Graham McFee

Main page  
Baker and Hacker without Baker? Wittgenstein: Text and Context
"On first looking into ...": The placing of the 'slips of paper' ... Wittgenstein, Systematicity and the Use of Philosophy
A Nasty Accident with One's Flies [Inaugural Address] More on the 'Third Wittgenstein' Idea
Everything Goes with Beer

Everything Goes with Beer

This paper encapsulates many of my concerns about exceptionlessness and occasion-sensitivity: as it notes, the style of concern with context it displays is far removed from much that goes on under that title in analytic philosophy (if one puts aside the work of Charles Travis).
A version of this paper appears as an appendix Truth, ethics & knowledge in sport research: an epistemology of sport. (2010) Routledge: pp. 215 (ISBN 978-0-415-49314-7); and a revised version in Philosophy and the ‘dazzling ideal’ of science. (2019) Palgrave/Macmillan: pp. 327 (ISBN 978-3-030-21674-0).

On one conception, philosophy is concerned with conceptual connections (all right so far!) which are exceptionless: so that, first, one can refute supposed or claimed conceptual connections by noting counter-examples (and hence raising exceptions); and then, second, the preferred conceptual connections is entailment -- which is exceptionless. And, third, we can recognise definitions as concise yet comprehensive accounts accounts of such-and-such (having an 'exact fit': McFee 1992 pp. 16-17; and McFee, 2003a). Then this view of definitions would explain why a (supposed) definition could be rejected either if it allowed in cases that were not of the idea defined or if it excluded cases that were. So such defnitions were regarded as logical equivalences; and this is another way to identify them as exceptionless.

Equally, there are places in philosophy -- which might found a different conception of it -- where the conceptual connections (although real enough) are not exceptionless. Thus Austin (1962 p. 144) asks us to:

consider the question of whether it is true that all snow geese migrate to Labrador, given that perhaps one maimed one sometimes fails when migrating to get quite the whole way.

For it seems to be true both (let us suppose) that snow geese migrate to Labrador and that this claim is not exceptionless. If there are conceptual connections (of the kind explored by philosophy) that are not exceptionless, then the model of definition just noted cannot be the dominant force in philosophy that it sometimes seems. Similarly, since we will be giving less weight to counter-examples, we will also give less weight to the sorts of thought-experiments that are refuted by the finding of an exception. So these are sets of considerations which raise, for us, the issue of the place of exceptionlessness in philosophy.

Part One: Introduction

On some occasions, 'all' and 'every' (and similar terms) are not used exceptionlessly: hence analysing their use as (always) exceptionless may mislead. In my non-professional moments, I am happy to assert:

(a) Everyone likes Chinese food;

(b) Everything goes with beer.

That is to say, I have had occasion to assert each of them; and in contexts where, if pressed, I would claim the truth of what I said. (With regard to the second, I especially recommend chocolate with beer.) And I will return to the contexts in which I used these claims later (Part Four). In those contexts, saying that these claims were true amounts to recognising that anyone who wanted to dispute them -- probably just one of those friends who studied philosophy -- should (rightly) be dismissed as a pedant, with no place at our celebration.

But whenever I feel privately happy with such claims, the demon of philosophy appears on my shoulder, asking me (first) whether it is really true that every single person likes Chinese food. Surely someone, somewhere, does not. And, even if this is currently not true in practice, surely it must be true in principle: it seems logically possible that someone who did not like Chinese food might be born tomorrow -- even were my claim presently true of the world's population. So the demon's thought is that this claim about the pervasive attraction of Chinese food cannot be true because it admits of counter-examples: hence that it is either false as it stands or (since it might be or become false) not the sort of claim one should accept just like that.

Equally, for my second case, the scope of the everything must be restricted: roughly, to foods and drinks . Then what does and does not count as beer for these purposes must be determined (factoring in, for example, the claim by Canadian friends that American beer is the result of filtering Canadian beer through a horse). Even when such points are accommodated, there are some foodstuffs which -- at least in the minds, or on the palates, of some judges -- do not go with beer. Now, our concern here is not (in either case) with these as matters of taste, or discernment, or discrimination. Once such considerations are put aside, we are again left with potential counter-examples: with the thought that my claim is not exceptionless. And, again, this is urged in principle, if not in practice.

All this concerns the demon, of course, just because (from his perspective) philosophy does -- or anyway should -- concern itself with the exceptionless, since exceptionlessness is one mark of (conceptual) necessities.

At this point, the demon reminds me that students in my predicate-logic class, faced with a statement such as "Everyone likes Chinese food", would treat it as urging that all persons like Chinese food (since this is one way to produce a categorical version). And would in turn treat that claim using the universal quantifier: "For all X" . So their account of the claim would cast doubt on its truth -- at least in principle -- since that account takes (or stresses) exceptionless as a feature. Moreover, such students (it might seem) typify one powerful inclination within the project of philosophy.

My debate with the demon comes to this: in making my claims, did I really say or assert what he is keen to deny? Or, even if I did not, should I have done so, if I were only more thoughtful, more consistent, and so on. In part, this debate concerns how to translate English into quantifier-speak. And I shall not pursue that point. But also the debate concerns when we raise objections, and what sort of objections, to general claims: and ultimately a debate about the scope or purpose of claims made in philosophy -- what is their scope?

Part Two: Ziff's Cheetahs

It has long seemed to me that the answer to the demon's queries is that I did not claim, say, or assert what he is denying. And that the reasons for this, in addressing the place of exceptionlessness in the project of philosophy, might be philosophically revealing . Philosophy has known for some time that there are questions here concerning the exceptionlessness of claims beginning "all" or "every". Thus, as Austin (1962 p. 143) reminded us, first, faced with the claim that all snow geese migrate to Labrador. we remain unfazed "that perhaps one maimed one fails when migrating to get quite the whole way". Equaly, we understand the person who -- defending the claim that all swans are white -- asserts that he was not talking about swans in Australia: for clearly we was not intending to discuss swans everywhere: he would, with justice respond that he was making not claim "about possible swans on Mars" (see Austin, 1962 p. 143). But comments such as Austin's have not been widely explored: thus. first, it is still urged that the primary feature of statements -- of which the claims just mentioned are surely clear examples -- is to be either true or false. And, second, much philosophy continues as though the units of truth and falsity here, the statements, were really just sentences; that is just linguistic units. And disputing the first of these (at least conceived detached from contextual standards of truth and falsity) will do something to unsettle the second.

