Wittgenstein and the Possibility of Meaning

Alexander Miller’s book is a valuable contribution to the debates initiated by Saul Kripke’s reading of Wittgenstein’s rule-following considerations. It offers an excellent presentation of the problems, an appealing account of the historical development of some of the main attempts to deal with these problems, and a lucid proposal for how to address some of the difficulties faced by these strategies. The book is organized in five chapters and an epilogue.Chapter 1 sets out some basic parameters for the problems to be discussed. It presents Paul Boghossian’s ACER account of rule-following, summarized thus by Miller: “An agent following a rule can be said to have accepted [A] the rule in such a way that his actions are partitioned into those that are correct [C] and those that fail to be correct relative to his acceptance of the rule, and, moreover, explained [E] and rationalised [R] by that state of acceptance” (4; letters in brackets added). Miller’s book assumes that the ACER model is correct and treats its satisfaction as an admissibility condition for a candidate account of rule-following. He argues that his own proposal comes sufficiently close to meeting this demand. He also maintains that candidate accounts of linguistic meaning and mental content have to satisfy suitably adapted versions of the ACER model. He defends this view against the claim, which some find in Wittgenstein’s Philosophical Investigations, that language should not be seen as a rule-guided activity.Chapter 2 provides an excellent presentation of Kripke’s Wittgenstein-inspired argument for the conclusion that there can’t be facts about what a speaker means by a linguistic expression. On Miller’s presentation, the conclusion is derived from the impossibility of satisfying the ACER model. He pays particular attention to dispositionalist and nonreductionist accounts of meaning facts. Defending a version of the latter is Miller’s ultimate goal. The chapter also offers insightful accounts of the obstacles faced by error-theoretic and nonfactualist responses to Kripke’s reasoning.Miller takes the arguments presented in chapter 2 as successfully establishing that rule-following, linguistic meaning and mental content cannot receive satisfactory reductionist accounts—accounts that specify in nonsemantic terms, necessary and sufficient conditions for someone to follow a particular rule, attach a particular meaning to a linguistic expression, or have a propositional attitude with a particular content. It follows that a nonreductionist approach offers our only hope of providing a satisfactory account of semantic facts. Miller is well aware that nonreductionism is itself plagued by problems, including some that Kripke himself had raised, but he proposes to assume provisionally and for the sake of the argument that these problems can be overcome and that a nonreductionist account can survive Kripke’s arguments against the possibility of facts about rule-following, meaning, and content.Chapter 3 takes this assumption as its starting point in order to concentrate on arguments that have been claimed to threaten the possibility of semantic facts even if nonreductionism survives Kripke’s challenge. The arguments, versions of which have been developed by Paul Boghossian and Crispin Wright, relate to the notion of blind rule-following. They purport to show that any account of semantic facts, reductionist or nonreductionist, will be incapable of satisfying the rationalization component of the ACER model—on any account, a subject’s acceptance of the relevant correctness conditions won’t rationalize her responses. The conclusion is meant to apply even to accounts that have survived Kripke’s challenge. Chapter 3 is devoted to a detailed presentation of these arguments.We can present the general idea of the blind rule-following problem, along the lines of Boghossian’s version, by considering the circumstances under which my doing A could count as following a rule of the form “If C, then do A!” This would seem to require that I have an intention with this rule as its content and the belief that C now obtains. It would also seem to require that my action results from an action-guiding state with the content “Do A!” and that this state is produced by an inference from my intention and my belief. But performing this inference would require following a different rule: If you have the intention “If C, then do A!” and the belief “C,” then conclude “Do A!”And if concluding “Do A!” is to count as following this rule, then an action-guiding state with the content “Conclude ‘Do A!’” will have to be produced by an inference, which will require following a different rule, or a different application of the same rule. A regress looms.Chapter 4 presents Miller’s proposal for dealing with the blind rule-following problem. It is based on the Wittgenstein-inspired idea that there is a way of grasping a rule that is not an interpretation. Grasp of a rule can also arise, Miller contends, from “training in a custom, practice or institution” (6), and this alternative, noninterpretive form of rule grasp can still guide the rule follower in her application of the rule. Miller argues that the possibility of noninterpretive rule grasp blocks the regress that gives rise to the blind rule-following problem. If my acceptance of the rule “If C, then do A!” is noninterpretive, then I can count as applying the rule by doing A without being guided by the inferential rule that starts the regress.In chapter 5, Miller discharges the assumption introduced in chapter 3 that a nonreductionist account can survive Kripke’s arguments. He argues that the kind of noninterpretive rule-following that his position contemplates has the resources for blocking Kripke’s arguments against nonreductionism, including, prominently, worries about the ability of a nonreductionist account to explain the first-person authority of self-knowledge of meaning and intention.The epilogue briefly discusses two outstanding issues for the brand of nonreductionism that the book advocates: