https://orcid.org/0000-0002-1509-0649
![Sample our Humanities journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5939/TOC-Subject-Sample-2021-Humanities.jpg"})
ABSTRACT
An influential way of distinguishing inferential from non-inferential processes appeals to representational states: an agent infers a conclusion from some premises only if she represents those premises as supporting that conclusion. By contrast, when some premises merely cause an agent to believe the conclusion, there is no relevant representational state. While promising, the appeal to representational states invites a regress problem, first famously articulated by Lewis Carroll. This paper develops a novel account of inference that invokes representational states without succumbing to regress. The key move is to reject the tempting idea that the relevant representational states are causally prior to inferences. I argue, instead, that an inference constitutes the relevant representational state. To infer is thus—in the very drawing of the conclusion—to represent the premises as supporting the conclusion, and thereby to commit to that support relation.
Disclosure Statement
No potential conflict of interest was reported by the author.
Notes
1 I take the term ‘mental jogging’ from Broome [Citation2013: 226–7].
2 Proponents of the Standard View include Thomson [Citation1965], Wedgwood [Citation2006], and Boghossian [Citation2014].
3 We must understand the tracing non-factively. The Standard View allows an agent to infer from
even though the premises do not support the conclusion.
4 The Standard View faces a version of the infamous problem of deviant causal chains [Davidson Citation1973]. The account that I develop here avoids the problem, but this is an issue for another occasion [Blake-Turner Citationmanuscript].
5 Several other theorists have recently developed similar views [Neta Citation2013; Valaris Citation2014; Hlobil Citation2019; Koziolek, N. Citationforthcoming; Marcus Citationforthcoming]. Briefly considering these views will help bring mine into relief. But that will have to wait until after my own view is on the table. See note 24.
6 Taking should be sharply distinguished from Boghossian’s [Citation2014: 5] ‘Taking Condition’. Boghossian defends a version of the Standard View: his Taking Condition entails both Taking and Priority.
7 Several distinct regresses can be extracted from Carroll’s text [Moktefi and Abeles Citation2016; Besson Citation2018].
8 Since inference is a piece of mentation and not a relation between abstract contents, ‘premise-conclusion transition’ is shorthand for ‘premise-beliefs-conclusion-belief transition’.
9 I’ll sometimes drop the ‘causally’ qualifier on ‘prior’ to avoid tedium, but it should be understood.
10 Making (C) self-referential would not fix the deep problem. Since Taking is only a necessary and not a sufficient condition on inference, the challenge still arises of distinguishing between non-inferential premise-conclusion transitions that happen to involve taking states, and genuine inferences.
11 Wright [Citation2014] argues, by way of a dilemma, that it is insuperable. I don’t think that Wright’s dilemma is decisive, but it is suggestive: without a clear origin for taking states, the threat of a new regress looms large.
12 Hlobil [Citation2014: 424–6] raises a similar complaint, in a slightly different context.
13 Korsgaard explores a general version of this question in much of her work (e.g. [Citation2009]).
14 A Standard-View theorist might construe the causal relevance of an inference’s corresponding taking state in a variety of ways. A taking state’s jointly (along with the premises) causing the conclusion-belief is just one of these ways, which it is useful to focus on for expository purposes.
15 For similar concerns to the Explanation Problem and Rational Distinctiveness Problem, see Marcus [Citation2012] and Neta [Citation2019].
16 I use the first-pass version of Taking for readability, but the official construal is in section 2.
17 In addition to regress worries, some argue that the Standard View over-intellectualises inference [McHugh and Way Citation2016; Siegel Citation2017].
18 One might wonder how a transition can constitute a state. I construe inference as a causal process from (considering) premises to drawing a conclusion. Drawing a conclusion is an action, and actions can constitute states. (Some readers may prefer to construe an action as constituting the onset of a state. Nothing here turns on this.) Under the right conditions, the action your breaking the vase constitutes the state your being in deep trouble. Similarly, an inference is a process-terminating-in-an-action that constitutes a taking state. The whole process, rather than just believing the conclusion, constitutes the state in order to keep the connection between premises and conclusion tight enough to preclude deviant causal chains [Blake-Turner Citationmanuscript].