So, first, I will rehearse some of the discussion, from Paul Ziff (1972 ), of the assertion. that a cheetah can outrun a man. This is, I take it, something we would all be prepared to assert, at least in our pre-reflective moments. And, of course, in not using the words "all", "every", and so on, this claim does not so obviously intersect with the issue of exceptionlessness. But it clearly does have a kind of generality -- for instance, it is not just about Herbert, our pet cheetah. Ziff's point is that this claim is not equivalent to "any cheetah (or every single cheetah ) can outrun any man", even if that is how we sometimes treat it. So it has generality without universality. And this is the thought I shall elaborate.

In discussing his claim, Ziff makes six crucial points, which I begin by merely stating (since later discussion will both elucidate and justify them):

  • the claim itself is "as standard as can be" (Ziff, 1972 p. 128);
  • we grant that the claim can be used to say something true ;
  • we are not excluding very fast human runners (Olympic champions, say), cheetahs with broken legs, cheetahs wearing leg-weights (what Ziff [1972 p. 128] calls "an encumbered cheetah"), or even the slow horned cheetah (Ziff, 1972 p. 139);
  • we are also not failing to consider such cases, as though they might have slipped our mind;
  • our remarks are not equivalent to asserting that, if we put a man and a cheetah on a racetrack, the cheetah will outrun the man -- for a cheetah turned loose might just sit "lazily in the sun" (Ziff, 1972 p. 128). As Ziff (1972 p.128) puts it, "'can' is not 'will.";
  • what we said cannot be 'translated' (without change) into some other form of words -- and especially it cannot be put in terms of merely "some", and so on. My point here (like Ziff's) is partly that we need a way of saying what we did in fact say, in saying that. Of course, this is not a discussion about words as such, but about the contrasts which (sometimes) those words are used to draw. So this point amounts to claiming that there is something importantly distinctive being said. And, then, to look for an 'analysis', or some such, would mean finding something equivalent.

This last insight is fundamental here; and really wants careful consideration. But a few (haphazard) examples make the point. Thus, the claim was not that some cheetahs can outrun men -- for then we could ask which; and we have no more specified that (nor intended to, nor needed to) than we have filled out to which cheetahs (or men) our original claim referred. So anything in that direction makes what was said specific in ways it presently was not; and thus cannot be equivalent to what was said initially. Hence it must be rejected for that reason.

A key feature of our original claim was that it did not imply some exceptionless connection or relation. And that point is quite general. For instance, Ziff (1972 p. 132) considers the claim:

"A tiger is a large carnivore".

This too is a standard sort of claim; and one we'd take to be true. But, as Ziff correctly urges, it will not be exceptionless -- especially given the powers appropriated to itself by philosophy. For, first, a newborn tiger is a tiger all right, but not a large carnivore (Ziff, 1972 p. 129). Once that case is granted, we have 'open season' for thought-experiments: we will have to consider (say) "an adult tiger shrunk to the size of a newborn tiger" (Ziff, 1972 p. 132), and so on. Faced with such a case, one might -- in the spirit of disambiguation (see later: Part Three) -- assert, "A normal unshrunken adult tiger is a large carnivore". But did we really mean that initially? Clearly not, for thoughts of, for instance, shrinking (or not shrinking) had not crossed my mind.

What goes for tigers as large carnivores goes for cheetahs outrunning men: we shall soon have to rule out not only encumbered cheetahs but (Ziff's list):

foot-bound [since birth] unencumbered cheetahs, three-legged cheetahs, cheetahs being forced to run after being force fed and so on and on. (Ziff, 1972 p. 129)

The "so on and on" is revealing in two ways. First, it highlights, quite correctly, that there is no finite totality of such (imaginable?) cases to consider: so that we could never turn that "so on and on" into a complete list of them. Instead, any list we draw up will (in principle) admit of yet more cases. Second, it recognises that there is no single principle for generating such cases. So we cannot seek to deal with all (apparent) counter-examples in a uniform way: for instance, by specifying that our cheetah is not encumbered. For that still admits of plenty of ways of identifying a particular cheetah that will not fit our (general) bill -- say, the three-legged cheetah, which is after all still a cheetah!.

Yet suppose we came up with some cheetah-ly equivalent of the general formula, "a tiger is a large carnivore", true in the ways that one is true. (And here our best example to date was a highly qualified claim about unshrunken adult tigers.) The thought would then be that, in this new claim (whatever it was), we had said of cheetahs what we originally claimed, in saying they could outrun men. Anyone happy with this transformation is being asked to accept that this new version was what he/she really meant all along. That seems at least one bridge too far.

For, typically, some highly qualified version (say, speaking of "some cheetahs", for instance) does not bring out what we meant to say all along. To elaborate this idea, Ziff (1972 p. 129) comments:

It is true that some -- never mind which -- cheetahs can outrun a man. But it is also true that some -- never mind which -- men can outrun a cheetah (namely those able ones racing against disabled cheetahs). And it is also true that some -- never mind which -- cheetahs cannot outrun a man and some -- never mind which -- men cannot outrun a cheetah. So, so far, men and cheetahs would seem to be on a par with respect to running. But they are not: a cheetah can outrun a man. [two commas added]

And it is this last assertion that we wanted to make all along. Since it does not amount to either or both of the others, we have not yet arrived at what we do want to say. Yet both of the claims rightly put aside by Ziff in that passage aim at exceptionlessness. Notice, too, that our 'summary' here is more or less a reiteration of our original assertion. But what did that assertion amount to? And how does it relate to the powers and capacities of extant cheetahs?

Suppose that, as a result of disease (say), only a few weak cheetahs were left in the world. All (or anyway most) men could outrun these poor specimens. Does that necessarily mean that, now, cheetahs cannot outrun men? No, it does not. To clarify that point, Ziff (1972 p. 129) asks us to consider the assertion:

"Cheetahs don't have horns".

Again, we would be willing to assert this; and doing so seems right. But what should we then make of a cheetah with grafted horns, say (Ziff, 1972 p. 128)? Or even a result of genetic engineering? Certainly, these are not typical cheetahs. Perhaps that thought can move us forward.

In this context, I am reminded of a kind of joke: asked, on average, how many arms a human being has, you might be inclined to answer, "two". And certainly Jo and Jane Normal, average human beings, have two each. But perhaps you are wrong (ha! ha!). For if we look, instead, to the average in the sense of the statistical mean well, I imagine that there are few (perhaps no) human beings with three arms, but there are definitely a large number with fewer than two arms -- as a result of injury and such like. So, for humans, the statistical mean for arms will certainly be less than two.

