‘Wittgenstein’ by Severin Schroeder, in: O’Connor & Sandis (eds), A Companion to the Philosophy of Action (2010)
The contributions of Ludwig Wittgenstein (1889-1951) to the philosophy of action consist mainly in his discussions of the concepts of a voluntary action and of a reason. In both cases he rejects the prevalent causal accounts.
1. Voluntary action
Inasking: ‘what is left over if I subtract the fact that my arm goes up from the fact that I raise my arm?’ (PI §621) Wittgenstein is bringing out how tempting it is to think of voluntariness as a psychological occurrence that needs to be present on top of the physical movements involved in an action. For, after all, the same physical movements could be voluntary or involuntary. Hence, the traditional philosophical account of voluntary action (versions of which were held by Descartes, the British Empiricists, William James, and Bertrand Russell) is this: For a bodily movement to be voluntary it must be causedby an act of will; without such a cause the same movement would be involuntary. Wittgenstein was to criticize this view as an instance of the common philosophical tendency to construe all words as names of objects, events or occurrences (cf. PI §1; Schroeder 2006, 128-34, 181-5). In this case words used to characterize an action as voluntary (like ‘will’ and its cognates) are uncritically taken to denote some mental occurrence. Wittgenstein offers three objections to the traditional causal account of voluntariness:
(i) The acts of will postulated by the theory do not exist. If we take an impartial look at what goes on in our minds whenever we move our body voluntarily, no suitable mental events causing the movements come to light. However, the elusiveness of acts of will tends to be obscured by philosophers’ selective attention, when they focus on only a few especially favourable examples, such as this one: ‘I deliberate whether to lift a certain heavyish weight, decide to do it, I then apply my force to it and lift it’ (BB 150). Here we have some occurrences that could, without absurdity, be thought to constitute willing: some anticipatory thinking of the action, an act of resolve, a sensation of bodily effort. And now we take our ideas about voluntary action from this kind of example and assume lightly that those ideas must apply to all cases of willing (BB 150). But of course not all cases are like that. We frequently do things without any such preliminaries. Ordinary speech, for instance, is often entirely unpremeditated and effortless, yet not for that matter involuntary.
(ii) Willing is thought to be a mental occurrence, but a mental occurrence must be either voluntary or involuntary. That leads to a fatal dilemma: If the mental act of willing is itself subject to the will, in order to be proper willing it would have to be willed. But then we are launched on an infinite regress: For the event of willing to be voluntary it has to be caused by an earlier event of willing; but that earlier event, too, in order to be voluntary would have to be caused by yet an earlier event of willing, and so on ad infinitum — which is absurd (cf. Ryle 1949, 67). So it seems more promising to deny that willing itself could be subject to the will: ‘I can’t will willing’ (PI §613). But that sounds odd as well. For now it would appear that ‘willing too is merely an experience … It comes when it comes, and I cannot bring it about’ (PI §611). But now the whole idea of voluntariness, of being in control of one’s actions, seems to be lost. That must be wrong too (PI §612). The dilemma shows that the whole question (whether or not willing can be willed) is misbegotten. Willing is not the sort of thing of which it makes sense to ask whether it is voluntary or involuntary. ‘Willing’ is neither the name of an action, nor of a passive experience. It is not the name of a mental occurrence of any kind.
(iii) According to the traditional account, a voluntary bodily action is a bodily movement caused by a mental act. Thus, on this theory I bring it about that, say, my arm rises. But in fact, Wittgenstein objects, I don’t (PI §614). I don’t do anything else as a means to effect the rising of my arm. In particular, it cannot be said that I contract certain muscles in order for my arm to go up, for I don’t even know which muscles need to be contracted for the arm to go up. (It is rather the other way round: I could raise my arm in order to bring about the contraction of whatever muscles are involved in the process.) Nor do I bring about bodily movements by acts of wishing or deciding. Wishing that something may happen is actually incompatible with doing it voluntarily (PI §616). The word ‘wish’, like ‘hope’, implies that one is not fully in control of what will happen. If I wish my arm to rise and, lo! it does — it wouldn’t be my own action and I’d be very surprised (Z §586b). A decision to raise my arm, on the other hand, is of course likely to lead to my raising my arm; but it does not just cause my arm to go up. Again, I’d be rather surprised if it did. It would not be my own doing (PI §627). A decision to do something occurs before the action and cannot be regarded as part of it. Hence it cannot figure in the analysis of the concept of a voluntary action.—
The traditional causal account of voluntary action must be rejected. Words like ‘voluntary’ or ‘willing’ do not stand for some distinctive mental occurrence that must precede and cause a movement for it to be voluntary. How, then, is the word ‘voluntary’ used? According to Wittgenstein, we should not expect the answer to be an exciting revelation. The concept is a familiar one, so its philosophical elucidation can only be a reminder of what in practice we are all familiar with. ‘Voluntary movement is marked by the absence of surprise’ (PI §628). I am not a third-person observer to my own behaviour: I cannot look on with interest to see what will happen next, and then perhaps be surprised by it. Voluntary movements are characterised by a special surrounding of intention, learning, trying (Z §577); one can be ordered to do them (Z §588), and one can carry them out in different ways, instantiating familiar patterns of expressive behaviour: readily, reluctantly, hesitatingly, cheerfully, carefully or carelessly (Z §594).