19 For more on constitution and similar notions, see Wilson [Citation2014].
20 The notion of a goalscoring kick is not useful enough to soccer fans to be something of which anyone keeps track. But it is worth sacrificing soccer usefulness on the altar of clarity in our pursuit of an account of inference.
21 One crucial disanalogy between soccer and inference is that whether or not a kick is goalscoring doesn’t seem to have much to do with the intrinsic nature of the kick. By contrast, whether a premise-conclusion transition is inferential is intrinsic to the transition. See section 5.2.
22 I construe functional roles liberally, so that even transitions can have functional roles by having certain inputs or outputs: for inferential transitions, the output is of constituting a commitment to the premises’ supporting the conclusion. But, if one prefers, one can think of drawing the conclusion itself as playing the functional role, rather than the whole premise-conclusion transition as doing so.
23 As I construe it, whether a premise-conclusion transition plays the hereby-commit functional role is due to the intrinsic nature of that transition. This is importantly different from whether a kick is a goalscoring one.
24 It’s worth outlining how my account differs from other recent attempts to reject the Standard View without abandoning Taking. Neta [Citation2013, Citation2019] is neutral on whether taking states are efficient causes of drawing conclusions, whereas I deny Priority. Moreover, the contents of Neta’s taking states are very sophisticated. They are self-representing, and depict relations of ex post justification. These complicated representations are grist to the mill of those who complain that Taking over-intellectualises inference. My account relies on much simpler taking states, which committally represent the premises as supporting the conclusion. Valaris [Citation2014] and Marcus [Citationforthcoming] argue that taking states constitute, rather than cause, inferences. On my view, by contrast, taking states are constituted by inferential transitions. This allows my account to avoid problems that Valaris and Marcus have with generalising their accounts beyond deduction. Koziolek, N. Citationforthcoming disjunctive, knowledge-first, account has taking states (rationally) caused by inferential transitions, whereas on my view taking states are constituted by such transitions. My account has two main advantages over Koziolek’s: mine is unified, rather than disjunctive; and it avoids having to take knowledge as primitive. Finally, Hlobil [Citation2019] rejects Taking, arguing that takings are not states at all. Rather, he holds, to take some premises to support a conclusion is to attach ‘inferential force’ to the argument from those premises to that conclusion. There are advantages to construing takings representationally, however, and Hlobil’s case against doing so is not decisive. Moreover, despite his best efforts, the notion of attaching inferential force remains somewhat obscure.
25 This point stands even if, ultimately, we cannot spell out what it is for a transition to play the hereby-commit functional role in terms completely independent of inference. Of course, it’s open to Boghossian to elucidate the notion of rule following while leaving it unanalysed, but that needs to be done.
26 Neta [Citation2013, Citation2019], Hlobil [Citation2019], and Koziolek, N. Citationforthcoming pursue similar strategies.
27 We should not overstate the force of the intuition, however. Compare action. While some intentional actions involve prior intentions, others involve only intentions-in-action: the agent’s intention is manifest through her acting, rather being prior to it [Anscombe Citation1963]. Similarly, it’s intuitively better to construe some inferences as involving only what we might call takings-in-action: the agent’s taking is manifest through her drawing the conclusion.
28 Thanks to Aliosha Barranco Lopez, Michael Della Rocca, Dan Greco, Maxime Lepoutre, Moya Mapps, Laurie Paul, and John Phillips for helpful discussion. I’m especially grateful to Matt Kotzen, Joanna Lawson, Ram Neta, Silvan Wittwer, and Alex Worsnip for comments on multiple drafts. Finally, thanks to two anonymous referees for this journal, whose generous feedback greatly improved the paper.