This is relevant, of course, in highlighting something about our target when we claimed that a cheetah can outrun a man: we were looking for a kind of average man and a kind of average cheetah -- but not in the statistical sense of the term "average". (We might even think of this, on a parallel with dog shows, as the 'breed standard' for cheetahs, or whatever.)

For it is typical cheetahs in typical conditions -- don't want anything too hilly, perhaps -- that can outrun typical men. In addition to its not being what was said initially, such a claim seems rendered 'true by definition', as it were: that what we make of all these typicallys guarantees it. Whereas my original claim might have been false: at the least, it seems contingent. (After all, that I commented on Chinese food and not, say, on Indian food was supposed to reflect differentially 'how the world is'.)

It may help to explain that thought a bit. For might we have varied concerns here? Suppose we are teaching a biology class, and want to show our students a cheetah: which of the powers, capacities, features, properties, and so on, of cheetahs -- that is, instantiated in some cheetah or other, including the drugged and encumbered ones -- will we select?

Notice that we certainly do not require for our purposes all of the properties of extant cheetahs, both because we recognise that, for many of these properties, lacking them would not mean that the animal was not really a cheetah and because some will conflict with others (say, height or age), but without this being problematic -- these will not be features our chosen cheetah exemplifies to our class. So, to answer this concern, some but not all of the characteristics of extant cheetahs are required. But the cheetah we choose will, of course, instantiate some characteristics not, in this way, required of it as a sample: it will, for instance, be of a certain age. (We only rule out very young and very old -- for a cheetah!) And it is the context that creates the requirements. Thus, for these classes, we will certainly want an unencumbered, four-legged, two-eyed, unhorned cheetah -- that one will (if it wants to) outrun the kind of man we'll need when we get round to showing these same students an example of homo sapiens.

Complications arise, though, when we want something other of our cheetah: for instance, we believe that a guard-cheetah would keep our property safe. Now we might prize fierceness, or perhaps the appearance of it, over some of the features mentioned above. Our cheetah isn't required to run-down any burglars, just to scare them away. We might, therefore, not care if this cheetah could outrun the man mentioned previously. Indeed, we might be happy to find that it stayed put, and kept guarding us. Again, we might be happy with a fierce but one-eyed cheetah: it would be pretty useless in our biology class, but perfectly 'adapted' to the tasks we are now setting it (see McFee, 1992 pp. 108-109). Is it a cheetah? Certainly it is! But does it have all the properties of the abstraction from cheetah-hood, the 'breed standard' for cheetahs? No, it does not. And, in particular, this cheetah would not, and could not, outrun the 'breed standard' for homo sapiens. But who every said it would or should? I certainly did not, although -- as above -- I agree that a cheetah can outrun a man.

Note, first, that this conception of a 'breed standard' is therefore a normative one, highlighting some features, good of cheetahs (or whatever) in some contexts; second, that we cannot give it genuine descriptive substance while we insist that it be exceptionless; third, that our picture of generality here cannot be expanded by expanding the 'breed standard' either in length or into a disjunctive form (say) -- there is no finite totality of such properties, such that, in noting them, we would say all we could say about (in this example) cheetahs: there is no all here.

I draw four morals from this consideration of ideas developed from Ziff's chapter :

1. That the claim made (about the cheetah) is not equivalent to the disambiguated versions, those using "some", and so on -- to say more about that, I turn (in the next section: Part Three) to some more general comments on strategies that use disambiguation.

2. That there is something revealing here in the fact that no finite totality of properties circumscribes either the 'exceptions' to our original general claims (that everyone likes Chinese food; that everything goes with beer) or the 'class' of cheetahs.

3. That there is something revealing in this case that applies to many occasions either when we do use the terms "all" or "every" or when we might be thought (say, by die-hard logicians) to imply the use of such terms. And, again, I'll comment on this later (Part Four).

4. That, as Ziff (1972 p. 129) puts it, "[t]hese cheetah cases are not curious, special or rare" [my comma]. Hence that our conclusions here extend beyond these few cases.


Part Three: Defeating Disambiguation -- some morals from Travis

It might seem (it seemed to Ziff [1972 p. 131]) that one response here is to make explicit certain further specifications which -- it might be claimed -- were implicit in what was said: this might be the route of disambiguation. And clearly there are some cases where this is just what is required: say, where you confuse a question about the temperature of your Indian food with a similar-sounding (because orthographically indistinguishable) comment about its degree of spiciness -- "Is that vindaloo hot?". Moreover, it might seem that a kind of implicit disambiguation applies in the case of, say, the assertion:

"I want to buy some alligator shoes." (see Ziff, 1972 p. 63)

For only occasionally do we need to add, "Yes, of course for my alligator". The key point for these cases is that there is a clear and recognisable ambiguity (at least once it is pointed out). Disambiguation in such cases may do all that we require. That may give the illusion that such a disambiguation strategy will always work: that is, in every case, we can clarify what was really said or meant with disambiguation.

Further, this strategy might seem suitable to deal with our cases concerning the general attractiveness of Chinese food or the compatibility of beer. For it might seem that the way to make these claims true -- once one recognises the exceptions -- is to disambiguate them so that, on this new understanding, they will be exceptionless. And, of course, this strategy can work in cases where, say, "all" or "every" are replaced with "some". Not all swans are white, but some (and perhaps most) are.

We have already had cause to be suspicious of the general force of such a disambiguation strategy. As Ziff showed us, when faced with (apparent) exceptions to the claim about a cheetah outrunning a man, that claim was not readily disambiguated in a way that preserves our sense of it as true or appropriate. But our response to the thesis about disambiguation should be careful. If the issue is whether, faced with a particular case, generating a particular puzzlement, we can always (or, at least, generally) resolve the perplexity through disambiguation, I am inclined to say, "yes" -- and that is (for my money) much of the usefulness within philosophy of the drawing of distinctions. But often more is required: a disambiguation such that no perplexity whatever can remain. So, can we deal with all perplexity in this way, globally rather than case-by-case?

Here, I would answer, "no", with my reasoning reflecting, in two different (though related) ways, the fact that there is no finite totality of candidate counter-cases here. And both aspects bear a fairly obvious debt to the writings of Charles Travis, who (therefore) I shall cite in this context. The more direct response points out that whatever disambiguations between notions or 'senses' are adopted, they will deal only with this case (the case before us), since the manner of drawing the distinctions here is set by this context: that is, by this puzzlement or perplexity. There can be no guarantee that it could (or would) deal with other cases. And new cases, unconsidered cases, can always arise. So, as Travis (1997 p. 119) notes, this "is a dead end", because our newly-drawn contrast must now be confronted with "novel cases, which it may count as describing correctly or not" (Travis, 1997 p. 119) .

