ABSTRACT
Inference, understood as a form of conscious and active belief-revision, has recently attracted much interest among philosophers of mind. Many writings on the topic depict inference as a kind of process. However, this assumption is, to my knowledge, nowhere explicitly justified or even reflected upon. In this paper, I argue that the assumption is wrong: while processes take time, it is not possible that inferences take time. Both claims are conceptual observations. It is therefore conceptually impossible that the mental act of inferring is a process.
Disclosure Statement
No potential conflict of interest was reported by the author.
Notes
1 In the current literature, ‘reasoning’ and ‘inferring’ are mostly used interchangeably. See, e.g., Winters [Citation1983: 202].
2 See, e.g., Wedgwood [Citation2006: 660], Boghossian [Citation2014: 2], and Müller [Citation2019: 2]. Interestingly, this assumption is also made outside the analytic tradition: see, e.g., Arendt [Citation1967: 360].
3 Compare Boghossian [Citation2019: 101]: ‘What happens when we reason our way from one proposition to another? This process is usually called “inference” and I shall be interested in its nature.’ There are some exceptions to this rule, however. Hieronymi [Citation2006, Citation2009] and Boyle [Citation2009, Citation2011] suggest that the kind of mental agency that we exercise in judging and reasoning is different from the kind of agency that we exercise in intentional action. This makes room for the idea that judging and reasoning are not processes, a conclusion for which Boyle explicitly argues. Rödl [Citation2007] and Engstrom [Citation2009] urge that reasoning and judging are acts of simultaneously holding thoughts, or parts of them, together. Similarly, Held [Citation2020] rejects what he calls process theories of reasoning and develops a Kantian account of inference as synthesis. Taking their lead from White [Citation1971], Rumfitt [Citation2011, Citation2015] and Valaris [Citation2019], propose a distinction between inferring and deducing, where inferring consists in adopting or revising an attitude such as a belief because it follows from previously held attitudes, and deducing consists merely in tracing the logical connections among propositions. While inferring, according to Rumfitt and Valaris, is neither a process that takes time nor a performance under one’s intentional control, deducing is both.
4 Mourelatos [Citation1978] develops a similar three-fold distinction between states, events and processes, building on the verb categories proposed by Vendler [Citation1957] and Kenny [Citation1963]. Mourelatos, in turn, influenced Steward [Citation1997: ch. 3], Thompson [Citation2008: part 2], and Rödl [Citation2012].
5 The following argument is indebted to Boyle [Citation2011: 7–15].
6 When I presented this argument to an audience, I was given the following counterexample. ‘Suppose that you are working out your receipts for the week. From your records, you form the belief that on Monday you received $429.24. Again from your records, you form the belief that on Tuesday you received $83.90. You add these two amounts in your head. Next, from your records, you form the belief that on Wednesday you received $638.19, which you add to the total. And so on. You end up believing that your receipts for the week were $1,943.74. You have inferred this conclusion from the contents of your beliefs about your daily receipts. But, by the time you reach the conclusion, you might have forgotten what you received on Monday. You might even believe erroneously that your receipts on Monday were $233.39.’ This, however, is not a counterexample, because, in working out the total, you proceed in steps, working out subtotals on the way. And, for each of the steps, you must have all of the sums of money that go into the subtotal present to your mind when you work out that subtotal.
7 Compare Müller [Citation1979, Citation1992] for a similar argument.
8 See, e.g., Broome [Citation2013: 237–42].
9 Boghossian [Citation2014] offers another rule-following account of inference.
10 Compare Boghossian [Citation2014].
11 Compare Held [Citation2020: ch. 7].
12 See, e.g., Setiya [Citation2013: 185].
13 Thanks to Jonas Held, Annemarie Kalis, Erasmus Mayr, Anselm W. Müller, Dawa Ometto, Sebastian Schmidt and Hannes Worthmann for helpful conversations and comments on earlier versions of this paper. I would also like to thank two anonymous referees for their thoughtful comments, and Aaron Shoichet for improving my English.