2. Reasons & causes
Wittgenstein’s principal claim about reasons (and motives) is that they must be distinguished from causes, for one can give the reasons for one’s actions with authority, whereas one’s statement of the causes of one’s actions can only ever be a fallible hypothesis (BB 15). In the early 1930s, he considers the following example:
Let us suppose a train driver sees a red signal flashing and brings the train to a stop. In response to the question: ‘Why did you stop?’, he answers perhaps: ‘Because there is the signal “Stop!”’. One wrongly regards this statement as the statement of a cause whereas it is the statement of a reason. The cause may have been that he was long accustomed to reacting to the red signal in such-and-such a way or that in his nervous system permanent connections of pathways developed such that the action follows the stimulus in the manner of a reflex, or yet something else. The cause need not be known to him. By contrast, the reason is what he states it is. [VW 110-12; transl. changed; cf. PLP 121]
In the discussion of this example Wittgenstein makes three points to set reasons apart from causes:
(i) A reason, in this case, is a rule that justifies the action. It is not a hypothesis as to what happened, which might be falsified through further observations. The driver, in this example, ‘could have also given this rule if he had not gone by it’, and it would have been equally correct (VW 110-12). — The rule in this case is: ‘When a red signal is flashing trains must stop.’ However, what the driver gave as his reason was not exactly this rule, although he referred to it by speaking of ‘the signal “Stop!”’. Rather, he made the empirical claim that that signal was flashing at that moment. By itself the semantic rule doesn’t provide a reason to do anything; it needs to be combined with the observation that its antecedent is fulfilled at the time. Of course it may not be necessary to point that out when one’s interlocutor is already aware of it. In such a case, the reason explicitly stated may indeed be nothing more than a rule.
However, this point cannot be generalised, since, obviously, not all reasons involve semantic conventions or rules. Often, giving one’s reason involves citing a causal regularity instead (e.g., ‘Whisky gives me a headache’).
(ii) A reason or motive given in answer to the question ‘Why did you do that?’ is comparable to an answer to the question ‘How did you get here?’. It is ‘the specification of the route one has taken, hence the description of a singular process, not the specification of a cause which always involves a whole host of observations. For this reason we say too that we know the reason for our action with certainty … but not the cause of an act’ (VW 424; cf. BB 15). — Indeed, there is nothing hypothetical about describing the route one has taken, say: ‘It occurred to me that p, and then I did X’. But for one thing, not all causal judgments are hypothetical and in need of confirmation by repeated observations (PI §169; CE408). For another thing, mere succession of a thought and an action that could be justified by that thought is not sufficient to make the content of that thought the person’s reason for acting. For example, looking at half a bottle of wine it may occur to me that since it’s been opened it won’t keep, and then I proceed to drink it all up. Yet it’s easy to imagine a context in which that thought would not in fact have contained my reason for drinking the wine (perhaps I’m well aware that I would have drunk two glasses regardless of whether it meant finishing up a bottle or not).
So far then it would appear that a causal construal of the relation between reason and action cannot be ruled out. However, pace Donald Davidson, a causal link cannot be the criterion for something’s being the reason that’s operative, for (as Davidson himself was forced to admit) the thought of a reasonmay (not just precede, but) trigger an action without being or containing the agent’s reason for it. The thought may, for example, just serve as a reminder which leads the agent to act, but for a different reason. Moreover, although the occurrence of the thought that the wine in an open bottle won’t keep may cause me to act, what could be invoked as a reason, as a justification of my action, is not the occurrence of the thought, but the content of the thought: that the wine in an open bottle won’t keep. As Wittgenstein explained to Waismann:
The attending to the rule can indeed be the cause for the rule being followed. … [But] the cause of an action can never be referred to, to justify the action. I may justify a calculation by appealing to the laws of arithmetic, but not by appealing to my attending to these laws. The one is a justification, the other a causal explanation. [PLP 123]
In any case, as Wittgenstein was to realize later, it is not always true that giving one’s reason is like describing the route one has taken. Not all reasons are brought to mind before the action. In general, knowledge, beliefs, interests and preferences can inform our actions without having to be brought to consciousness prior to their behavioural manifestations. There are countless things I believe without ever wasting a thought on them (e.g., that the chair I’m sitting down on is sufficiently stable to support me (PI §575)), which, however, I may bring up when asked to give reasons for some of my past behaviour. ‘The reason may be nothing more than just the one he gives when asked’ (AL 5; cf. PI §479). And that, in fact, is the point to stress in order to account for the grammatical difference between reasons and causes:
(iii) Agents have first-person authority about their reasons for their actions: What they sincerely claim to betheir reason is what we call their reason (VW 30f., 110f.).