This can introduce Travis's view of the occasion-sensitivity of what is said: that the same string of words might, on a different occasion, amount to something different -- indeed, might even have a different truth value. Thus, consider the contribution of colour-shades, stains, holes, fading and such like that would permit correctly asserting, on one occasion, that a particular curtain was red, but where, on another occasion, that same combination would make it false that the curtain was red, given who now was asking, or the interest in redness on that occasion. So these different occasions set different constraints on the redness of the curtains. As Austin (1970 p. 130) noted:

The statements fit the facts always more or less loosely, in different ways on different occasions for different intents and purposes. What may score full marks in a general knowledge test may in other circumstances get a gamma. And even the most adroit of languages may fail to 'work' in an abnormal situation.

The thought of some exceptionless account of the relation of statements to facts here seems unsustainable.

The other response (which I take more directly from Travis) focuses more closely on the disambiguation strategy. For this can seem to exploit a natural or obvious way to distinguish cases. But is this really always so? Faced with the recognition that what we had previously called "gold" (and taken for one substance) is actually two substances -- with different chemical composition, atomic number, and so on -- it is not obvious which of them I should continue to call "gold", which I should call something else; for instance, "fools' gold". As Dummett (1978 pp. 428-429) noted:

What is clear, at any rate, is that the word "gold" did not, in advance of the introduction of a theory and technique of chemical analysis, have a meaning which determined the course to be followed.

For that decision might suggest that, by the term "gold", I always really meant the element with such-and-such atomic number, rather than the compound. Indeed, we might continue to call both of them "gold" (as though it were some larger covering concept), and then distinguish within that category: say, between assay gold and fools' gold. But must one of these be what was meant all along? For the disambiguation strategy assumes that there were two different ideas, each clear, distinct and non-occasion-sensitive, which need to be distinguished from each other. But that is not our view.

Moreover, it seems implausible that semantics alone will distinguish, for all cases, one option (and not the other) as true. Consider a different example (Travis, 2008 p. 112): "the leaves are green", said of painted leaves -- now, is this just plain true or plain false? Part of the difficulty is that we cannot decide which, since it seems natural (on a particular occasion) to say one and, on another occasion, the other. But if we cannot decide which is to be (obviously) right, then neither can be obviously right .

Disambiguation might seem to offer a way forward: we could talk of, say, "painted-green leaves" -- although this too might not be unproblematic; for instance, some (otherwise) green leaves might be painted green. These would then be, as it were, double-green. Still, at first blush, such a strategy might look promising.

To consider the adequacy of the disambiguation option, then, we turn our attention to one of the (apparently) disambiguated pair: if we can generate occasion-sensitivity in respect of it, then the disambiguation strategy looks unhelpful. So our argumentative tactic is, for each of the difference 'senses' of "green", to find it still amenable to occasion sensitivity. Of course, one cannot prove that such still-contentious cases can always be found; but it looks promising. As Travis (1985 p. 200 ) urges, "[d]istinctions in the same and other veins can be proliferated indefinitely as ambiguities in expressions of a language cannot". Hence occasion sensitivity should be distinguished from (mere) ambiguity.

First, if disambiguation sorts out this case, how could I be sure that it could sort out all such cases? I could not. For any level of 'grid', why might a yet finer mesh not be needed to avoid, once and for all, unclarity or ambiguity? And, if this is so, we need the idea of an ultimately fine mesh -- but we have no basis for such a conception.

Second, if disambiguation works for these purposes or in this context, we could view it as dealing ('completely') with the issue in this context: for our puzzle is the one now, the one in this context. As a result, the puzzle depends on the context. Then both the puzzle and the relevant 'disambiguation' are occasion-sensitive:

If you call a leaf green, the leaf must count as green for what you said to be true. But when must it so count? There is more than one thing to be said in calling a leaf green. (Travis, 2008 p. 278)

As Travis notes, there are a wide variety of cases here, including the term "green" used to mean inexperienced ("But that is a different kind of green!"). Still, even sticking to some version of (roughly -- but much turns on how rough) green-coloured and leaves, we have green leaves painted green as well as brown leaves painted green; we also have leaves dyed; and we have 'naturally occurring' green leaves. In this last camp, we have those uncontentiously green, and those contentiously green ("Isn't this one a little yellow?"; "Hasn't that one started to turn red?"). And many more. For some purposes, what I am happy to call "a green leaf" -- and right to do -- will not count for you: as an example for your biology class, neither my painted leaf nor my yellowing (but still green) leaf fits the bill. And you have no truck at all with my jade leaves!

A case which is both simpler and nearer to that desired by the advocates of the disambiguation strategy will help to consolidate our response. Notice that, in some cases, the hearer may not know how to take an utterance, although it is in fact clear once the context is taken into account -- while it might have misled me, it is not misleading. Suppose I work as assistant to both a marine biologist and a cook (McFee, 1992 p. 121: Travis, 2008 pp. 189-190; Austin, 1962 pp. 65-66); both are interested in red fish -- the cook in a surface swimming red fish, the biologist in a deep swimming one. Now the instruction, "Bring me a red fish", is clear once I know which of my masters uttered it. Although I might end up confused (I don't know who said it), it is not confusing: I might be unable to fulfill the task but -- in this case -- what I am being asked is perfectly clear (Travis 1989 pp. 18-21), in the sense of prescribing one behaviour as a satisfactory response. The differential interests of, say, biologist and cook yield different interests (or different questions, or different issues) even though both result, for me, in the request, "Bring me a red fish". But particularising the matter offers this clarity: once I know who said it, then whatever follows will follow. So, in each 'world', there is really no ambiguity. There is, in general, no prior way of guaranteeing in which of these 'worlds' we find ourselves,: still (in practice), participants in that 'world' will not be perplexed.

Further, one cannot simply deal with the issues raised either by dismissing them as merely pragmatic (rather than semantic ) or by treating them as amenable to solution through disambiguation -- which may have been a first thought faced with (say) the 'red fish' case: that one could just speak more exactly of "surface-swimming fish". And then, the thought might be, the occasion-sensitivity has disappeared: what I hear determines which fish to bring with no need to explore the details of the occasion or the speaking. But, as we have seen, there are a number of reasons why this answer is inadequate:

(a) What is said was always clear in these cases; once we know what occasion this is, or what 'world' we are in, the order or claim becomes clear (and clearly either satisfied in such-and-such way or truth/false [respectively]). So there is really no ambiguity to be remedied.

