As explained in my earlier posts, the unifying thread in my work has been a concern with agency, and especially cognitive agency. In my first post I sketched some ideas about the sense in which judgment, even if it is not the outcome of explicit reasoning, can count as an exercise of agency on our part. But, of course, explicit reasoning is probably the most easily recognizable example of cognitive agency. In my final two posts as FA for the cocoon I want to say a few things on this topic, which is also one that I am actively working on right now.
Now, the terms “reasoning” and “inference” are often applied quite liberally. For example, humans are quite good at judging each other’s emotions on the basis of subtle facial and behavioural cues. In general, however, we have no special first-personal insight into how we do this; we can just tell that so-and-so is angry, for example, without being able to point to the evidence that makes us think so. Some people would happily call the information processing leading up to such judgments a case of unconscious reasoning or inference.
I think, however, that such usage obscures an important distinction. The processing leading up to the judgment in a case like this is, properly, sub-personal. So far as you are concerned, you just see the other person as angry (or happy or whatever). You don’t have to do anything other than endorse (or decline to endorse) this appearance. This sharply contrasts with cases of personal-level reasoning — cases in which you actively work out what to think about a given topic. (For example, having judged that your friend Mary is angry, you might engage in a bit of active reasoning to figure out why she is angry.) My focus is exclusively on reasoning of this latter, active sort.
So what is involved in the claim that reasoning is an exercise of cognitive agency? A central part of what is involved is this: necessarily, if you infer p from a set of premisses R you take it that R supports p (in some relevant sense), and come to believe that p in part because you take R to support p. Following Boghossian (2014), call this the “Taking Condition”. My work on reasoning has focused on showing how this condition could possibly be met. As it turns out, giving an account of reasoning that can meet this condition leads to some rather surprising results. (Boghossian [2014] comes to this conclusion as well; unfortunately, as I will explain in the next post, he remains too committed to certain assumptions about the nature of reasoning that make the Taking Condition impossible to satisfy.)
The goal of my recent paper “Reasoning and Regress” (Mind, 2014) was to respond to a very familiar style of regress argument against conditions of this sort (the argument is usually traced back to Lewis Carroll’s story of Achilles and the Tortoise). I still think that the main argument of this paper is sound. As I will explain, however, I now think that there is more that needs to be done.
The argument begins with a challenge: if the Taking Condition is true, then it must be possible to give an account of what reasoning is that explains why it is true — i.e., that shows what role such “takings” play in reasoning.
Now, on standard approaches to reasoning, reasoning is a causal process in which some of your beliefs (the premiss beliefs) combine to cause you to acquire a new belief (the conclusion belief). So the question becomes where, in this causal chain, we can find room for the relevant “takings”.
For present purposes I will just try to give a quick sense of why this question seems very hard to answer. If we were to say that they are among the causes of the conclusion belief, then we would effectively be assimilating their role to that of the premisses. Familiarly, however, this way lies an infinite regress. Suppose we claim that in order to reason from the premises A and “If A then B” to B you need a further premiss, perhaps as follows: “If you believe A and ‘If A then B’, then you may infer B”. Then, as Lewis Carroll famously observed, consistency would require that, in order for you to reason from these three premisses to B, you would also need a fourth premiss, to the effect that your existing three beliefs license the conclusion that B. And so on. The result is that reasoning could never get started.
How to respond to such regress arguments? My own response is to reject the idea that the “takings” required by the Taking Condition must feature as causal factors in a causal process of reasoning. Reasoning is not, in the relevant sense, a causal process. If you believe R and come to recognize that p follows from R (in the relevant sense), then your reasoning is done: you thereby believe p. Barring conditions like inattention or irrationality, believing R and that p follows from R just is a way of believing p.[1]
In the paper I defend this claim by pointing to a version of Moore’s paradox. Claims of the form “R, and p follows from R, but I do not believe p” appear to be unassertable in just the same way that claims of the form “p, but I do not believe p”.
Standard explanations of Moore’s paradox focus on the idea that asserting the first conjunct of the Moore sentencenormally conveys that the subject believes p — i.e., the very same mental state disavowed in the second conjunct. This is why it would be difficult to understand what state of mind someone who asserted a Moore sentence might be expressing.
It is important to note that a mere regular causal connection between two mental states does not make for Moore-paradoxicality. For example, there is nothing paradoxical with an assertion of “this cake looks delicious, but I have no desire to eat any”. Since “R, and p follows from R, but I do not believe p” is Moore paradoxical, this is strong evidence that the mental state expressed by “R, and p follows from R” is not merely causally related to believing p. As I already mentioned, my own view is that, barring conditions such as inattention and irrationality, believing R and that p follows from R simply is a way of believing p.
How does this help with the regress argument? Recall that the regress argument begins with a challenge: the challenge of explaining what role beliefs about support might play in reasoning. On the traditional causal picture of reasoning, this challenge has bite, because it looks as if the only role such beliefs could play in reasoning would be as extra premisses: they would have to figure among the contributing causes of the conclusion belief. This is exactly the assumption that I reject: beliefs about what follows from what play an essential role in reasoning, but that role is not a causal one.
I still think that this move works against this Lewis Carroll-inspired argument against the Taking Condition. Nevertheless, loose ends remain.
Perhaps the most obvious of these is that I have given no account of where the relevant beliefs about what follows from what might be coming from. This is a challenge that I tried to address in the paper, but — as I now think — not terribly convincingly.
Going a little deeper, in the paper I gave no positive account of what the mental state of belief must be like if it is to function in reasoning in the way I am suggesting. What is it for beliefs to stand in the sorts of relations to each other I am suggesting they do stand? In the paper I argued that standard accounts of belief should be compatible with this claim, but I did not elaborate. But I now think that in order to make progress in the theory of reasoning — and especially the challenge suggested in my previous paragraph — a more explicit account of belief is required.
In my next (and final) post I will sketch my current thinking about these issues.
[1] What type of mental state do the “takings” required by the Taking Condition exemplify? I am going to speak of them as beliefs, for one very important reason: they are open to epistemic assessment. To see this, notice that the following type of case is possible: a subject can fail to be justified in believing p, even though she infers p from R in a situation where R really is good evidence for p. Intuitively, this will happen in a situation in which the subject is not justified in taking it that R is good evidence for p. Suppose that, steeped in superstition, I take certain spots on Bob’s face to be evidence that he has been cursed, and conclude that his life is in grave danger. As it happens, the spots on Bob’s face are the result of advanced disease, and so they really do constitute evidence that Bob’s life is in grave danger. Nevertheless, it seems intuitively clear that I, in my ignorance, am not justified in taking them to constitute evidence that Bob’s life is in grave danger; accordingly, my conclusion that Bob’s life is in grave danger is not justified.
Notice that this sort of case tells against views (such as John Broome's [2013] and Chris Tucker's [2012]) which construe the “takings” required by the Taking Condition as intellectual seemings, and hence as not open to epistemic assessment.
Comments