Austin on Sense-Data: Ordinary Language Analysis as 'Therapy' more

Grazer Philosophische Studien 70 (2005), xxx–yyy. AUSTIN ON SENSE-DATA: ORDINARY LANGUAGE ANALYSIS AS ‘THERAPY’ Eugen FISCHER Ludwig-Maximilians-University, Munich Summary The construction and analysis of arguments supposedly are a philosopher’s main business, the demonstration of truth or refutation of falsehood his principal aim. In Sense and Sensibilia, J.L. Austin does something entirely different: He discusses the sense-datum doctrine of perception, with the aim not of refuting it but of ‘dissolving’ the ‘philosophical worry’ it induces in its champions. To this end, he ‘exposes’ their ‘concealed motives’, without addressing their stated reasons. The paper explains where and why this at first sight outrageous aim and approach are perfectly sensible, how exactly Austin proceeds, and how his approach can be taken further. This shows Austin to be a pioneer of the currently much discussed notion of philosophy as therapy, reveals a subtle and unfamiliar use of linguistic analysis that is not open to the standard objections to ordinary language philosophy, and yields a novel and forceful treatment of the sense-datum doctrine. J.L. Austin’s attack on sense-data, his lecture series Sense and Sensibilia, is in many respects a model of clarity. The points made are individually straightforward, the structure of argument is neither convoluted nor elaborate, the language is plain and devoid of technicalities. But, even so, the text notoriously puzzles its readers and gives rise to the two questions this paper is to answer: ‘Though not an obscure text, it is certainly a rather strange one: what is it for? What does it do?’ (Warnock 1989, 12). I.e.: What are Austin’s aims? What is his approach? Pursuit of these questions will lead us to a rather surprising discovery: In his later, primarily ‘positive’ or constructive papers, Austin is often thought to have proceeded to establish philosophical claims by addressing the question, what we should ordinarily say when. Indeed, this approach became associated with his name. At least in these early, primarily ‘negative’ or critical lectures, however, we shall see his aims and methods to be, to a large extent, ‘therapeutic’ in a sense sometimes associated with some things done by the later Wittgenstein. The philosophical pay-off of these historical findings is three-fold: The examination of Austin’s aims will familiarise us with a crucial but little noted philosophical predicament that lends ‘therapeutic’ goals a particular point. Study of Austin’s approach will provide us with one viable way of coping with that predicament: We will become acquainted with a little understood form of linguistic analysis that is radically different from the approach ordinarily associated with Austin’s name, and not open to the standard objections against ‘ordinary language philosophy’. Taking this approach a bit further than Austin, we will, third and finally, obtain a novel and particularly forceful treatment of the sense-datum doctrine of perception. 1. Austin’s Therapeutic Aims Sense and Sensibilia (Austin 1962, henceforward: S&S) is a critical examination of the doctrine that we never see or otherwise perceive, or anyhow never directly perceive, material objects but only sense-data (S&S 2). This paper is to explain the, at first sight outrageous, aim and approach Austin states at the outset. Austin is quite clear on what he wants to do — and what not: I am not, then — and this is a point to be clear about from the beginning — going to maintain that we ought to be ‘realists’, to embrace, that is, the doctrine that we do perceive material things (or objects). This doctrine would be no less scholastic and erroneous than its antithesis. The question, do we perceive material things or sense-data, no doubt looks very simple — too simple — but is entirely misleading … So we are not to look for an answer to the question … What we have above all to do is, negatively, to rid us of such illusions as ‘the argument from illusion’ [on which the sense-datum doctrine is frequently based] … — an operation which leaves us, in a sense, just where we began. In a sense — but actually we may hope to learn something positive in the way of a technique for dissolving philosophical worries (S&S 3–5). I.e.: Austin’s chief aim is not to establish any general philosophical claim or theory, but to dissolve the philosophical worry of those who 23 ‘find’ the sense-datum doctrine ‘disturbing’ (S&S 3). Second, he wants to ‘reduce a bit’ the ‘constant repetition of assertions that are … not even faintly sensible’ (S&S 5), in particular, the repetition of answers to the ‘entirely misleading’ question, what kind of thing we perceive: He wants to get the philosophers he addresses to give up looking for an answer to it (S&S 4). We shall see that these statements are to be taken perfectly seriously: Austin addresses philosophers disturbed by a doctrine they find compelling even though it is, he judges, ‘not even faintly sensible’ for them to maintain. He wants, first and foremost, to (i) liberate these philosophers from their feeling of intellectual unease and disquiet, and, ‘besides’, (ii) make them give up the pursuit of bad questions. These aims he intends to pursue by means of a striking approach: Instead of examining the stated arguments explicitly adduced to support the disturbing doctrine, he proposes to proceed by ‘exposing a wide variety of concealed motives’ (S&S 5). The above aims, rather than the approach he proposes, make Austin’s efforts therapeutic. Philosophers frequently think of therapy as necessarily opposed to argument. But this is not so: Even exemplary psychotherapies like cognitive therapy (e.g., Beck 1995) and rational emotive therapy (e.g., Ellis 1994) crucially involve argument and the assessment of evidence (to establish, for instance, whether the depressed patient is really as utterly inept as he thinks). Therapy may proceed by argument. What makes an effort therapeutic are not the (merely supposedly non-argumentative) means employed, but the aims pursued: wanting, first and foremost, not to demonstrate certain facts or conclusions, but to (i) remove distressing or disquieting emotions and (ii) modify dysfunctional behaviour. Therapeutic argument may of course demonstrate conclusions. Therefore, therapeutic and ‘cognitive’ aims are not mutually exclusive, but hierarchically ordered: The therapist wants to demonstrate facts not as an end in itself, but to achieve those two therapeutic effects (e.g., relieve distress and end avoidance behaviour, by demonstrating to the depressed patient that he is nowhere near as inept as he thinks). Accordingly, we can characterise efforts as more or less therapeutic, depending upon the extent (exclusively, mainly, partially) to which they are motivated by the desire to attain therapeutic effects. We shall see Austin’s efforts to be mainly (but not exclusively) therapeutic. A common metaphilosophical dogma would dismiss all therapeutic efforts, and in particular the removal of disquiet, as philosophically 24 irrelevant: ‘Philosophy is the pursuit of truth, in quest of knowledge. Feelings are of no concern.’ Like many edifying phrases, this sounds fine in the abstract but does not stand up to confrontation with specific examples. Consider, for instance, philosophical paradoxes. They typically engender intellectual disquiet: Those who actually believe the paradoxical conclusion are disquieted by the conflict between this conclusion and the common convictions it is at odds with. Those who find the paradoxical argument compelling but are unable to believe its conclusion (a common attitude towards sceptical paradoxes) feel unease in the face of the conflict between its — highly natural — assumptions and the — firm — convictions at odds with the conclusion, which appears to be brought out by the argument. Either way, the apparent inconsistency between propositions we believe or find compelling makes us feel uneasy. Various philosophical efforts are undertaken also, or even primarily, to dispel such appearances of inconsistency and the feelings of unease they cause. Thus, philosophers typically respond to sceptical paradoxes with attempts to remove the appearance of inconsistency between their attractive assumptions and the common knowledge-claims their conclusions put into question. This is done by showing in various ways that the sceptical arguments do not manage to derive paradoxical conclusions from attractive assumptions: by showing that, in addition to the explicit assumptions, these arguments tacitly rely on presuppositions or principles that are false or invalid; or that they tendentiously place unduly substantive readings on merely otherwise attractive assumptions; or that they employ words in a way that deprives them of their meaning, etc. By and large, little independent interest attaches to such findings (say, that in the course of the particular argument a substantive falsehood was read into the expression of a truism). These efforts are therefore clearly motivated, at any rate in part, by the desire to ‘take the sting out of paradox’: to dispel the unease caused by the contradictions it apparently reveals in the philosopher’s beliefs — to ‘reduce cognitive dissonance’. The strength of this motivation varies considerably from one philosopher to the next. Thus, theorists of knowledge studying sceptical paradoxes typically (though not invariably) seek, first, to expose in these arguments mistaken assumptions and presuppositions about the nature and structure of knowledge, and, second, to formulate accounts of this concept which do not permit the construction of the familiar sceptical 25 arguments. Some such theorists pursue the project of analysing this concept as an end in itself and treat the resolution of sceptical paradoxes primarily as a test for their analysis; the removal of the worry induced by paradox is, to these theorists, but a minor by-product. Other, ‘paradox-oriented’, philosophers, by contrast, find the analysis of the concept of knowledge a rather stale exercise when undertaken for its own sake and