(b) There is no natural or obvious way to distinguish cases. My red fish story might seem to support the misunderstanding: in trying to simplify the case I indicated how the expressions might be taken. But that is a red herring (!): we do not in general have a sense of the precise number of possibilities here -- for this reason, my artificial cases (with, seemingly, only two outcomes possible) may induce recognition of occasion-sensitivity while clouding our understanding of it.

(c) If there really were just two options here, we should be able to say which is the correct one on a particular occasion, which is wrong. As Travis (2008 p. 9112) notes, "one must choose in a principled way. What the words mean must make one or other disjunct plainly, or at least demonstrably, true." And there seems little hope of this.

(d) If we take some term in English to be ambiguous (and therefore amenable to disambiguation), "there must be a way of saying just what these ambiguities are: so a fact as to how many ways ambiguous they are" (Travis, 2008 p. 112). And, again, there seems no hope of finding some fixed number here.

(e) Further, the 'new' terms, now suitably disambiguated, are still amenable to occasion-sensitivity -- as when the cook would ask for a surface-swimming, red-skinned fish if he were making fish tacos and a surface swimming, red-fleshed fish if he were making fish stew. Having disambiguated the expression "red fish" once, I am still left with two occasions, with different satisfaction-conditions, where I bring red fish; and on which the word "red" means red, "fish" means fish, and so on. And the artificiality of the case should not disguise its power for us. For we cannot in general predict how certain expressions might be used (that is what Chomsky calls "creativity" [Lyons, 1970 p. 86]: that particular terms can be understood in indefinitely many sentences, most of which we have not previously encountered); moreover, we cannot predict which of the many understandings (Travis, 2000 p. 4) of a particular situation is the appropriate one for a particular occasion -- although (consonant with earlier points) that we would typically understand it when we encountered it!

As this line of reply emphasises, the notion of an 'ideal language' that could deal with all cases (including the new ones) is a fiction, because there is no finite totality of conditions to be met in describing a particular scene; and, relatedly, that even if a particular issue could be accommodated by, say, modifying what was urged (so that it covered exclusively some of the cases originally envisaged: for instance, by specifying which kind of red fish), the problem simply recurs -- the point is that there is no basic level of description or explanation here (see McFee, 2000 pp. 130-131). So that we cannot simply 'disambiguate' [see (c) above] down -- or up -- to that level: there will always be a sense of, say, the term "green" which may escape such disambiguation:

To grasp a thought is to grasp what it would be for things to be in a certain way -- that way things are according to it. To grasp that by grasping what being green is, one must grasp the appropriate way of counting as green or not; a particular way in which facts about [say] the leaf may count for or against its being as it is according to that thought. (Travis, 2008 p284)

As Travis (2000) highlights, this conception of meaning and understanding would be radically revisionary of many of the questions or issues or assumptions of much philosophy of language (at least, in the Anglo-American analytic tradition). Of course, this is not the place to explore such topics.


Part Four: "All", "Every", and Contexts

The argument of the previous part extends the considerations from Ziff (Part Two), by showing us flaws in the general strategy of aiming at disambiguation to clarify what was said -- say, in claiming that a cheetah can outrun a man.

What has all this discussion shown us in respect of the implication of all or every which the cheetah example showed us to be misplaced? In particular, what (on that basis) can we say about the apparent counter-cases to the claim that everyone likes Chinese food or that everything goes with beer? We might start by returning to Ziff (1972 p. 136), who notes:

Speaking of a cheetah, as one does when one says 'A cheetah can outrun a man', is like modelling a cheetah in clay or like doing a pictorial representation of a cheetah.

So that, whereas certain things must be true of any particular cheetah (either it is or is not one metre high at the shoulder, say), these need not be determinate for the cheetah in "A cheetah can outrun a man" -- they amount (in terminology from Ziff [1972 p. 136]) to conceptions of a cheetah. We arrive at our counter-cases by making determinate some or other feature of this conception, a feature that will be determinate for any extant cheetah. And this allows us to see that claims about such conceptions are not exceptionless if that is read to mean that there can be no counter-cases to them. But that is not how such claims should be read. For (recall) there is no finite totality of properties here, either as part of this conception or applicable to extant cheetahs. Hence the conception cannot cover all cases. As Ziff (1972 p. 139) notes:

No picture captures everything: even the best picture of a cat won't purr.

The point is also that there is no all or every here: there are features which, given the interest we are taking or the question we are asking (or answering), are important or central -- and there are other features important or central on other occasions, in respect of the raising of other issues. (This is just occasion-sensitivity: see Part Three.)

My thought, then, locates the scope of the "all" or "every" via a consideration of the context in which the assertion (or question) is raised: there will be occasions when "Everyone does such-and-such" will mean "Every single person, without exception, does such-and-such" , but there will be other contexts too.

So, in what contexts of their utterance might my candidate claims, about Chinese food or about beer, be appropriately thought true? And how exceptionless do these contexts require them to be? For it might have seemed -- ever since I introduced this idea (from Ziff) in Part Two -- that there was some sleight-of-hand implicit in my repeated assertion that (once properly understood) the claims of mine were true . My hope is to discharge that obligation here, by offering at least one context for each of them. Thus I offer just one brief sketch for each, presented through my dialogue, while granting that a large number might be constructed, in all directions.

(a) "Let us buy Chinese food for this motley crew of visitors -- we do not know much about their specific requirements or tastes; but everyone likes Chinese food."

(b) "No, it is not essential to open a bottle of (expensive) wine to go with this food, especially as -- if you do so -- you will need a bottle of a different wine with the next course (if you are consistent). Beer is OK, honestly -- everything goes with beer." [And, in imagining the beer in front of us, we can take this as a context where the content of the term "beer" is clear!]

Recall here my earlier characterisation of the claim that a cheetah can outrun a man: that this claim has generality without universality. This seems to fit exactly what we would also say, in the contexts in which they are plausible, better true, of my two claims:

(a) Everyone likes Chinese food.
(b) Everything goes with beer.

It would be misplaced to offer, in criticism of me, the claim that everyone just means "every single person", that everything just means "every single foodstuff", and that I should revise my comments accordingly (or read them that way). Such a suggestion faces two problems here. First, what I meant is certainly not equivalent to these revised claims -- as the possibility of counter-cases to these claims (but not to mine) illustrates. And if it is urged, instead, that my claim too is susceptible of counter-cases, that is to revise the terms of our example. For commonsense took my original claims to be true.