With certain qualifications such first-person authority applies even to reasons given for one’s past actions. Elsewhere Wittgenstein considers the remarkable confidence with which we are able retrospectively to state our intentions (i.e. one kind of reason for our behaviour), especially what we meant to express by our words or what we were going to say (PI§§633-63). The following points emerge:
(a) One knows what one was going to say or wanted to say, and yet one does not read it off from some mental process which took place then and which one remembers (PI §637).
(b) My words do not report what happened on that occasion, they are a conditional statement about the past. ‘They say, for example, that I shouldhave given a particular answer then, if I had been asked’ (PI §684).
(c) My utterance is a reaction to what I remember of the situation (PI §§648, 657, 659). That is to say, remembering the context, the situation and a certain amount of details, I will now say: ‘I wanted to φ’; or, ‘I did it because p’.
(d) This is the language-game: We ask people for their reasons, and under certain conditions the explanations they give, even if retrospective, enjoy a privileged status. The conditions are: First, the agent’s claim as to his reason must be sincere and not conflict with what he expressed (by words or deeds, including the action in question) at other times. Secondly, the reason cannot have been a fact of which the agent was not aware, nor a supposed fact which the agent did not believe to (or knew not to) obtain. If these conditions are taken to be fulfilled, an agent’s avowed reasons will be accepted. More than that, they will, as a matter of fact, be the agent’s reasons, for the concept of an agent’s reason is the precipitate of this language-game together with the considerations given by those conditions.
The point of such a concept is easy to see. An agent’s proffered reason will give us an insight into his character. It tells us what considerations he regards as justifying the action in question (at least in a weak sense of ‘justify’: as making the action understandable from the agent’s point of view), or would so regard given the information and interests he had at the time. Assuming that people’s general views and dispositions remain fairly stable over short periods of time, we can generally trust people to be reliable in expressing subsequently what they would have been able to say at the time of action. Anyway, the justificatory aspect of explanations in terms of reasons is of paramount importance to us. Asking people to give reasons for their behaviour we challenge them to justify it; to tell us (if they can) why it wasn’t a bad (or silly) thing. The question when this justification was (or would have been) thought of for the first time may be quite irrelevant.
Wittgenstein’s principal claim that we can always be wrong about the causes of our actions, but not about our reasons, is most persuasive where reasons and causes are logically independent. Thus, the train driver’sstated reason for stopping (that the stop signal was flashing) is independent of any causal account to explain:
(a) What made the agent become the kind of person who responds to such reasons.
(b) Physiological processes involved in the action, rather than causing the action.
However, there are some causal explanations of a different kind whose falsity would seem difficult to reconcile with the truth of an agent’s sincere statement of his reasons. Where the reason given is that a certain event occurred (e.g., the flashing of a signal) it seems plausible to hold that it also implies a causal explanation:
(c) What perceptible event occasioned the agent to act.
Thus, from the driver’s professed reason it is natural to draw the causal explanation that the flashing of a red signal caused him to stop the train. Wittgenstein, at any rate, believed that although we are more interested to explain human behaviour in terms of reasons, it mayalso be susceptible of correspondingcausal explanations. Hence although signals and linguistic utterances are primarily taken to give people reasons to respond to them, we can also regard language as a mechanism:
It is clear that language is used for occasioning [veranlassen] people to take actions. It is used for purposes like a mechanism and it is a mechanism. [VW 100f.; cf. PLP 122f., PI §495]
In the present example, Wittgenstein suggests, as a parallel causal account, that perhaps the driver’s ‘action [of braking] follows the stimulus [of the flashing light] in the manner of a reflex’ (VW 112f.).
Now the following problem arises: How can our fallibility about causal explanations be reconciled with our first-person authority about our reasons in cases where our reason seems to imply a certain causal explanation? In other words, if I may be wrong in thinking that a certain perceived event caused me to act, how can I be safe from error in declaring the occurrence of the event to be my reason for acting? Let us consider separately the two ways in which the causal statement may be false:
(i) The event in question did not in fact occur.
(ii) The event occurred, but did not in fact cause the action.
Ad (i): Suppose the driver was mistaken in his impression that there was a red signal. Would it still be correct to say that the reason why he stopped was that a red light was flashing? Jonathan Dancy thinks so. He argues that explanations such as ‘His reason for doing it was that p’ are not factive: they report the considerations the agent regarded as justifying the action, without committing the speaker to the truth of those considerations. For ‘a thing believed that is not the case can still explain an action’ (Dancy 2000, 134). Dancy concedes, however, that there may be something like a conversational implicature to the effect that if the speaker simply reports the agent’s reason he is naturally taken to endorse it. Hence, if we regard an agent’s professed reason as false we will normally distance ourselves from it by inserting an expression like ‘he believed that’. So in the error case we may prefer to say: