution to the truth-conditions of propositions in which it occurs) to the pragmatic dimension, i.e., that having to do with doing things with words beyond merely using them to state facts.

A “linguistic phenomenology” of claims to know

Wittgenstein’s analysis focuses on the pragmatics of knowledge-claims. That is, he describes the “language-game” in which we use the phrases “I know”, “you know”, “s/he knows”, etc. In so doing, Wittgenstein breaks with the whole tradition of epistemology. Traditional epistemology (and mainstream epistemology as practiced today) defines “knowledge” as appropriately justified true belief, and assumes that appropriate justification is some special relation between a mind, or subject S, and a proposition p. Traditional epistemology assumes that this relation can be defined, and attempts to define it. (For some examples, Plato thought that justification is a matter of a soul’s intuiting Forms. Descartes held that justification is a matter of a thinking substance’s “clearly and distinctly” perceiving an idea in the mind. Modern “coherentists” hold something like the view that someone S is justified in believing a proposition p if p is consistent with a sufficient number of S’s other beliefs. Etc., etc.) In merely restricting his remarks to our use of claims to knowledge, Wittgenstein refuses to play the game of traditional epistemology. That is, he is not committed to holding that “knowledge” refers to any particular definable relation between a mind and a proposition. In general, Wittgenstein thinks that traditional epistemological problems, such as skepticism, arise because we “reify” (i.e., try to make into a thing) our talk of knowledge by assuming that knowledge is a thing with an essence. An actual “linguistic phenomenology” of the use of our claims to know demonstrates that traditional epistemology is predicated on a misunderstanding of our use of words.

Here is an important basic observation about the pragmatics of knowledge-claims:

When someone S in a given context C claims to know some proposition p, S commits him- or herself to

  1. responding with reasons that are good in C to objections that are relevant in C to S’s claim to know p

and

  1. abandoning the claim to know p if S cannot meet condition (a).

(Note that no such commitments are made when someone simply expresses his or her personal beliefs, including those of which they are subjectively certain. Religious beliefs might be an example of such subjective certainties.)

The social context of argumentation determines what counts as a good reason and a relevant objection under normal conditions. In holding this view, Wittgenstein holds a contextualist view of justification. Note that several different social contexts of argumentation may be in play at the same time, and that the same speaker might be involved in different contexts at different times. For example, surely different standards of justification operate in everyday discourse, scientific journals, examinations for awarding academic degrees, etc.

The social context of argumentation also determines the difference between (1) empirical hypothesis and (2) frame propositions that hold fast. (In this way, the social context of argumentation plays an analogous role to logical form in the Tractatus.) That is, empirical hypotheses are ones to which the social context of argumentation permits relevant objections under normal conditions. For this reason, it makes sense for someone to doubt an empirical hypothesis, since s/he may have a relevant objection to the hypothesis. Also for this reason, it makes sense for someone to claim to know an empirical hypothesis, since one can make good on one’s commitment to respond with good reasons to relevant objections.

Frame propositions that hold fast, on the other hand, are ones to which the social context of argumentation permits no relevant objections under normal conditions. Frame propositions are those that persons are entitled to believe without justification. For this reason, it does not make sense to doubt a frame proposition unless one also has a reason to doubt that the conditions are not normal. Similarly, it does not make sense under normal conditions to claim to know such a proposition, since it would not be possible to make good on one’s commitment to respond to relevant objections to it; after all, there are no good objections to frame propositions under normal conditions. (This is the basis of Wittgenstein’s criticism of Moore’s defense of common sense by proving the existence of the external world. Moore wrongly assumed that it did make sense under normal conditions to claim to know frame propositions.)

Like tautologies in the Tractatus, frame propositions are objectively, not subjectively, certain. This is because whether or not a proposition is a frame proposition has nothing to do with whether or not an individual happens to be sure of it. Rather, it is a matter of the social context of argumentation. Furthermore, just as the Tractatus maintains that logical form shown in a truth-table, so too is the social context of argumentation shown in which propositions persons in a given context take as certain frame propositions, as empirical hypotheses, as relevant objections, or as good reasons.

To see how this works, imagine someone who claims that a frame proposition is false without giving a good reason to believe that the conditions are not normal. It would be perfectly right to question whether such a person actually understands the meanings of the words s/he is using. After all, we teach the meanings of words to someone by uttering frame propositions (although not by claiming to know them). What better way is there to get someone to understand what “hand” means than by pointing to a hand and saying “Here is one hand”? Or how else should we explain what “physical object” means but by pointing to various physical objects and saying “This is a physical object”? (Wittgenstein calls such statements used in teaching ostensive definitions.) In this way, someone must believe such frame propositions under normal conditions in order to have mastered the meanings of the terms involved. Since no one can express a doubt using words whose meanings s/he does not understand, we have no reason to take seriously someone’s apparent doubt of a frame proposition without having a good reason to believe that the conditions are not normal.

How to turn a frame proposition into an empirical hypothesis

Now if Wittgenstein held that the social context of argumentation were something absolutely fixed and impervious to criticism, then it would be possible to accuse him of relativism (in which we would have good reason to believe that neither of two competing contexts of argumentation contained any more truth than the other). It is true that Wittgenstein’s initial response to the skeptic in ##1-65 of On Certainty seems to leave him open to the threat of relativism. This response is based on claiming that the proposition accepted by both skeptic and realist – that if we know that here is one hand and here is another, then we know that physical objects exist – is just nonsense. This is because both the propositions “here is one hand and here is another” and “physical objects exist” are frame propositions, and it is not appropriate to claim either to know or to doubt frame propositions under normal circumstances. But after all, the relativist can argue, people in the past have held many frame propositions that have turned out to be false (e.g., that the earth was flat, that the earth was at the center of the universe, etc.). Thus what reason do we have to think that we are in any better position now? In other words, Wittgenstein’s initial response to the skeptic looks somewhat dogmatic. To see how we can modify this response into a subtler one that avoids the relativist’s charge of dogmatism, let us first examine two examples of how there can be reasons to change what counts as a frame proposition in a given context.

