From the ALWS archives: A selection of papers from the International Wittgenstein Symposia in Kirchberg am Wechsel

I. Introduction

When David Hume said that reason is (and ought to be) only the slave of the passions (Hume 1978 [1739], II.iii.3, p. 415), he started a tradition of thought which has become known as Humeanism. Although Hume’s own dictum can be read both as a thesis about explanation and as a thesis about justification, Humeanism has always been mainly concerned with explanation. A satisfactory explanation of an action, Humeans insist, must involve an appeal to a passion, or else remain incomplete.

Humeans’ main argument for their thesis starts with the finding that agents sometimes fail to act, although they have, and know that they have, reasons to act. From this thesis, it is inferred that if an explanation is to be complete, it must appeal to something else besides the agent’s reasons. From here, the way to the passions – or so Humeans think – is not far.

Instead of directly going into the debate over Humeanism and playing out the well-known arguments, this paper aims for an indirect contribution by placing the debate within a larger setting. In particular, it aims to trace a few interesting analogies between Humeanism and two doctrines which have come to be the targets of famous pragmatist attacks during the last century: the conception of doubting as an inner doing, famously attacked by Ludwig Wittgenstein; and the idea of sense-data as the foundation of all empirical knowledge, famously attacked by Wilfrid Sellars. The paper argues that in all three debates, the insistence on an invisible extra something (passions, in case of Humeanism) is tied to a blindness to the pragmatic role played by the relevant attributions – attributions of beliefs and doubts (in Wittgenstein’s case), of perceptual evidences and seemings (in Sellars’ case), and of practical reasons that do or don’t motivate (in the Humean question). In all three cases, the attraction of the classical, pre-pragmatist, theory vanishes once we pay attention to that pragmatic role, because doing so yields a reversal of our view of what is the background, and what is the aberration in need of explanation. In the case of Humeanism, we learn that motivating reasons are the ordinary case, while reasons that fail to motivate are the strange things in need of explanation.

The essay will proceed in three steps. The next section will sketch Wittgenstein’s treatment of doubts and doubting, and give some first hints about how talk of non-motivating reasons is like talk of doubtful beliefs. The third section will discuss Sellars’ critique of sense-data and seems-talk, and sketch out some further analogies. The last section will give a fuller account of the analogy, establishing that doubtful beliefs, seemings, and non-motivating reasons are all in the same boat with respect to the pragmatics of their attributions. In particular, all three attributions can plausibly be seen as late additions to our linguistic tool-box which are logically dependent on more fundamental attributions, and all three deal, in different ways, with human limitations.

II. Wittgenstein on doubting

From the earliest to the latest stages in Wittgenstein’s intellectual development, an important part of his philosophy has been the insistence that some sentences cannot meaningfully be doubted. This is not because of their subject matter, as one might think. Rather, the reason for which certain statements cannot be put in doubt is that they have no subject matter at all. Wittgenstein’s point concerned the logical – or, in his later thought: the grammatical – statements, whose role is not to convey information about the world, but merely to settle the meanings of the symbols or expressions appearing in them, in a way just like truth tables. These statements cannot meaningfully be doubted, because they express nothing less than the conditions of meaningful speech.

Of course, this philosophical insight of Wittgenstein’s is usually presented as a lesson about different kinds of statement. But it can also be presented as a lesson about the activity of doubting: doubting, whatever exactly it is, cannot – as a matter of logics, not psychology – take just any statement as its object.

As Wittgenstein’s philosophy matured during and after the Philosophical Investigations, this idea slowly grew into a large-scale project, namely a detailed study of the grammar of doubting. The grammatical aspects uncovered by Wittgenstein within this project are essentially threefold. (See, among many other passages, Wittgenstein 1972, §§ 67ff., 74ff., 105, 115, 122, 140ff, §§ 231, 275ff., 325, 333, 354, 472, 609ff., 648) Firstly: doubts over a statement (also doubtful beliefs) can only be attributed to a speaker to whom undoubted beliefs are also (already) attributed. Secondly, something similar is true for the attributor: only someone who already masters talk of (undoubted, ordinary) beliefs can learn to master talk of doubting (of doubtful beliefs). And finally: the doubter (as well as, trivally, the attributor of doubt) must belong to the sort of agents who can articulate their beliefs – they must be speakers, in other words. (At least close relatives: I do not mean to deny that we often treat some non-speakers as quasi-speakers.) Wittgenstein repeatedly warns that a violation of any of these constitutive rules would amount to changing the subject from doubts to mere doubting behaviour like scratching one’s head or raising one’s arms.

The upshot of Wittgenstein’s findings, which we can mark with the slogan “doubt comes after belief” (Wittgenstein 1972, § 160), is that doubting must not be conceived as an inner doing whose attribution does not commit the attributor to any more than, say, the predication of a colour does. It is a rather more complex manoeuvre, and whatever exactly it is, it rests, in some way, on the simpler manoeuvre of attributing un-doubted beliefs.

Why is all this relevant to the debate over reasons and motivations? To speak cautiously: as a hypothesis, it is not implausible that the very same grammatical features structure talk of non-motivating reasons, as well. Firstly: you can only attribute to an agent non-motivating reasons if you also attribute to her ordinary, motivating, reasons. Secondly, you can only master talk of non-motivating reasons if you also master talk of ordinary, motivating, reasons. And thirdly, someone who has reasons, and knows about them, but remains unmotivated by them (and also, trivially, someone who attributes such a state to someone) must be a speaker. They must generally be capable of articulating a great deal of their reasons, rather than just act on them. Again, failing to adhere to any of these constitutive rules amounts to changing the subject from reasons to the outward behaviour which characteristically accompanies the following of reasons.