In justification of such 'commonsense' (if justification is thought necessary), we might even go some way towards explaining why these are just the things to say in this context. For instance, I want to be as inclusive as possible for those to whom I am giving this food: they are not just my Friday philosophy class, nor my friends from next door, but a wider (and unspecified) group. And yet others may arrive. Moreover, if they do, my claim is that they will be happy with my selection. The practice of ordering Chinese food in such contexts is, I am asserting, a sound one -- for everyone likes Chinese food. So there will be no dissenters ; there will be some food for the vegetarians, and so on. Since no particular group is identified (such that I could say, "All my Friday philosophy class like Chinese food."), I must find something general and expansive to say: that is my thought here. But the issue of its being exceptionless does not arise in this context. (Who would raise it but a philosopher-manqué -- and my people know better than literalism!) And if the issue of the exceptionlessness of my claim did arise, I could console myself by pointing out that my expansive, general everyone does not, after all, invoke a finite totality. So it is not as though -- had I wished to ascertain the truth of my claim about Chinese food -- that I should (or could) explore all the people who could (in principle) show up for dinner, to see where they stand vis-à-vis Chinese food: for their likes or dislikes are beside the point -- I need only consider those who might plausibly show up . (If you like to put it formally, there is a principle of total evidence at work here .) So I cannot be criticised for, as it were, failing to consider some group of only logically possible dinner guests.

Then, second, we have shown that we had a point here: we were not merely 'flapping our gums'. Hence we do need some way to assert what I said, in claiming:

(a) Everyone likes Chinese food.
(b) Everything goes with beer.

And we have seen that many candidate assertions -- including one with the term "some", or one that is shown false by counter-cases (as mine is not) -- are not equivalent to that from which I began. So one cannot simply reject (as somehow ill-formed) my original assertions. And that is one aspect of Ziff's recognition that such claims are "as standard as can be" (Ziff, 1972 p. 128): they do not turn on complexities of grammar or syntax, for example.

Moreover, we do have a reason to say what was said. And we can imagine it being challenged :

"Well, in this vegetarian enclave, that is not true -- since all the Chinese food you can buy near here contains meat."

In context, this challenge would make me withdraw my claim here. Yet it would not preclude my making a claim in the same words (hence, the same claim) on another occasion (as long as I was in a different place!). Nor would it incline me to add, as a parenthetical remark, "Except for such-and-such". For that would give my assertion false specificity, in two related ways: there are, no doubt, other such enclaves of which I am unaware (so the specification to just this one is misplaced) and, again, my claim in the new place does not bear on the old enclave, one way or another -- its members are too far away to get here for dinner. So I am neither including the case of that enclave nor failing to mention it: it is not relevant to the generality towards whom my new assertion is directed. Thus I am justified in sticking to my assertion (on the new occasion) even when I have encountered what others might take as a counter-case.

In this way, the force of my claim about Chinese food is clear: it is general but not universal or exceptionless. And, in context, it is true. Or, at least, that would be the best way to characterise it, if pressed.

Then, even if the expression is used (slightly aggressively?) to suggest that a dissenter is, somehow, weird -- because everyone likes Chinese food -- the thought is still to preserve a kind of generality: it cannot be to deny the existence of the dissenter, who is (after all) right there in front of the person making the assertion. So one way not to be 'weird' would be to accord with the 'breed standard' in liking Chinese food.

At this point, we can return to the thought that my claims are contingent (see Part Two). For what is really happening? In Ziff's case, we extract some features from extant cheetahs in claiming that a cheetah can outrun a man. Since we begin from the extant cheetahs, we will not (say) assert the possession of horns -- although, of course, cheetahs might have been different than they are. In that context, this is an empirical matter: this process is a kind of generalisation across cheetahs. Then, we are 'in the hands' of the changes in our world -- as it changes, we will revise what we say about it. As Austin (1962 pp. 76-77) remarked about talking cats:

Suppose that one day a creature of the kind we now call a cat takes to talking. Well, we say to begin with, I suppose, 'This cat can talk.' But then other cats, not all, take to talking as well; we now have to say that some cats talk, we distinguish between talking and non-talking cats.

So our comments assume (or build in) a certain stability to the powers and capacities of the cheetah population, real or imagined (or future). In this sense, my claim must be revised (just as Ziff's must be) were these generalisations no longer broadly true: that is, if cheetahs could not, in general, outrun men or if people did not show a marked preference for Chinese food.

However, the assertion itself (in either case) is not such a generalisation: as I put it earlier (Part Two), Ziff's remark should be read as pertaining to 'breed standard' cheetahs and 'breed standard' men. And that is a way of saying that the claim is not, after all, straightforwardly empirical, if that means that it must be susceptible to counter-cases (or, better, it is a reason to reject that conception of the empirical: compare Travis, 2004).

Further, as Austin (1962 p. 77) goes on to illustrate, the contours of concepts here too can be responsive to the values or priorities of people:

we may, if talking becomes prevalent [among cats] and the distinction between talking and not talking seems to us to be really important, come to insist that a real cat be a creature that can talk. And this will give us a new case of being 'not a real cat', i.e. being a creature that is just like a cat except for not talking.

So we cannot completely detach the scope of what is said from the interests and concerns of those saying it: and this goes for our claims about all or every as much as for the remarks about cheetahs (or cats).

Am I saying, using Ziff's terminology, that we have at work here a conception of all or of every? Perhaps I prefer to put the point another way, since the terminology does not quite fit. But, at the least, we have identified both the generality of my claims and the fake universality in some of the exclusions noted earlier. For it seemed as though I was not entitled to claim everyone or everything unless I had checked -- and put aside -- all the potential counter-cases. But that is mistaken. Since there is no finite totality of such cases (no all here) that is a spurious requirement. We can never consider all cases (or every case) because there is no all (nor every): unconsidered cases could always arise, as we saw earlier. Thus we cannot legitimately be criticised for failing to consider them all. (Indeed, a more sophisticated account of the philosophy of logic here might suggest that our view -- that these claims here are not exceptionless -- is the only defensible one.) Thus, if we took all and every always to be necessarily exceptionless (as above) there would always be exceptions, in some contexts, to claims deploying them. Hence, in those contexts, the terms would be useless. Yet they are useful in my contexts: they say (in these contexts) just what I need said.