take an interest in it precisely to the extent to which it contributes to the resolution of sceptical paradoxes. Such philosophers are more interested in dispelling the disturbing appearance of inconsistency than in formulating a ‘true account’ or accurate analysis of knowledge. Of course they want both: to remove their disquiet, and to establish philosophical (here: conceptual) claims. But, at any rate when studying paradoxes, the first aim matters more to them than the second. Austin faces a similar problem and displays a similar attitude: As supported by some of the familiar arguments (‘from illusion’, ‘from relativity’, etc.), the doctrine of sense-data constitutes a philosophical paradox: The doctrine appears to contradict common sense and is derived, in an apparently straightforward way, from intuitively plausible assumptions that are apparently trivial. Austin’s response is that of a paradox-oriented philosopher: First and foremost, he wants to ‘dissolve the philosophical worr[y]’ the paradoxical doctrine induces; as a welcome by-product, he also ‘hopes to learn something … about the meanings of some English words … which, besides being philosophically very slippery are in their own right interesting’ (S&S 5). His attitude is certainly more common among philosophers, and his chief therapeutic aim more commonly (if implicitly) accepted as philosophically relevant than it might appear at first sight. What distinguishes Austin from so many other philosophers is not his chief therapeutic aim, but that, at this stage, he recognises it as clearly as the later Wittgenstein (cp. Fischer 2004), who seems to have gone even further, in declaring: ‘As I do philosophy, its entire [!] task consists in expressing myself in such a way that certain disquietudes disappear’ (Wittgenstein 2000: 421), and ‘Thoughts at peace. That is the [!] goal someone who philosophises longs for’ (Wittgenstein 1998: 50). The reason the clear recognition of this goal is essential is that it allows Austin to dissociate the aim of dissolving disquieting philosophical worries from the aim of establishing truth and refuting falsehood, with which it is usually pursued in one breath. And this, in turn, enables him to pursue the therapeutic aim also by other means than the demonstration 26 or refutation of philosophical claims or arguments. We will now see that this is absolutely necessary: There are important cases in which this traditional proceeding gets us nowhere, and the requisite endeavours are ‘therapeutic’ in a stricter sense than the one explained so far. 2. Austin’s Approach The paradox of sense-data is in one striking way different from the sceptical paradoxes we just adverted to: Philosophers studying sceptical arguments sometimes feel compelled to accept them as valid or even sound — but fail to believe their conclusion for a minute. By contrast, precisely some of the most important sense-datum theorists firmly believed their doctrine — even though they did not accept the supporting arguments as sound or even valid. This apparent indifference to the argumentative support of the disturbing doctrine creates a difficulty for the pursuit of Austin’s aims. Taking note of the indifference and the resulting difficulty will let us understand the unfamiliar approach with which he proposes to pursue his aims: the exposure of ‘concealed motives’ (S&S 5). The indifference is evident already in Berkeley. In the first of his Three Dialogues between Hylas and Philonous (henceforward: 3D), and only there, he defends the doctrine that the objects of our immediate perception exist only in the mind. He develops the three then most familiar arguments: the identification argument, the argument from relativity, and the argument ‘from science’. The latter is explicitly based on a ‘scientific’ theory Berkeley is careful not to put into the mouth of his spokesman Philonous, and condemns as downright unintelligible (sic) in the third dialogue (3D 208–10, 242). The second argument is criticised as ineffective in an earlier work (Principles, section 15) and based on an assumption Berkeley explicitly rejects in a later dialogue (3D 245). Even should he not have realised that his remaining, first, argument blatantly presupposes its conclusion, it would still be remarkable that Berkeley should have maintained so fervently so paradoxical a doctrine on so feeble a basis, through which he had seen to such an extent. An even ‘purer’ case (which we shall examine in detail in the next sections) is that of A.J. Ayer, the arguably most important analytic philosopher to champion phenomenalism, whose second book (Ayer 1940) 27 is the main object of Austin’s scrutiny: As we shall presently see, Ayer explicitly bases the doctrine of sense-data on one single argument, the argument from illusion. He then ‘evaluates’ this argument as merely illustrating a new rule of language, rather than warranting any new factual belief at odds with common sense. After carefully establishing this as the main point of his first chapter — he devotes the bulk of his book to the development of a phenomenalist theory that is to show, against appearances to the contrary, that our common ways of speaking and thinking are compatible with the truth of the very factual and paradoxical conclusion that the argument, according to Ayer, utterly fails to support. Clearly, he continues to adhere to this conclusion and to worry about its apparent conflict with common sense. Thus, Austin rather under- than overstates the case when characterising the ‘argument from illusion’ (under which label, following Ayer, he apparently subsumes also the argument from relativity) as ‘an “argument” which those (e.g., Berkeley, Hume, Russell, Ayer) who have been most adept at working it … have all themselves felt to be somehow spurious’ (S&S 4, my italics). Ayer’s combination of critical insight and doctrinal loyalty renders at any rate his belief in the sense-datum doctrine ‘not even faintly sensible’, namely, absurd in a quite strict sense of the frequently overused term: His belief does not merely ‘go against reason or common sense’. (This much need not be wrong: Some scientific theories go with reasons against common sense.) Rather, Ayer’s belief goes against common sense, without reason. In this crucial sense, the term “absurd” appraises the warrant a particular subject has for a belief, and does not entail either that what is believed is false or that nobody has pertinent warrant. Rather, it marks a particularly severe lack of personal or subjective warrant: A subject’s belief is absurd iff (i) the belief conflicts with common sense, thus placing the subject under an obligation to support it by strong and specific evidence or argument; and (ii) the subject is unable to come up with any relevant evidence or argument. Relevant are evidence and arguments that make at least a prima facie case for the truth of the belief, and are accepted by the subject. (His paradoxical belief is absurd if he maintains it either when he clearly ought not to have been taken in, or when he was not taken in by faulty 28 arguments.) When a subject’s belief is absurd because he himself fails to accept what evidence or arguments he can think of, let’s call it a ‘deliberate absurdity’. The absurdity of Ayer’s belief in the sense-datum doctrine makes Austin’s endeavour ‘therapeutic’ in a stricter sense than the one explained in the last section. Its deliberate absurdity creates a difficulty for the pursuit of his therapeutic aims. We noted that an effort is therapeutic if it seeks to remove distressing emotions and to modify dysfunctional behaviour. In a stricter sense of the term, however, we speak of ‘therapy’ only when at least most of the emotions and behaviours addressed are not only distressing or dysfunctional, but also (subjectively) unwarranted. The relevant difference is one of degree: Even such an exemplary psychotherapy as the cognitive therapy of depression will typically address warranted, in addition to unwarranted feelings (as an example from Beck 1995 illustrates: Once her unwarranted feelings of anxiety prevented her from studying effectively, the depressed student’s fear to fail the next exam is quite reasonable — and is addressed by the therapist, even so, as it exacerbates her state). Accordingly, we can classify efforts as more or less strictly therapeutic, depending upon the strength of warrant of the attitudes they seek to modify. In view of the absurdity of Ayer’s belief efforts to put an end to the worry it induces and the theorising it motivates are strictly therapeutic. Austin’s ends are to be achieved by making the addressee shed his belief in the doctrine of sense-data. But when someone’s belief is, like Ayer’s, both paradoxical and a deliberate absurdity, it is impossible to achieve this task in what would seem to be the most straightforward ways: We cannot make a philosopher shed a paradoxical belief he sincerely holds, by ‘direct’ refutation, i.e., by adducing the familiar facts at odds with it. For given that the philosopher is, all along, every bit as familiar with these facts as anyone else, they clearly do not move him. And if his belief is a deliberate absurdity, we cannot proceed, either, by ‘indirect’ refutation, namely, by showing that his stated arguments do not warrant his conclusion. For of this he is already well aware, anyway. Austin therefore has to turn from Ayer’s stated arguments to his ‘concealed motives’ (S&S 5). Since Austin accuses nobody of insincerity, he does not mean reasons the sense-datum theorist is aware of but conceals, but, rather, intends to refer to whatever urges the theorist to maintain his belief independently of what reasons he is aware of: to 29 the motives that have him cleave to his belief even once he has seen through his stated reasons. For those motives may be ‘concealed’ in the sense that the thinker himself is unable to identify them. Austin suspects that the sense-datum doctrine owes its tenacious plausibility to ‘a mass of seductive (mainly verbal) fallacies’ (S&S 5): We find the paradoxical doctrine overwhelmingly plausible even once we have seen through the arguments that purport to