If true, the hypothesis of an analogy between doubts and non-motivating reasons in the sketched way has important philosophical consequences for the debate over Humeanism. Since the Humean view of motivation is that reasons by themselves do not motivate, it seems that Humeans are committed to the reverse of the priority theses just sketched. For them, non-motivating reasons are the baseline, and motivation is something which is explained later. Might it be that Humans get the priority theses wrong in an analogous way to those who take doubting to be an inner doing?

III. Sellars on Seems-Talk

Let us keep this question in mind as we continue to Wilfrid Sellars’ discussion of perception and seems-talk at the centre of his Empiricism and the Philosophy of Mind. In his discussion, Sellars takes aim at two common opinions: firstly, the view that (all of) our perceptual beliefs are ultimately justified by sense-data, and secondly, the view that the subject matter of ordinary seems-claims is nothing else than these sense-data. On this opinion, seems-talk is not only logically independent from, but actually prior to is-talk: we can only attribute is-beliefs if we are also prepared to attribute seems-beliefs; and we can only learn is-talk if we also (already) master seems-talk.

Not unlike Wittgenstein, Sellars argues that the common view of the relationship between is-talk and seems-talk gets it exactly wrong. However, Sellars, who is less resistant to systematic philosophical theorising than Wittgenstein, gives a much fuller account of seems-talk which includes a characterisation of its characteristic point – and which, it turns out, can help to clarify our understanding of both talk of doubts and talk of non-motivating reasons. According to Sellars, seems-talk essentially enables the conveying of rich information regarding the score of the language game.

Sellars’ idea can be studied in the following sequence of claims. “Peter sees a red tomato” — “Peter sees a tomato which seems red to him” — “It seems to Peter that there is a red tomato” (This is adapted from Sellars, 1997, pp. 50ff.). In all three examples, Peter is credited with the (non-inferentially elicited) propositional attitude (belief) that there is a red tomato in front of him. But the speaker – the attributor of the propositional attitude – takes on very different commitments regarding the appropriateness of Peter’s belief. In the first sentence, the attributor endorses the proposition that there is a red tomato; in the second one, she endorses only the proposition that there is a tomato but resists commitment to the proposition that it is red; while in the third one, she endorses no part of the proposition she attributes to Peter. It is the transmission of this information which seems-talk makes possible, and which makes seems-talk into a powerful pragmatic manoeuvre.

It is obvious that the dependence of seems-talk on is-talk (in other words, the negation of the classical idea that seems-talk is independent from, and indeed prior to, is-talk) is directly implied by Sellars’ characterisation of the pragmatics of seems-talk. After all, making seems-claims, on Sellars’ view, involves the (self-)ascription of is-beliefs. But this is not the only way in which Sellars goes beyond Wittgenstein. Unlike Wittgenstein, Sellars clearly shows that the logical pull towards the insistence on sense-data is the consequence of a misunderstanding of seems-talk (which includes a wrong view about its logical priority). Sellars’ big idea is that the point of both is-talk and seems-talk is to render speakers’ or agents’ conduct understandable by tying it to relevant practical aspects of the world, but in crucially different ways. While is-talk works in ordinary situations without informational imbalances, seems-talk is adapted to situations in which some third parties lack access to data on which the commenter can draw. The function of seems-talk is to render these parties’ conduct intelligible, although it is, or at least might be, ill-adapted to its practical surroundings. Why does Peter grab the stone? He grabs it because it seems to him that it is a red tomato. We know that it isn’t, and yet we have successfully demystified Peter’s formerly mysterious action by tying it to the tastiness of ripe tomatoes.

IV. Conclusion

It is here, I want to suggest, that the three debates touched on in this paper have their shared core. In all of them, we have pairs of a simple attribution and a complex attribution. Both serve to tie an agent’s conduct to practical aspects of her surroundings, while the more complex one is adapted to a specific human limitation. We do not always have reliable perceptions (although many people can often rely on their perceptions) and we sometimes have contradictory evidences (although many people often have good, simple, evidence). Seems-talk deals with the former limitation, doubts-talk deals with the latter. Both render human conduct intelligible in spite of this limitation. And both are logically posterior to the simple manoeuvre from which they have been developed.

An analogous story, now, can be told with respect to talk of reasons that do not motivate. The human limitation that this kind of attribution is meant to deal with is weakness of the will. Peter ought not to eat meat and knows it. His understanding of his reasons shows not only in what he says, but also in what he does. He does not mistreat dogs, he donates money to animal welfare groups, he never throws food away and actively tries to minimise his ecological footprint. And yet he violates his reasons in his practical life by eating meat: he just cannot resist the taste of bacon. Does the taste of bacon constitute a good, countervailing reason for Peter? No – not even Peter himself would claim that. Peter is locally irrational, and it is this human limitation for which talk of reasons – genuine reasons! – that locally fail to motivate has been invented. It is a remodelled version of talk of ordinary, motivating, reasons, and it reflects the view that sometimes, the best interpretation of an agent is one that does not portray him as rational, and yet that falls short of seeing him as completely crazy. Sometimes, we make most sense by attributing the reasons we endorse to people who fail, through complex defects of their characters, to act on them as one normally would (for ordinarily, that’s what reasons do: they motivate).

It seems that this story not only vindicates our hypothesis of an analogy between the grammar of doubting and the grammar of reasons. It also supports the thesis that seeing the pragmatic point of non-motivating-reason-talk aright erases the temptation to postulate passions, just like seeing the point of seems-talk aright erases the temptation to postulate sense-data. We can elucidate action without appealing to the passions, just as we can elucidate perception without appealing to sense-data.