The tie shop:[*] Imagine a client of a tie shop asking to purchase a green tie. The clerk sells the client a tie that appears to be green. When the client takes the tie into the light of day, however, the client realizes that the tie really is blue. It only appeared to be green in the tie shop because of strange yellow lighting. Naturally, the client will demand a refund or an exchange. When he does so, the clerk is at first entitled simply to repeat his claim that the tie is green. After all, persons with normal perceptual systems who know the meaning of color terms are entitled to make claims about colors without further justification – under normal perceptual conditions. Such beliefs about colors are frame propositions. The problem is that the conditions are not normal. What the client needs to do is to give a good reason to believe that the conditions in the tie shop are not normal. To this end, the client can point out the yellow filters on the lights, have the clerk bring several ties from the shop out into the light of day, and so on. Once the clerk concedes that the conditions are not normal, he has thereby also conceded that his belief that the tie is green is not a frame proposition (to which there can be no relevant objections), but rather an empirical hypothesis (to which there can be relevant objections). At this point, the clerk theoretically has two options. First, he could abandon his belief that the tie is green, and simply admit that the client is right in saying that it’s blue. Second, however, particularly stubborn clerk might claim to know that the tie is green, even though the conditions in the shop are not normal. It’s difficult to see how this second option could be successful, however, since in claiming to know that it’s green he commits himself to giving good reasons against relevant objections – and the strange yellow lighting of the shop is certainly a good objection to the claim to know that the tie is green.

Scientific revolutions: Thomas Kuhn (in his 1962 book, The Structure of Scientific Revolutions) has given the title “scientific revolutions” to major shifts in scientific “paradigm”, which includes the basic concepts and methods of a science. (Examples of scientific revolutions include the Copernican revolution in astronomy, the overthrow of Aristotelian physics by Newton, Darwin’s theory of evolution in biology, Einstein’s theory of relativity in physics, and the introduction of quantum physics.) Put in Wittgensteinian terms, a scientific paradigm is a social context of argumentation that defines a particular scientific discipline at a given time. A scientific revolution is opposed to what Kuhn calls “normal science”, which characterizes the activity of the great majority of scientists. In normal science, scientists work within the accepted “paradigm” for doing science, and solve particular problems, or puzzles, that come up within that paradigm. (An example of normal science would be adding new epicycles to the Ptolemaic, or geocentric, system of astronomy that was dominant through 1500.)

Let’s assume that a major goal of science is to have theories that account well for the important observational data. (We need the caveat “important” because there are always some, usually fairly insignificant, observational data that a paradigm cannot account for well.) This goal allows us to distinguish periods in which a scientific paradigm is functioning under normal conditions from periods in which these conditions are abnormal. Note that here, unlike the tie shop example given above, the relevant conditions of normalcy or abnormality are not perceptual conditions, but rather the conditions of the social context of argumentation itself (i.e., the scientific paradigm). Let us say that a scientific paradigm is functioning under normal conditions just in case its theories account well for the important observational data. Meeting this goal is what makes normal science normal. In normal science, the paradigm’s basic theories are frame propositions. That is, they are beliefs that scientists are entitled to believe without further justification.

For Kuhn, the transition from normal science to a scientific revolution is always preceded by a crisis. A scientific paradigm is in crisis just in case there are important observational data that the paradigm cannot account for well. (Kuhn calls such data “anomalies”.) That is, a scientific paradigm is in crisis just in case it is functioning under abnormal conditions. In pointing out important observational data that the paradigm’s theories cannot account for well, someone thereby offers a good reason that the paradigm is not functioning under normal conditions. This is analogous to the client in the tie shop example pointing out that the perceptual conditions in the shop are not normal. When the client points this out, the clerk’s belief that the tie is green changes from being a frame proposition to being an empirical hypothesis. Similarly, pointing out that a scientific context of argumentation is not normal changes the paradigm’s basic theories from frame propositions into empirical hypotheses. This sets the stage for being able to make objections to these basic theories that are relevant within the current social context of argumentation. In times of scientific crisis, the current social context of argumentation actually allows for relevant objections to aspects of itself. One such relevant objection to a basic theory is that they can account for important observational data only by adding ad hoc theories, i.e., theories lacking any principled reason. (The Ptolemaic practice of adding epicycles within epicycles is a good example of an ad hoc theory.)

For Kuhn, a scientific crisis is resolved by a scientific revolution, in which a new paradigm is proposed whose basic theories can account for the observational data better than the basic theories of the old paradigm. Once the new paradigm is accepted, its basic theories become frame propositions, i.e., propositions that its practitioners are entitled to believe without justification under normal conditions. Certain basic theories of the old paradigm, namely the ones that are contradicted by those of the new paradigm, become the opposite of frame propositions (analogous to contradictions in the Tractatus), i.e., propositions that the practitioners of the new paradigm are entitled under normal conditions to reject without justification.

There is an important difference between the case of the tie shop and the scientific revolution. In the case of the tie shop, the social context of argumentation remains unchanged, but abnormal perceptual conditions make what would have been a frame proposition into an empirical hypothesis to which there can now be relevant objections. In the case of the scientific revolution, however, the abnormal conditions of the social context of argumentation make the basic theories of that very social context of argumentation itself, which were formerly frame propositions, into empirical hypotheses. In a period of crisis, scientists are therefore able to make relevant objections to the basic theories of their paradigm.