Of course, I am not asserting that there are no contexts in which terms such as "all" or "every" should be taken exceptionlessly. Instead, my position is to wait and see. And, of course, I might be happy that in certain contexts -- say, those of natural science (see McFee, 2000 pp. 126-131) -- exceptionlessness was secured by fiat. But I an denying that we can always argue seamlessly from the presence of the terms "all" or "every" (and so on) to a conclusion about exceptionlessness.

Thus far, I have urged both that my claims about (say) everyone were, in context, not to be read exceptionlessly and that this reflected a wider issue concerning the scope of general claims. At root, both these points are explained in terms of occasion sensitivity. What we have seen, though, is a kind of constraint on such occasions -- that we read them charitably. Hence, if it makes no sense to talk (exceptionlessly) of all people in this context (as liking Chinese food) -- because there is no finite totality -- then that cannot be a way to understand what was said in using those words on that occasion.


Part Five: Conclusion

Ziff (1972 p. 141) offers a useful slogan, as a summary of my overall point here; namely that "[o]ne speaks and hopes to be understood." So that at issue in considering (say) my major claims about the attractiveness of Chinese food and the general compatibility of beer with other foodstuffs is (first?) what I said -- in saying those words on that occasion -- and (then?) the truth of what I asserted. For, in line with Ziff's slogan, this is what you must understand in order to understand what I said. To understand what I said is, in part, to take from my saying those words, in that context, both what I meant and where I was right (when I was). Here, I have discussed some pitfalls to this kind of understanding: in particular, those associated with assumptions of exceptionlessness, especially in making sense of the generality in my claiming (in context) both that everyone does such-and-such -- in my case, liking Chinese food -- and that cheetahs have so-and-so properties (in this case, that they can outrun men). And my plan was to draw on insights from the second to help us with the first. In neither case should we see the claim as merely about some, or this or that specific group. And, in both cases, what was said was (in context) arguably true. Further, you will misunderstand if you take them differently: in particular, if you assume that such general claims were (or were supposed to be) exceptionless.

A thesis of this paper has been that terms such "all" and "every" do not necessarily indicate exceptioinlessness, and that other locutions operate similarly: our example was, "A cheetah can outrun a man". Moreover, that this is an unproblematic feature of our understanding -- and hence of our language. So that would suggest a commitment here to occasion-sensitivity, at least as an agenda for research. But why should we care? Does this leave us anywhere useful? For I take it that philosophy is not, in general, much vexed by either the speed of cheetahs or the tastiness of Chinese food.

The short answer is that there is a tendency within (especially) philosophical readings of some claims to treat every claim of this kind as exceptionless. And a symptom here would be the search for counter-examples to these claims. Then it can seem a philosophical error to have made such an exceptionless claim: in saying men did such-and-such, was I really ignorant of the two men who did not? How could I miss (or, worse, ignore) such an obvious counter-example? Well, the 'error' requires that I say "All men" (for example, are selfish) -- and I did not say that. In part, then, the concern is to capture what was said or meant in context, rather than to import from outside a 'reading' of what was said. This is one place where the occasion-sensitivity of the remarks is crucial. Thus, as Travis (2008 pp. 300-301) recognises:

The point is: whether that is so depends not merely on the fact that it is this that is to be so or not, and on the way things are, but also on what one is to count as things being that way, where this last is a genuinely substantive question.

Nor can we always resolve the matter by disambiguation: as he continues:

Just how black does toast need to be to count as blackened? The concepts expressed in the words used to call the toast black do not answer such questions univocally. There is not just one thing that might count as toast being blackened. And so it is in general. (Travis, 2008 p. 301)

So, first, my asserting, "Men are selfish" is not equivalent to claiming that all men are selfish, where that "all" is read as exceptionless: this is the moral from Ziff's tales of cheetahs Then, second, even if I had said, "All men are selfish", that too should not (in all contexts) have been taken as an exceptionless claim: that is the moral from my claims about beer, or about Chinese food -- saying these things need not commit one to the exceptionless reading. In particular, having recognised occasion-sensitivity, we should be more hesitant about raising accusations of failing to be exceptionless. Then these are points to convey to my philosophical demon.


Coda: one example -- Freud's claims

The imposition of an exceptionless 'reading' on an author's remarks sometimes takes place ever when there is (or seems) compelling evidence in the opposite direction. To give one brief example, Freud's ideas are commonly presented as offering exceptionless generalisations -- and these are the assailed with counter-examples. A focus on the finding of exceptions offers one way to read the Popper-inspired criticism that Freud's work is not falsifiable (compare Cioffi, 1970). But in general, and throughout much of his work, Freud was fairly careful as to the scope of his claims: in particular, he was aware that those who offered summaries of his chief ideas tended to falsify those ideas in the process. (And he understood why.) As he wrote:

qualifications and exact particularisation are of little use with the general public; there is very little room in the memory of the multitude; it retains only the bare gist of any thesis and fabricates an extreme version which is easy to remember. (Freud, [1905] 1953 p. 267)

So that what is taken from his theoretically-precise accounts becomes, instead, something inexact when presented to a general audience. Thus, for instance, Freud carefully denied the idea of a 'Freudian symbol' for dreams , although that appears one 'legacy' of his work.

In fact, much of the general, abstract criticism of Freud (or of psychoanalytically-inclined psychology more generally) is precisely directed at its failures to generalise fully or at its (supposed) spurious exceptionlessness. In particular, consider in this context the criticism that Freud's work could never be scientific because, by its nature, science works with exceptionless generalisations ("laws"); and then Freud is criticised either for not dealing in such exceptionless generalisations or, in aiming to deal with them, for producing general claims that were false (because not exceptionless) . I hope that we now know enough about spurious exceptionlessness to need a fuller scholarly treatment of exactly what (say) Freud was really asserting before we would, in these ways, convict him of the sin of claiming exceptionlessness for some thesis that admits of exceptions (and especially obvious exceptions).

I should stress that I hold no particular brief for Freud or for psychoanalytically-inclined psychology: that is just an example here. And certainly meeting the criticisms of Freud noted here would not be to meet all lines of criticism. My point is only that a judicious weighing of the claims of Freud involves us reading him aright. And that takes us back to Ziff's slogan, quoted at the beginning of this section. Elsewhere (McFee, 2005) I have noted that judicious attention requires the sort of close reading of Freud we would offer to other theorists, as a criterion of our taking them seriously. Here, I am recording another constraint on understanding (with Freud as my example): that we must not mistake for an exceptionless claim one that is not (and was not meant to be). And, in particular, that -- having recognised occasion-sensitivity -- we should be more hesitant about raising accusations of failing to be exceptionless. These are points, too, my philosophical demon should recognise.