the extent to which the position should treat rule-following as involving an essentially social element and its ability to ascribe genuine causal efficacy to content properties.This is a nuanced, tightly argued book, demanding deeper engagement that would be possible in this review. In the remainder, I’m going to restrict myself to flagging two issues. I want to focus first on the role that training in a custom, practice, or institution is supposed to play in the account. I think the issue can be seen in the following way. Suppose S has the intention to do A if C and the belief that C now obtains. Now S forms an action-guiding state with the content Do A! and proceeds to do A. What conditions would have to be satisfied in order for S’s doing A to count as acting on her intention to do A if C? Miller assumes, plausibly, that some condition will have to be imposed—that having the relevant intention, belief, and action-guiding state is not sufficient for doing A to count as acting on the intention. One condition we could impose is that the production of the action-guiding state involves an inference from the intention and the belief, but imposing this condition, Miller maintains, renders the regress unstoppable. An alternative condition that we might want to impose is that the action-guiding state needs to be caused by the intention and the belief. Miller takes this to be a necessary but insufficient condition for doing A to count as acting on the intention. The reason why it’s insufficient is the possibility of deviant causal chains (105)—there are causal routes from the intention and the belief to the action-guiding state that wouldn’t make doing A count as acting on the intention. Ruling out deviant causal chains is Miller’s reason for the bracketed qualification in the formulation of the condition he proposes: “Following a rule R is a matter of being caused (in the right kind of way) to act in a manner that conforms to R in virtue (inter alia) of having an intentional state with the rule R as its content” (98–99).The point I want to highlight is that in this formulation of the condition there is no mention of customs, practices or institutions, and my question is what role these are supposed to be playing in the proposal. Here’s Miller’s answer to my question: An intention to uphold a rule combining with a suitable belief to cause an action “in the right kind of way” is a matter of the relevant agent’s having been suitably trained or brought up in an appropriate custom or practice. So the exclusion of deviant causal chains via the inclusion of “in the right kind of way” in our account meshes well with an account of rule-following in which interpretation is superseded by training, custom, and practice, etc. (107)One way of understanding this is as presupposing an independently intelligible notion of deviant causal chain and making a substantive claim to the effect that when an agent has been suitably trained or brought up the relevant causal chains are never deviant. If this is the right reading of the passage, Miller’s claim would need more support than is offered. Wouldn’t deviant causal chains still be possible when the right training is in place? In Miller’s own example of a deviant causal chain (105), there is no reason to think that the agent hasn’t undergone the right training.Alternatively, we could read the passage as providing a definition of when a causal chain from an intention and a belief to an action-guiding state counts as nondeviant—namely, when the agent has received the right training. If the passage is read in this way, then the qualification “in the right kind of way” in Miller’s formulation above should be understood as “when the subject has been suitably trained.” But if we read it in this way, Miller’s proposal for when doing A counts as acting on the intention to do A if C is much more vulnerable, as it is far from obvious that suitable training will rule out all causal chains of the kind that intuitively would not count as making the agent act on her intention, even if Miller’s definition classifies them as nondeviant.The second point I want to highlight, more briefly, is the role that Miller assigns to abilities in his account of rule-following. He writes, “Let’s say that accepting a rule consists in having an ability to suit one’s practice to its dictates and intending to do so” (119). Does accepting a rule really require the ability to suit one’s practice to its dictates? Miller points out, correctly “that it is possible to have an ability yet be prevented from exercising it” (120). This observation neutralizes some potential counterexamples to Miller’s claim, but he maintains that in cases in which the ability is really not present, “the agent in question has not after all managed to accept the rule” (121). I don’t think this is in line with our intuitions about when someone has accepted a rule. Suppose a novice sailor is given the following rule for upwind helming: If a gust comes, head up. It is perfectly natural to say that he can accept the rule even if he is too incompetent to suit his practice to its dictates—he can’t reliably recognize the arrival of a gust or he often bears away when he intends to head up. However, Miller can’t easily abandon the claim that accepting a rule requires having the ability to suit one’s practice to its dictates, as it is needed in his explanation of the noninferential nature of first-person knowledge of rule-following: “If S possesses the ability, S can have non-inferential knowledge of the identity of the rule she is following” (126).Miller has written an excellent book, certainly compulsory reading for anyone interested in the rule-following debates. His proposal will be of interest, in particular, to those of us who are convinced that reductionist accounts of semantic notions are not to be had, but are worried about the obstacles faced by nonreductionist views

Argument (complex analysis) · Meaning (existential) · Mental state · Presentation (obstetrics) · Reading (process) · State (computer science) · Wittgensteinian philosophy and applications