establish it, because it is suggested to us, independently of these arguments, by characteristic fallacies that pervade practically all our abstract thinking about perception. We commit these fallacies without realising that we are making a mistake or how the move supports our doctrine. Thus, these fallacies constitute ‘concealed motives’ for endorsing it. To expose and correct the pertinent verbal fallacies, Austin attempts ‘to make clear that our ordinary words are much subtler in their uses, and mark many more distinctions, than philosophers have realized’ (S&S 3). He analyses, in particular, what is entailed, implied, or suggested when these words are employed in different contexts. To do so, Austin repeatedly considers the notorious question, what we should say when. As a result, some of Sense and Sensibilia may look at first sight similar to the sort of ordinary language analysis that became associated with Austin’s name. But, to recapitulate: Here, the analysis of ordinary ways of speaking does not serve to establish — or refute — any philosophical claims. Rather, Austin considers paradoxical doctrines some philosophers maintain on the basis of ‘arguments’ they have already seen through themselves, to varying degrees — and perhaps even so completely that nothing more is required for the demonstration of the absurdity of their belief. Accordingly, Austin focuses on the subsequent task of exposing and removing what concealed motives have them continue to worry about their doctrine, against better knowledge. And we shall see that, once it is put to this use, ordinary language analysis is no longer open to any of the familiar objections that even critics sympathetic to Austin found forceful. Let’s now develop in the requisite detail what has been briefly outlined so far. I will establish first that Austin’s quest for concealed motives is indeed called for (sections 3–4), and shall then explain how he proceeds (sections 5–6). To prove the quest necessary, we need to establish the at first sight outrageous claim that A.J.Ayer, the arguably finest sense-datum theorist within the analytic tradition, maintained the paradoxical doctrine of sense-data as a matter of deliberate absurdity. 30 To do so, we will consider first his once famous ‘linguistic evaluation’ of his preferred argument for the doctrine. Then we will turn to his real view that is so squarely at odds with his evaluative results. 3. Ayer’s ‘Linguistic Evaluation’ Notoriously, the argument from illusion turns on the non-standard use of terms (as in this currently up-to-date formulation, from Robinson 2001: 57–8): (1) In some cases of perception, physical objects appear to possess sensible qualities that they do not actually possess. (When you look at a straight stick partially immersed in water, it appears bent.) Whenever something appears to a subject to possess a sensible quality, there is something of which the subject is aware which does possess that quality. (When something appears bent to you, there is a bent spot of which you are aware.) Hence: In some cases of perception there is something of which the subject is aware which possesses sensible qualities which the physical object the subject is purportedly perceiving does not possess. (There is a brownish spot which is bent, while the stick is not.) (2) (3) Whence it follows, by Leibniz’s Law, that in perception we are sometimes aware of something other than physical objects, namely of ‘sensedata’. But in the ‘delusive’ cases the argument starts out from, our visual experience is qualitatively so similar to the experience we have in those ‘veridical’ cases in which things do appear to us the way they actually are, that we must be aware of the same kind of thing in both. Whence it is finally inferred that in perception we are always aware of immaterial sense-data. As any dictionary reminds us, “aware of” is commonly used to mean ‘having knowledge or understanding of’. Like the claim that someone knows something, the claim that he is aware of something entails that certain facts obtain; and like the attribution of knowledge, it entails no claim about the ‘experience’ the subject ‘enjoys’: When aware of a problem, risk, opportunity, etc., we know that there is a problem or risk, 31 etc., and may be aware of it throughout, even when we do not think of it the whole time. Similarly, when you assert that ‘Smith is aware of the noise Jones’ lawn-mower makes’, you imply that Jones’ lawnmower is in fact noisy, and that Smith knows this; and your assertion need not even suggest that there is something to be heard at the time: ‘Well aware of the noise Jones’ lawn-mower makes, Smith closed the windows already the moment Jones approached his tool shed.’ Even when there is a noise to be heard, someone who does not hear it may be said to be aware of it: ‘The invigilator moved the examination to another hall, because he was aware of the distracting noise’ is true also in case the almost deaf invigilator was alerted to the noise only by a written note. Accordingly, ‘There is something that is bent of which you are aware’ would commonly be understood to express a claim not about any ‘experience’ you enjoy ‘in perception’, but about facts you know and became aware of ‘through perception’ or otherwise: There is in fact something bent lying around to be seen or felt, and you know this (because you see or feel it, or, e.g., someone tells you). On this reading, the notorious ‘Phenomenal Principle’— (2) above — is clearly false: When the straight stick appears bent, there is nothing of which I am aware that is bent. But, equally clearly, its proponents do not want to talk about our knowledge of things actually lying around, competing for space with sticks and coins. They give the term “aware” a new use, if they do not resort to the new phrases “directly aware” or “directly perceive” from the start. Typically, neither innovation is explained in any detail. What gives reason for concern is of course not the novelty as such, but the lack of explanation. Unlike many other sense-datum theorists, Ayer is aware of the novelty, and seeks to remedy the lack, in his Foundations of Empirical Knowledge (Ayer 1940, henceforward: FEK): Properly spelled out, the Phenomenal Principle emerges as the rule of language that introduces the new phrases. Ayer treats “x is directly aware of y” as synonymous with “x directly perceives y”, which he presents as a new, third, sense of perception-verbs that philosophers introduce in addition to two allegedly established senses of those verbs. Ayer infers the existence of these two from the fact that different perception-reports are equally legitimate: ‘For example, a man will say that he sees a distant star which has an extension greater than that of the earth; but if he is asked to describe what it is that he is actually seeing, he may say that it is a silvery speck no bigger than a sixpence’ (FEK 22). In this case, says 32 Ayer, there need be no contradiction, as the man may use the verb “to see” in two different senses: In one sense [A], the sense in which the man can say truly that he sees the star, it is [i] necessary that what is seen should really exist, but [ii] not necessary that it should have the qualities it appears to have. In another sense [B], which is that in which the man can say truly that what he sees is no bigger than a sixpence, it is [ii] not possible that anything should seem to have qualities that it does not really have, but [i] also not necessary that what is seen should really exist (FEK 23). ‘What the advocates of the sense-datum theory have done’— namely, by upholding the Phenomenal Principle which is plainly false when interpreted in accordance with the rules of ordinary language — is ‘to decide’ to introduce a further, third sense of verbs of perception, which combines feature (i) of Sense A with feature (ii) of Sense B: They ‘define [“direct perception” or] “direct awareness” in such a way that if someone is directly aware of [or perceives] an object x, it follows that x exists and that it really has whatever properties it is appearing to have’ (FEK 61). ‘At the same time’, they thus ‘decide to apply’ these words in this sense ‘to delusive as well as to veridical experiences’ (FEK 24). ‘The word “sense-datum” [then] is to stand [quite simply] for any object of which it is conceivable that someone should [perceive it in this new sense, i.e.] be directly aware’ (FEK 59). Accordingly, Ayer accepts the argument from illusion only in this ‘linguistic’ form (cp. FEK 24–5): (1) When we [in sense A] see a stick partially immersed in water, the stick appears to us to have a property it does not really possess (bent instead of straight). (2) Let there be a sense of “to perceive” in which, whenever a subject perceives anything in any sense [A or B], there actually is something that is being perceived, that really has the properties it appears to have. (3) By (1) and (2) and Leibniz’s Law: When we [in sense A] see the stickin-water, there actually is something other than the stick, the physical object, that is being perceived (in the new sense of that verb). (4) Apply the term “sense-datum” to whatever objects are being perceived in the new sense of the verb. (5) By (3) and (4): When we [in sense A] see the stick in water, we perceive (in the new sense) a sense-datum but no physical object. 