[1] And I am not committing myself to the thesis from Sondheim's Sweeney Todd: The Demon Barber of Fleet Street (1979) that "everybody goes down well with beer".

[2] Thus, for instance, characterising the view of the second edition of Principia Mathematica, Ramsey (1931 p. 114) writes: "Mr. Russell would say "Unpunctuality is a fault' really means something like 'For all x, if X is unpunctual, x is reprehensible' ".

[3] Importantly, this is not the typical debate about contexts or occasions current within philosophy: thus, for instance, of the special issue of The Philosophical Quarterly devoted to "Contextualism: Problems and Projects" (Vol. 55, No. 219, April, 2005) only the paper by Charles Travis (and the commentary on it) address the issues raised here. Thus I would agree with Travis (2008 p. 14), when reprinting his piece, that "[m]uch that currently goes under the name of contextualism about knowledge I want no truck with."

[4] Indeed, much of this paper of mine might be redundant if more students of philosophy read carefully that chapter of Ziff's work.

[5] See also Austin, 1980 pp. 143-145 on "France is hexagonal", and so on -- we might agree that "Lord Raglan won the battle of Alma" (which means, 'agree with the person who said it', or, 'agree that it was true') while granting that "there would no question of giving Raglan a medal for it" (Austin, 1980 p. 144).

[6] There is a great deal of other insight in Ziff's chapter, rewarding detailed attention.

[7] As Travis (2008 p. 99 note) says (of another example): "I will not pause to argue against the heroic view that that just means that no one can ever speak truth in calling something round". If "round" were an absolute term, he might have in mind Unger, 1975 pp. 66-67: see Stroud, 2000 pp. 42-43.

[8] The remainder of this section reworks, with various expansion and contractions and other revisions, McFee, 2004 pp. 48-52.

[9] This passage was excluded when the paper was 'reprinted'.

[10] Another useless strategy: to deny that these are genuine meaning-connections, invoking "implicature" (Grice, 1989 pp. 24-31): such ideas are annihilated by Travis 2008 pp. 19-64; pp. 65-93.

[11] Amusingly, someone reading this claim over my shoulder asked, "What about the married ones?".

[12] Recall that Austin (1980 p. 143) commented, of a similar case, "[i]t is a rough description; it is not a true or false one". Yet he goes on to grant that true should be thought of as standing for "a general dimension of being the right or proper thing to say as opposed to the wrong thing, in these circumstances, to this audience, for these purposes and with these intentions" (Austin, 1980 p. 145). But of course this is just the kind of occasion-sensitive understanding which makes what I said, on that occasion, (plausibly) true.

[13] Notice that, for this reason, "Most people like Chinese food" would not do.

[14] NB A parallel with the demands of scepticism: compare Stroud, 2000 pp. 46-49, esp. p. 48; McFee, 2000 pp. 117-118.

[15] As Carnap (1950 p. 211) states it:

in the application of inductive logic to a given knowledge situation, the total evidence available must be taken as the basis for determining degrees of confirmation

So what we cannot presently know we must treat as irrelevant.

[16] The example recalls the occasions when my friend Bob Goldman wanted to voice just such an objection.

[17] Thus he did write "this symbolism is not peculiar to dreams" (Freud [1900] PFL Vol. 4, 1976 p. 467). For there is a fund of such symbols in myths, folk-lore, and so on, from which we learn symbols -- which partly explains generality; so 'we' read the same myths, etc. and myths drawn on what is human (feelings, etc.). Further, "many of the symbols are habitually or almost habitually employed to express the same thing" (Freud [1900] PFL Vol. 4, 1976 p. 469). But asked whether we should interpret all dreams in this fashion, Freud replied: "No, not at all" (Freud [1933] PFL Vol. 2, 1973 p. 4).

[18] Both kinds of criticisms, and others, are re-cycled with animus in Gellner, 1985 (see especially pp. 150-203) -- although his topic is the Freudian or psychoanalytic movement, rather than Freud specifically.



Austin, J. L. Sense and Sensibilia. Oxford: Clarendon Press, 1962.

--- Philosophical Papers (Second Edition). Oxford: Clarendon Press, 1970.

--- How to Do Things with Words [Second Edition] Oxford: Clarendon Press, 1980.

Carnap, Rudolf Logical Foundations of Probability. London: Routledge & Kegan Paul, 1950

Cioffi, Frank "Freud and the Idea of a Pseudo-science" in R. Borger & F. Cioffi (eds) Explanation in the Behavioural Sciences. Cambridge, Cambridge University Press, 1970 pp. 471-499.

Dummett, Michael Truth and Other Enigmas. London: Duckworth, 1978.

Freud, Sigmund "On Psychotherapy" [1905], in The Standard Edition of The Complete Psychological Writings. Volume VII London: Hogasrth Press, 1966 pp. 256-268.

Grice, Paul Studies in the Ways of Words. Cambridge, MA: Harvard University Press, 1989.

Lyons, John Chomsky. London: Fontana, 1970.

McFee, Graham Understanding Dance London: Routledge, 1992.

--- Free Will. Teddington: Acumen, 2000.

--- Sport, Rules and Values. London: Routledge, 2004.

--- "The Artistic and the Aesthetic" British Journal of Aesthetics Vol. 45 No 4, 2005a pp. 368-387.

Ramsey, F. P. The Foundations of Mathematics. London: Routledge & Kegan Paul, 1931.

Stroud, Barry Understanding Human Knowledge Oxford: Oxford University Press, 2000.

Travis, Charles "On What is Strictly Speaking True" The Canadian Journal of Philosophy, Vol. 15 1985 pp. 187-229.

--- The Uses of Sense Oxford: Clarendon Press, 1989.

--- "Reply to Simmons" Mind Vol. 106 1997 pp. 119-120.

--- Unshadowed Thoughts. Cambridge, MA: Harvard University Press, 2000.

--- "The Twilight of Empiricism" Proceeding of the Aristotelian Society Vol. CIV 2004 pp. 245-270.

--- Occasion Sensitivity: Selected Essays. Oxford: Clarendon Press, 2008.

Unger, Peter Ignorance: A Case for Scepticism. Oxford: Clarendon, 1975.

Ziff, Paul Understanding Understanding. Ithaca, NY: Cornell University Press, 1972.




Return to top of page