33 Cast into this form, the argument presents no paradox. It no longer even appears to bring out a contradiction in our common beliefs: Between the conclusion that, in one (new) sense, we ‘perceive’ no physical object but a sense-datum, and the assumption that we then ‘perceive’ a stick, in another (established) sense, there is no more of a contradiction than between the two statements of the star-watcher. Instead of establishing a paradoxical conclusion, the ‘argument’ merely illustrates the use of two new rules of language that introduce a new terminology: The philosopher who says that he is seeing a sense-datum in a case where most people would say that they were seeing a material thing is not contradicting the received opinion on any question of fact … What he is doing is simply to recommend a new verbal usage. He is proposing to us that instead of speaking, for example, of seeing a straight stick which looks crooked … we should speak of seeing a sense-datum which really has the quality of being crooked, and which belongs to a straight stick (FEK 25). The generalisation from delusive to all cases of perception turns on the ‘contention that if [delusive and veridical] perceptions are not qualitatively distinguishable the objects perceived must be of the same kind’ (FEK 24–5). Also this ‘contention’ Ayer accepts only ‘as a rule of language’ (FEK 25): One and the same label should be applied to those objects. As the label “sense-datum” is to be applied to whatever is being ‘perceived’ in the new sense — (4) above — , the further rule actually is to use perception-verbs in this new sense, regardless of whether the perception is delusive or veridical. We may note (even if Ayer does not) that, in the latter case, what we perceive in the new sense are of course physical objects, which then ‘really exist and have the properties they appear to have’. Thus, the new rules have us apply the label “sensedatum” also to physical objects: In some situations, the ‘sense-data’ we are said to ‘sense’ are material things. Therefore, Ayer is perfectly right to conclude his evaluation by stating that the final conclusion of the argument, ‘What we see, or otherwise directly experience, are never material things, but only sense-data’, does not express ‘a proposition about an empirical matter of fact’, viz., about which of two different kinds of thing we perceive, but rather a ‘resolution about the usage of words’ (FEK 28), viz., about which of two different sets of terms to use: To say what you perceive, always use the terminology of ‘direct perception’ and ‘sense-data’. 34 Thus properly understood, the so-called ‘argument from illusion’ is no argument at all: It is not even meant to establish the truth of any claim. Rather, it is a ‘procedure’ for introducing a new ‘terminology’ (FEK 25): ‘[T]he expressions “direct awareness” and “sense-datum” are to be regarded as correlative; and since each of them is being used in a special, technical sense, it is not satisfactory merely to define one in terms of the other. It is necessary first to employ some other method, such as the method of giving examples’ (FEK 61). Such examples are provided by the ‘delusive’ cases the ‘argument from illusion’ starts out from, where one is tempted to point in the direction of the (straight) stick or the (round) coin, and say, ‘Do you see the bent speck / the elliptical spot? That is what I mean by a “sense-datum”.’ The first half of the ‘argument’ seeks to explain this in a less objectionable form. The second half then motivates the recommendation that the new terms be applied in all cases of perception. Ayer stresses that in endorsing the sense-datum ‘theory’ he endorses but this recommendation: ‘[T]he acceptance of this theory involves nothing more than a decision to use a technical language’ (FEK 57). Indeed, ‘this important point’ holds quite generally for ‘what philosophers call theories of perception’ (FEK 48). None of them puts into question anything we believe about matters of fact. Ayer’s evaluation thus illustrates Austin’s more general introductory remark: ‘[S]trange though the doctrine looks, we are sometimes told to take it easy — really it’s just what we’ve all believed all along. (There’s the bit where you say it and the bit where you take it back.)’ (S&S 2). We now turn to the bit where Ayer reinstates what he here takes back. 4. Ayer’s Real View Ayer’s evaluation of the argument from illusion and the sense-datum ‘theory’ it seeks to establish appears, at first sight, to lend substance to the claims with which he introduces his own approach to ‘the problem of specifying the relationship of material things to sense-data’ (FEK 229): There is, indeed, a sense in which it is correct to say that both sense-data and material things exist, inasmuch as sentences that are used to describe sense-data and sentences that are used to describe material things both very frequently express true propositions. But it would not be correct to 35 infer from this that there really were both material things and sense-data, in the sense in which it can truly be said that there really are chairs as well as tables, or that there are tastes as well as sounds. For whereas, in these cases, the existential propositions refer to different empirical “facts”, this does not hold good in the case of sense-data and material things (ibid.). Certainly — according to his evaluation, talk of sense-data was introduced to re-describe familiar facts about material things (e.g., that the straight stick looks crooked when immersed in water), indeed, introduced in such a way that material things sometimes are (being referred to as) ‘sense-data’. The empirical facts are exactly what we took them to be all along; the sense-datum theorists merely introduced a new way of talking about them that is useful ‘in philosophising about perception’ (FEK 26). But, it quickly emerges, this is not at all what Ayer here means to claim: The reason why it is ‘not correct’ to say that ‘there really are both material things and sense-data’ is that, upon final analysis, ‘material things are nothing but collections of … sense-data’ (FEK 231)! Of course, Ayer assures the reader, material things are not literally constituted of immaterial sense-data ‘as a patchwork quilt consists of different coloured pieces of silk’ (FEK 232); rather, they are ‘logical constructions out of sense-data’ (FEK 237). Initially, ‘what is being claimed is simply that the propositions which are ordinarily expressed by sentences which refer to material things could also be expressed by sentences which referred exclusively to sense-data’ (FEK 232). But already in the two very next sentences Ayer sets out to show something far more ambitious: As for the belief in the “unity” and “substantiality” of material things, I shall show that it may be correctly represented as involving no more than the attribution to visual and tactual sense-data of certain relations which do, in fact, obtain in our experience. And I shall show that it is only the contingent fact that there are these relations between sense-data that makes it profitable to describe the course of our experience in terms of the existence and behaviour of material things (ibid.). I.e.: All we actually ‘describe’ when talking of material things is ‘the course of our experience’. And it is only due to the fortunate coincidence that the sense-data that make up this experience stand in certain relations, that it is even ‘possible for us successfully to employ the physical terminology that we do’ (FEK 243), i.e., to speak as if there 36 were substantial things like sticks and stars — even though, in fact, our experiences are all there is. Austin, who bases this claim on a much earlier part of the text (FEK 11–19), is therefore perfectly right to maintain that ‘the apparent sophistication of Ayer’s ‘linguistic’ doctrine really rests squarely on the old Berkeleian … ontology of the ‘sensible manifold’’ (S&S 61). Ayer believes that all there is and all we can perceive (and talk about by means of our ‘physical terminology’) are sense-data. I.e.: Ayer upholds not only the very factual, paradoxical, claim about perception that, according to his evaluation, fails to be established by the ‘arguments’ philosophers typically adduce for their ‘theories of perception’. He also maintains an even stronger, ontological, claim that entails that doctrine about perception. Far from lending the latter any support, however, the — rather quickand-dirty — reasoning that leads Ayer to the ontological claim tacitly presupposes the sense-datum doctrine of perception: Ayer believes that propositions that refer to material things must be expressible in terms of sense-data, because those propositions are ‘empirically significant’ (FEK 231). Indeed, he seems to assume that the ‘empirical’ or ‘factual significance’ of propositions about material things can be spelled out exhaustively by means of propositions about sense-data. Whence the former somehow boil down to the latter, and (facts about) material things really are (facts about) sense-data. The crucial assumption follows from a definition and an epistemological doctrine: We say that a sentence is factually significant to any given person, if, and only if, he knows how to verify the proposition it purports to express — that is, if he knows what observations would lead him, under certain conditions, to accept the proposition as being true, or reject it as being false (Ayer 1946: 35). And the only observations that lead us to accept or reject propositions about material things are observations of sense-data; our evidence for those propositions ‘will always consist in the occurrence of … sensedata’ (FEK 231, cp. FEK 17). The most influential motivation for accepting only ‘facts about our experience’ (FEK 17) as evidence is the quest for certainty. But, far from embracing it, Ayer explicitly rejects this quest, as based on a conceptual confusion: ‘The notion of certainty does not apply to propositions in which we express our perceptual judgements’ (FEK 45). So why does 37 he maintain the epistemological doctrine, even so? The above passage continues: ‘Indeed, there is nothing else [but the occurrence of sensedata] in which one can legitimately suppose [our empirical evidence] to consist, once one has accepted the rule that the word “sense-datum” is to be used to stand for whatever is, in fact, observed’ (FEK 231). But of course Ayer had explicitly introduced the word “sense-datum” to stand for whatever is directly perceived. Thus, he manifestly makes a leap of thought that presupposes that only what is directly perceived is in fact observed. Failing to support his epistemological doctrine by any independent argument, Ayer here tacitly and (presumably) unwittingly relies on a ‘factual’ version of the sense-datum ‘theory’ of perception, of the sort he has explicitly rejected. Arguably, the same happens, e.g., when Ayer asserts without argument, as if it was entirely uncontroversial, that ‘facts about our experiences’ constitute ‘the evidence that is ordinarily thought [!] sufficient to establish’ assumptions about material things (FEK 17): Without noticing it, I submit, he conflates the standard explanation ‘Empirical evidence is the evidence accessible to our senses’ and his doctrine ‘Empirical evidence is the evidence constituted by facts about our experiences’. In unwittingly moving from the truism to the contentious doctrine, he presupposes that only facts about sense-data are accessible to our senses. To sum up: In the end, Ayer sticks to everything he took back, and makes even stronger claims. He continues to rely on the very paradoxical doctrine he knows full well not to be supported by any argument, and thus arrives at claims that constitute the epistemological and metaphysical core doctrines of phenomenalism. To make Ayer give up the sense-datum doctrine, one clearly has to address more than the stated arguments he adduced for it — and dismissed. One also has to look for concealed motives that may make the doctrine seem compelling independently of all those arguments. 5. Illusion and Delusion In chapters III–V of Sense and Sensibilia, Austin analyses Ayer’s presentation of the ‘argument from illusion’, to identify a series of ‘seductive (mainly verbal) fallacies’ that may be effective as such motives, namely fallacies that may lead one to accept part or all of the sense- 38 datum doctrine independently of the ‘argument’ purporting to establish it. Accordingly, the fallacies at issue are not committed in the course of the ‘argument’ itself but rather surface, e.g., in its title and the description of the cases it starts out from. The first of these fallacies is the conflation of “illusion” (‘argument from illusion’) and “delusion” (‘cases of delusive perception’). In three steps, Austin explains why this is a fallacy, why we may commit it, and how it may suggest to us that we perceive something immaterial when looking, say, at the stick-in-water (S&S 22–5). First, Austin shows that there are important differences between illusions and delusions, that the concepts of illusion and delusion are not at all the same. To do so, he points out some genuine examples of illusion and delusion, respectively: optical illusions like the Müller-Lyer illusion or the familiar phenomenon of rapidly rotating wheels that look as if they were turning slowly in the other direction; as opposed to delusions of persecution or of grandeur, which ‘are primarily a matter of grossly disordered beliefs … and may well have nothing to do with perception’ (S&S 23). These examples he then uses to bring out what are for present purposes ‘the most important differences’ between the two, first and foremost: ‘The term ‘an illusion’ (in a perceptual context) does not suggest that something totally unreal is conjured up — on the contrary, there just is the arrangement of lines on the page … [or] the rotating wheels; whereas the term ‘delusion’ does suggest something totally unreal, not really there at all’ (S&S 23). Second, Austin provides those who found the ‘argument from illusion’ persuasive with reasons to believe that they actually confused the two concepts: He brings out that, and why, this confusion is seductive and points out that it surfaces in the argument from illusion. As he explains, it is easy to confuse ‘illusions’ and ‘delusions’ not only because ‘these terms are often used loosely’, but also because ‘people may have, without making this explicit, different views or theories about the facts of some cases’ (S&S 24), so that one and the same case is described in both terms — e.g., Ayer’s example of the ‘delusive perception’ of a mirage: Some (like Ayer) tend to think mirages visions conjured up by the crazed brains of thirsty and exhausted travellers (delusion); many others explain mirages as cases of atmospheric refraction (illusion). The confusion thus rendered seductive actually crops up in the ‘argument from illusion’: In its title and its talk of ‘delusive perceptions’, ‘there are two clear implications — (a) that all the cases cited in the argument 39 are cases of illusions; and (b) that illusion and delusion are the same thing’ (S&S 22). Third, Austin explains how this confusion may have been effective as a motive for endorsing the idea at issue. This might make those who previously acknowledged that they shared the confusion further acknowledge it as one of their motives for endorsing that idea. Austin explains: ‘So long as it is being suggested that the cases paraded for our attention are cases of illusion, there is the implication (from the ordinary use of the word) that there really is something there that we perceive’ (S&S 25). Whereas when we call these cases ‘delusive’, ‘there comes in the very different suggestion of something being conjured up, something unreal or at any rate ‘immaterial’. These two implications taken together may then subtly insinuate that in the cases cited there really is something that we are perceiving, but that this is an immaterial something’ (S&S 25), that, in Ayer’s words, ‘even in the case where what we see is not the real quality of a material thing … we are still seeing something … [but something that] is not part of any material thing’ (FEK 4). Having thus exposed a ‘seductive (mainly verbal) fallacy’ as a concealed motive for endorsing this idea, Austin needs to remove the motive by showing up the fallacy for what it is. Already the opening step of the exposure showed that, pace (b), illusions are not to be conflated with delusions. In the sequel (S&S 26–7), Austin examines briefly ‘some of the other cases Ayer lists’ (refraction, perspective, reflections), to show that, pace (a), none of them can be described as a case of illusion. This argument involves two assumptions: (i) Someone qualifies as being ‘under an illusion’ only if he is ‘taken in’; and (ii) something deserves to be termed an ‘illusion’ only if most people are under an illusion when confronted with it. Both assumptions are in line not only with common usage but also with the explicit definitions of “illusion” given by the philosophers Austin targets, such as this definition from Price (1932: 27): ‘An illusory sense-datum of sight or touch is a sensedatum which is such that we tend to take it to be part of the surface of a material object, but if we take it so we are wrong’. Indeed, but for the reference to ‘parts of surfaces’, this definition would capture no other idea than the one the present two assumptions articulate. Perhaps for this reason, Austin goes on to show that this reference should be dropped, as it prevents the definition from fitting paradigmatic cases of illusion (such as the Müller-Lyer illusion or the Headless Woman), which it is 40 surely meant to accommodate (S&S 27–8). The relevant upshot is that it is presumably from assumptions (i) and (ii) that the philosophers addressed meant to proceed to (a). Austin’s argument against the latter thus engages with his addressees: He exposed and refuted fallacies that constituted (one of) their concealed motives for adopting the specific doctrine that we perceive only immaterial sense-data in cases of delusive perception. 6. Look, Appear, and Seem In the following chapter IV, Austin exposes further concealed motives, mainly for directly adopting the general doctrine that all we ever perceive directly are immaterial sense-data. Here, he explicitly executes only the first two of the three steps involved in exposure. But we can easily work out on our own how the fallacies identified can be effective as a motive. (I) Again Austin starts out by drawing our attention to a possible confusion, viz., to that of looks, appears, and seems. To make us ‘avoid… misguided assimilations’ (S&S 34), he first shows that “looks” does not occur in all syntactic constructions in which “seems” figures, which in turn does not share all of the constructions of “appears” (S&S 34–6). Then he brings out what he calls the different ‘root ideas behind the uses’ of these three expressions: “Looks” is used quite simply to comment on the looks of things. “Appears” is used with reference to certain special circumstances, whereas “seems” makes an implicit reference to certain (less than conclusive) evidence. To bring this out, Austin considers ‘in just what circumstances we would say which, and why’ (S&S 36), given three sets of examples (S&S 36–7), including: (1) He looks guilty — he has the look of a guilty man. (2) He appears guilty — ‘I quite agree that, when he’s prevaricating over all those searching questions about what he did with the money, he appears guilty, but most of the time his demeanour (not just ‘his looks’) is innocence itself.’ (3) He seems guilty — ‘On the evidence we have heard so far, he certainly seems guilty.’ [The evidence of course suggests that he is guilty, not that he seems guilty. And it is less than conclusive; otherwise, I would have said less cautiously, ‘He is guilty.’] 41 Hence, ‘the root ideas behind the uses of ‘looks’, ‘appears’, and ‘seems’ are not the same’ (S&S 37). (II) It is, however, easy to run them together, as the use of these words sometimes does come down to pretty much the same thing: The use of “looks” (S&S 38) — and “looks like” (S&S 39–40) — usually does not involve implicit reference to particular evidence, but sometimes does. Whereas that of “seems” (S&S 38–9) — and “seems like” (S&S 39) — usually involves such reference but sometimes does not. Therefore, ‘it is not enough to simply examine the words themselves; just what is meant and what can be inferred (if anything) can be decided only by examining the full circumstances in which the words are used’ (S&S 41) on the particular occasion under consideration. In abstract philosophical reflection, we notoriously neglect this kind of exercise and thus tend to regard “look”, “appear”, and “seem” as always carrying pretty much the same implications. The ‘pretty free use’ Ayer makes of these expressions in setting out the cases on which his argument is based, ‘apparently in the manner of most other philosophers’ (S&S 33), suggests he is a case in point, as he writes, e.g.: ‘A stick that normally appears straight looks bent when it is seen in water’ (FEK 3) — rather than ‘A stick which normally looks straight appears bent (under special circumstances, namely) when it is seen in water.’ I.e.: The conflation of the three terms under consideration is seductive and there is textual evidence that Ayer was prone to it. (III) The resulting imprecision in the formulation is, by itself, completely unimportant. But the moment we assume that these words always carry the same implications, we may draw from a loose use of one word the implications typically warranted by this word’s strict use but not licensed in the present case. And this may subtly suggest to us that we always perceive but sense-data. Here is one way how (another will be spelled out in section 8). Suppose we loosely say ‘The coin seems silvery’, instead of ‘The coin looks silvery’, when viewing it in broad daylight, so that the looks of the coin are wholly conclusive. (What more does the coin need to do to be silvery than to look silvery under these conditions?) When we say the former, there is the implication (from the root idea behind the use of “seems”) that we are talking about less than conclusive evidence for the coin’s being silvery. But what evidence? As the situation is not one in which we ordinarily even consider the question of evidence, we are stumped for an answer and may jump to the next best thing: When we take ‘The coin seems silvery’ 42 to be tantamount to ‘The coin looks silvery’, there is the implication (from the root idea behind the use of “looks”) that we are commenting on the looks of the coin. Whence we conclude that these looks must be our less than conclusive evidence. This idea may be reinforced by another confusion which Austin discusses in passing (S&S 38–9): Utterances of such sentences as ‘He certainly seems to be guilty but he is not’ usually involve a shift in the evidence implicitly referred to: On the evidence presented so far (or publicly available) he seems to be guilty, but there is (or I have) further evidence that establishes that he is not. If one overlooks this shift, one may think that all the evidence one may possibly gather is compatible with ‘He is not guilty’. Applying this to the ‘evidence of our senses’ we supposedly have when looking at the coin in broad daylight, we may again come to think that the looks of the material thing are, even then, less than conclusive evidence. From these conclusions it is natural and tempting to jump to the general doctrine: We are always left to work out the ‘real qualities’ of material things. All we ever strictly speaking see are the looks or visual appearances of things that are but inconclusive evidence. These ‘visual appearances’ Ayer then equates with our ‘experience’ (FEK 17–18), and thus moves from shared to subjective ‘sense-data’, a move he regards as one between different equally ‘legitimate’ ways of individuating these entities (FEK 153–5). Hence, in the new terminology: All we ever perceive are sense-data. The argument from illusion infers this from the specific conclusion about delusive cases, which is suggested by the confusion of illusion and delusion, and from the assumption that delusive and veridical perceptions are qualitatively alike, which Austin considers blatantly wrong and attributes to another seductive fallacy, namely, the confusion of two different uses of “looks like” that have quite different implications (S&S 40–43). Without committing any of these fallacies in his explicit reasoning, Ayer was manifestly prone to them and may have jumped to the pertinent ideas in the ways indicated. If so, his final idea that all we ever perceive are sense-data would indeed be the result not of the ‘simple ‘argument’’ (S&S 4) he saw through, but of ‘a mass of seductive fallacies’ that continued to constitute ‘a wide variety of concealed motives’ that mutually reinforce each other (S&S 5). 43 7. The Relevance of Concealed Motives Austin’s considerations, however, did not much move their addressees. They mistook his efforts to expose concealed motives for attempts to refute the stated arguments (see, e.g., Ayer 1969). And of course the exposure, e.g., of confusion implicit merely in the title of the ‘argument from illusion’ does not even engage with this argument. As long as its relevance and function remain unclear, the exposure of concealed motives will not move anyone to give up an intuitively compelling doctrine. Exposure therefore needs to be accompanied by a forceful demonstration of its relevance — a second, in a way prior, job to which Austin did not, perhaps, pay sufficient attention (although he did not only make the comments quoted in sections 2–4 above, but provided further reasons for attending to concealed motives, in S&S 28–32). Advancing beyond Austin, I shall therefore now bring out more fully the relevance, first, of concealed motives and, in the next section, of the sort of fallacies that Austin suggested might be such motives. The sort of linguistic analysis Austin practiced can help us do the first job in a more systematic way than above (in section 2): Such analysis can help to reveal that intuitively compelling reasoning crucially involves leaps of thought that already presuppose the very claim the reasoning purports to establish. This finding implies that the subject has less warrant for this claim than he may have thought — and no warrant at all, in case the claim is (i) paradoxical and (ii) being presupposed by the subject in all arguments he actually adduces for it. This conclusion, in turn, would imply that the subject countenanced his idea already prior to all stated arguments, for concealed motives rather than for any of his stated reasons. The latter provide at best a semblance of justification for something he already believed, anyway. In this way, the identification of vicious leaps of thought can forcefully reveal that a thinker was driven by concealed motives to adopt an idea for which he has no warrant. I now want to show how leaps of thought that presuppose the sense-datum doctrine are crucially involved, at different stages, in the reasoning that leads Ayer to accept the ‘argument from illusion’ as a ‘legitimate procedure’ for introducing the sense-datum ‘terminology’ (FEK 25). (The traditional version of the argument is considered in the next section; the other arguments we mentioned are analysed in Fischer 2006.) As we have seen (in section 3), Ayer’s reasoning proceeds from the distinction between two allegedly established uses of perception-verbs, 44 which make for different descriptions of such cases of ‘qualitatively delusive perception’ as the ‘argument from illusion’ starts out from: Sense A, in which it is ‘[i] necessary that what is seen should really exist, but [ii] not necessary that it should have the qualities it appears to have’; and sense B, in which ‘it is [ii] not possible that anything should seem to have qualities that it does not really have, but [i] also not necessary that what is seen should really exist’ (FEK 23). This latter sense Ayer took the star-watcher to employ in saying ‘I see a small silvery speck’. But, first, Ayer explicitly assumes that there actually is a huge star to be seen and that the star-watcher sees it (in sense A). And, pace B(i), we would certainly not say that the man sees anything at all (in any sense), if nothing (no star, planet, satellite, or plane etc.) was to be seen in the region of the sky he was looking at (cp. S&S 94–5). Second, pace B(ii), we do not use the phrase “a small silvery speck” to describe anything as actually being small and silvery, in situations like the one envisaged (in which we do not speak of specks of grime or spots on a dress). In these situations, we rather use the idiom of ‘spots’ and ‘specks’ to pick out or characterise the things we see in terms of how they look or appear to us, and of how well we can see them. We resort to this idiom when we (or an interlocutor to whom we want to point it out) cannot discern of what kind the thing to be seen is; we say how it looks when we cannot tell what it is: ‘Do you see that faint, small reddish spot over yonder? Might that be a tower?’ As we approach, the spot then turns out to be a brick tower (or something else). We refer to the thing to be seen over yonder (the tower, if that is what it is) as a ‘spot’, as long as we, or our interlocutor, cannot tell whether it is (say) a tower. We then talk of a ‘small spot’ when this thing looks small from where we stand, and of a ‘faint spot’ when we see it only faintly. In a nutshell: A small reddish spot is a public thing that looks or appears small and reddish (from here, now), and not a private thing that is small and reddish. Accordingly, we say ‘I see a small silvery speck’ when we cannot tell whether what we see is a star, a planet, or a satellite. But we do then claim that there really is some thing, made out of gas or silicone or whatnot, while we do not claim that this thing actually is small. Adhering, by assumption, to the established usage, the star-watcher uses “to see” in Ayer’s sense A, both when saying that he sees a huge star and when saying that he sees a small speck: He is using the expressions “a huge star” and “a small speck” to refer to one and the same, ‘really existing’, 45 object of perception, viz., a celestial body. In one case, he says of what kind that object is, classifies it in terms of ‘qualities it really has’. In the other, he says how it looks to him. And in both cases it is, in Ayer’s slightly misleading words, ‘not necessary’ that the object ‘should have the qualities it appears to have’. What matters for present purposes is not that Ayer’s analysis of our established usage is wrong, but that it is based on a leap of thought that presupposes (part of) the sense-datum doctrine. Ayer takes the star-watcher to be using “to see” in two different senses, because he assumes that in saying ‘I see a silvery speck no bigger than a sixpence’ the man ‘say[s] truly that what he sees is no bigger than a sixpence’ (FEK 23). I.e.: Ayer thinks that the expression “small silvery speck” is being used to describe — correctly — something that is small (and silvery). Without any argument, he leaps to that idea from this description of the example: ‘a man will say that he sees a distant star … but if he is asked to describe what it is that he is actually seeing, he may say that it is a silvery speck no bigger than a sixpence’ (FEK 22). The latter assumption entails the conclusion not by itself, but in conjunction with an instance of (one half of) the sense-datum doctrine: (A) The man says ‘I see a silvery speck no bigger than a sixpence’ to describe what it is that he is actually seeing. (SD) The man is actually seeing some thing (a speck or sense-datum) that really is no bigger than a sixpence. Hence: (C) The man says ‘I see a silvery speck no bigger than a sixpence’ to describe some thing that really is no bigger than a sixpence. In declaring that the speaker here employs the perception-verb in sense B, Ayer leaps to the even stronger conclusion that the man means to describe some thing as really being no bigger than a sixpence. It is, Ayer seems to think, perfectly obvious to the man that what he is seeing is no bigger than a sixpence. But at the same time, Ayer seems to suppose, the — plain — man would say that the object of perception that fits the given description, the speck, merely ‘appears to exist’ or, at any rate, that it does ‘not really exist’. For sense B is framed to accommodate precisely this verdict. To sum up: At the heart of the first step of Ayer’s reasoning lies a leap of thought that already presupposes an instance of the epistemological half of the sense-datum doctrine, and leads Ayer to place a patently wrong interpretation on our established ways of speaking. 46 The next and crucial step, the introduction of the ‘new terminology’, presupposes the other, ontological, half of that doctrine. Recall that more is involved in the introduction of the new sense of perception-verbs than simply the combination of features Ai and Bii. For the adoption merely of this combination would have the plain man say that we ‘directly perceive’ something only in cases of veridical perception: Only in these, we would ordinarily say, does there ‘really exist’ some thing that ‘really has the qualities it seems to have’. It is therefore crucial for the argument that the sense-datum theorists also, second, ‘decide … to apply the word “see”’ in this new sense ‘to delusive as well as to veridical experiences’ (FEK 24). But of course they do not believe that their ‘decision’ creates an object that really is bent, in addition to the stick-in-water that merely looks bent. Clearly, therefore, when they ‘decide to apply’ the word in the new sense in the cases from which, indeed, their reasoning starts out, they already assume that this object, which actually is crooked, ‘really exists’. This is of course the very assumption the plain man, a moment ago, was taken not to share. If, like Ayer, we explicitly base our talk of sense-data on nothing but this ‘decision’, we simply presuppose, tacitly, an admittedly controversial ontological assumption, without argument. Thus, Ayer relates with approval: Having adopted it [the new sense], they [the sense-datum theorists] cannot say, in the case of delusive perception, that what is experienced is a material thing; for [a] either the requisite material thing does not exist, or else it has not got the requisite property. And so they say that [b] it [the thing they can say is experienced or perceived, in the new sense] is a sense-datum — and not one word is added to explain why they should then be able to say that anything is ‘experienced’, in the new sense (FEK 24).In thus moving, without further argument, from (a) on to (b), Ayer simply presupposes that (at any rate in cases of delusive perception) there exists an other than material thing that ‘has got the requisite property’. In this way, we come to see that the ‘argument from illusion’ suffers from an even more fundamental defect than the one Ayer made out: It does not only, upon proper evaluation, boil down to a ‘procedure’ for introducing a new terminology, which cannot, and does not even purport to, establish any unfamiliar facts (as explained above, in section 3). Rather, its execution involves leaps of thought that tacitly presup- 47 pose the truth of the factual claim that the ‘argument’, in its traditional, ‘pre-linguistic’ version, purported to establish. In a similar way, we have seen (in section 4), Ayer continues to presuppose the sense-datum doctrine of perception when leaping to the epistemological and metaphysical main tenets of phenomenalism. The persistence of these leaps suggests that adherence to that doctrine may be part-and-parcel of a comprehensive ‘urge to misunderstand’ (to borrow a term from Wittgenstein): A sense-datum theorist may make, unwittingly but systematically, leaps of thought that presuppose his core doctrine. Ayer’s leap from (A) to (C) above illustrates why the ‘urge’ to make those leaps is one ‘to misunderstand’: They may lead the thinker to place biased interpretations on a whole range of expressions, both ordinary and technical (“silvery speck”, “sees”, “perceives”, “aware of”, “empirical evidence” etc.). In this way and more directly, these unwitting leaps vitiate the ‘arguments’ he adduces for his doctrine, by having them presuppose what they are meant to establish. And, finally, the leaps may have the thinker rely on, and cleave to, the doctrine, even though he has (partially) seen through those ‘arguments’ and rejected the doctrine as unwarranted, namely when they occur in his further reasoning and thus take him to equivalent or even stronger claims. Where a paradoxical doctrine is being presupposed in this way throughout, in all pertinent arguments, we face the following situation: It is not merely (as suggested in section 2) at the end, once the stated ‘reasons’ have been seen through, that the doctrine is being maintained for concealed motives different from those ‘reasons’. Rather, it is being maintained for these hidden motives from the start. The doctrine grips our philosophical imagination, we do not quite know why, and then drives and biases all our philosophical reasoning on the topic at hand, without our quite realising how. Thus, it drives us to formulate ‘arguments’ for it that presuppose what is at issue and to construct ‘theories’ to reconcile it with our common convictions, regardless of whether or not we have, and believe to possess, warrant for it. In a nutshell: Everything important has already happened before the first argument for the paradoxical doctrine is being formulated; and once it has happened, we are set to build houses of cards (like Berkeley’s or Ayer’s system of phenomenalism), as long as we merely belabour the obvious ‘symptoms’ (the stated ‘arguments’) rather than turn to the underlying ‘causes’ (the concealed motives). 48 8. The Relevance of Verbal Fallacies Austin suggests that various seductive verbal fallacies may be effective as such motives. Some of these fallacies are particularly pertinent. First, and unlike, e.g., Ayer’s misinterpretation of our common ‘speck-talk’, they are not themselves manifestations of the given urge to misunderstand, i.e., of the propensity to make leaps of thought that presuppose the very doctrine at issue. Rather, the present fallacies may account for this propensity. And, second, they are not random mistakes either, but particularly likely to occur in philosophical reflection. In general terms, my idea is this: In abstract reflection, we often unwittingly start to use familiar words in new ways. Frequently, such innovation will be perfectly legitimate. But if made unwittingly, it is potentially misleading: We cannot change the inferential habits we acquired in decades of ordinary language use without making some effort. In particular when we do not even notice that we are using familiar words in a new way, we make no pertinent effort. In this case, we will therefore mindlessly continue to draw inferences licensed only by the established use of our terms, also from statements that employ these words in that new way. Thus, we may leap to entirely unwarranted conclusions, which may include the paradoxical doctrine. The core mechanism of unwitting linguistic innovation followed by mindless transport of ordinary implications to the new contexts is illustrated by the second conflation Austin discusses, of the root ideas behind the use of “looks” and “appears” or “seems”. In abstract reflection on perception in general, we frequently want to think or talk about how things look-or-sound-or-feel-or-smell-or-taste, rather than about how they do in any one specific sensory modality. Philosophers then commonly use “appears” and “seems” to stand for that disjunction. Since ordinary language is geared towards more mundane and specific pursuits, it is perhaps no surprise that this philosophical use is different from the common use of these terms, discussed above (in section 6). There, we considered the conflation of the root ideas behind “seems” and “looks” that comes from giving the former the new use indicated, without realising its novelty. And we saw how this conflation may take us to the sense-datum doctrine of perception. Consider now “appears”. Ayer, for instance, gives it the new use when writing: ‘A stick that normally appears straight looks bent when it is seen in water’ (FEK 3). The adverb “normally” cancels the com- 49 mon reference of “appears” to special circumstances; the latter is used as a synonym of “looks-or-etc.”. When we do not notice that we are deviating from the ordinary use, we may, however, draw an inference that is licensed — only — by the common use: As it invokes reference to special circumstances in which things may look different than they usually do, and are, it implies that what appears F need not be F, and, indeed, usually suggests that the thing probably is not F. Thus, we may unwittingly infer from formulations like Ayer’s that, even under the most normal circumstances, the stick may look straight but fail to be straight. We may, that is, leap to the conclusion that all circumstances under which we can view a thing are equal, and equally special, so that we may under no circumstances simply trust its looks: We can never tell just by looking what shape something has, but always have to work this out on the basis of how it looks (and feels) under various different circumstances (cp. FEK 266–9); all our eyes can tell us is how things appear (cp. 3D 174–5). I.e.: Visual appearances are all we strictly speaking see. Through a whole series of such fallacies which mutually reinforce each other, we may come to find the sense-datum doctrine utterly compelling. Compelling ideas, just like core beliefs, are want to pervade our thinking on the pertinent topic and bias our reasoning in various ways, which include belief and confirmation bias (see, e.g., the survey of Evans & Feeney 2004, with numerous references). In particular, we may tacitly presuppose or even explicitly assume the idea in the very arguments we construct to support it. The ‘arguments’ meant to establish the sense-datum doctrine are a case in point: Above (in section 7), we have seen how Ayer’s ‘linguistic’ version of the argument from illusion tacitly presupposes the doctrine for cases of delusive perception. In its traditional version, by contrast, the argument explicitly makes the (factual) assumption which the ‘linguistic’ reformulation allows to sweep under the carpet, namely, in the shape of the Phenomenal Principle: Whenever something appears to a subject to possess a sensible quality, there is something of which the subject is aware which does possess that quality. Even though this idea is at odds with common sense, we thus assume from the start and without argument that there is something that ‘has got the requisite property’, whether or not the material thing we contemplate 50 does. Indeed, the assumption tends to strike us as perfectly natural and uncontroversial, when we go along with the argument. This remarkable attitude may be due to a process that is, in a way, the converse of the one that had us find the sense-datum doctrine compelling, in the first place. At this second stage, after the doctrine’s adoption, further unwitting linguistic innovation may result from leaps of thought that presuppose the doctrine. A case in point is the leap from (A) to (C) which leads Ayer to place an other than ordinary interpretation on phrases like “small speck” or “elliptical spot”, namely, to regard them as actual attributions of that size or colour to something. We place this new interpretation on these phrases when we regard specific claims such as When a coin appears elliptical to us, (there is and) we are aware of an elliptical spot or speck. as illustrations of the Phenomenal Principle that speaks of ‘something … which does possess that quality’. But when we consider only the sample sentence, we may lapse back into our ordinary way of speaking of ‘spots’, ‘specks’, and what we are ‘aware of’, and interpret the sample sentence in line with the rules of ordinary language: ‘When a coin appears (i.e., looks under special circumstances) elliptical to us, we are aware of (and there is) some (perfectly tangible) thing that looks elliptical (from here, now).’ This is, of course, a truism. When we fail to realise that we are placing a different interpretation on the specific claim when treating it as an illustration of the Phenomenal Principle, we may therefore come to regard the latter as a generalisation from truisms. Obviously, this is a mistake: The principle can only be inferred from the — false — specific statements we obtain by placing the new interpretation on the sample sentence and its likes. Thus, unwitting linguistic innovation followed by, first, lapse into the established use of key phrases and, second, inference in accordance with their new use may lead us to mistake for a truism a substantive and paradoxical assumption that already assumes the most controversial part of what is to be established by the overall argument. To sum up: In philosophical reflection, we may unwittingly start to use familiar expressions in new ways. This may happen, first, when we want to formulate more general or abstract claims than the rules of ordinary language are geared to and, second, when an idea has captured 51 our philosophical imagination and biases our reflection. When we fail to realise that we have started to use familiar expressions in a new way, we may make two kinds of mistake. We may either draw inferences licensed only by the established rules from statements in which we employ key terms (“appears”, “seems”) in a new way. Or we may make inferences licensed only by new rules from statements in which we employ crucial expressions (“elliptical spot”, “small speck”) in established ways. Once the first sort of mistake had us adopt a philosophical doctrine, the second sort of mistake may make us overlook that we presuppose or assume that very doctrine in the arguments we subsequently construct to justify it. Thus, when we go along with Ayer’s linguistic argument, we do not realise that we tacitly presuppose most of what we want to show. And when we are persuaded by the traditional argument, we fail to notice that we explicitly assume this, as we mistake the crucial assumption for a truism. I suspect that these two things frequently happen when we find a philosophical paradox compelling: When a brief and apparently straightforward argument purports to derive a paradoxical conclusion from familiar truisms, chances are that much of what is purportedly being established is unwittingly being taken for granted from the start, and smuggled in, in ways we fail to notice. Presumably, various processes lead to, or facilitate, such inadvertent trafficking. In this section, we outlined two that result from unwitting linguistic innovation. 9. Conclusion The conflict of the paradoxical but unwarranted, hence absurd, conclusion with common convictions gives rise to groundless worries and to the construction of pointless philosophical theories aiming to reconcile the two. To satisfactorily cope with paradox in a way that removes the worry and forestalls the pointless efforts, we need to do two things: to show that the arguments explicitly adduced to support the paradoxical claim are unsound, and to expose and remove what concealed motives have us find the claim compelling, independently of all arguments. When a paradox arises through unwitting linguistic innovation followed by undue transport of implications, both tasks require a particular form of ordinary language analysis: an analysis of what familiar expressions ordinarily entail, imply, or suggest, in which contexts. By enabling us 52 to uncover processes of unwitting innovation and transport, such analysis puts us into a position, first, to identify in the arguments adduced assumptions or presuppositions that have us take for granted what is to be shown and, second, to explain why, for what concealed and fallacious motives, the paradoxical doctrine struck us as compelling. This explanation strips the doctrine of its intuitive attraction — i.e., enables us to give it up without qualms. Ordinary language philosophers have frequently, and often rightly, been taken to try to refute paradoxical conclusions by showing them false or devoid of linguistic meaning. To infer this without further ado from the finding that we would not ordinarily say this-or-that in suchand-such circumstances, one needs to assume that the verdicts of common sense are right, or that the rules of ordinary language constitute the bounds of sense. These assumptions are indeed dogmatic, and the former is precisely what is at issue in the case of philosophical paradox. Attributing this supposedly common aim and approach to Austin, even the most sympathetic critics therefore felt compelled to reject his work as failing to engage with the problems he addressed, and yielding linguistic, but no philosophical insight (e.g., Hanfling 2000: 32–3). In this paper we have seen that, at any rate in attacking sense-data, Austin had an entirely different aim and approach, which do not rely on any such dogmatic assumption: He addressed a doctrine to be rejected, first and foremost, not as false or meaningless, but as absurd, as it goes against common sense without reason. He tried to make us give up this doctrine by exposing concealed motives for endorsing it independently of reasons, namely, through an analysis of ordinary implications and the effects of their unwitting conflation. Advancing beyond Austin, we then saw where and why this is the best approach to paradox.1 1. For helpful comments on earlier versions of this paper I am indebted to Erich Ammereller, Peter Hacker, Oswald Hanfling, Marian David, and an anonymous referee for this journal. 53 REFERENCES Austin, J.L. (1962): Sense and Sensibilia, reconstructed from the lecture notes by G.Warnock, Oxford: Oxford University Press. Ayer, A.J. (1940): The Foundations of Empirical Knowledge, London: Macmillan. — (1946): Language, Truth, and Logic, 2nd ed., repr.: New York: Dover. — (1969): ‘Has Austin Refuted Sense-Data?’, in: K.T.Fann (ed.): Symposion on J.L. Austin, London: Routledge, 284–308. Berkeley, G. (1996): Philosophical Works Including the Works on Vision, ed. by M.R.Ayers, rev.ed., London: J.M.Dent (Everyman). Beck, J. (1995): Cognitive Therapy, New York: Guildford Press. Ellis, A. (1994): Reason and Emotion in Psychotherapy, New York: Birch Lane Press. Evans, J. and A. Feeney (2004): ‘The Role of Prior Belief in Reasoning’, in: J. Leighton and R. Sternberg (eds.): The Nature of Reasoning, Cambridge: Cambridge University Press, 78–102. Fischer, E. (2004): ‘A Cognitive Self-Therapy: Philosophical Investigations 13897’, in: E. Ammereller and E. Fischer (eds.): Wittgenstein at Work — Method in the “Philosophical Investigations”, London: Routledge, 86–126. — (2006): Philosophical Delusion and its Therapy, London: Routledge. Hanfling, O. (2000): Philosophy and Ordinary Language, London: Routledge. Price, H.H. (1932): Perception, London: Methuen. Robinson, H. (2001): Perception, London: Routledge. Warnock, G. (1989): Austin, London: Routledge. Wittgenstein, L. (1998): Culture and Value, rev.ed., Oxford: Blackwell. — (2000): ‘The Big Typescript’, Vienna: Springer. 54