On Smithies, Lennon, and Samuels on irrational belief
I've decided to write somewhat regular short pieces on interesting papers I've recently read. This one is about Smithies, Lennon, and Samuels (2022).
Smithies, Lennon, and Samuels (henceforth, SLS) criticise the view that there are a priori connections between having a belief with a certain content and other states that would be rational given this belief. A simple example of the target view says that believing P is being disposed to act in a way that would bring one closer to satisfying one's desires if P were true. A more complicated example of the target view, on which SLS focus, is Lewis's. According to Lewis, for a mental state to be a belief state with such-and-such content, the state must, under normal conditions, be connected in a certain way to behaviour, perceptual experiences, and other propositional attitudes. SLS deny this.
The argument goes as follows.
Consider a person with Capgras delusion. She believes, for no apparent reason, that her husband has been abducted and replaced by an imposter. In Capgras patients, this belief normally manifests itself in all sorts of ways. But we can imagine a case in which all these connections are blocked. Our patient doesn't behave as if her husband has been replaced by an imposter. But she has a strong feeling of conviction that her husband has been replaced. Intuitively, say SLS, she still believes that her husband has been replaced.
This intuition is compatible with Lewis's account. To challenge Lewis, SLS suggest that we can take two further steps.
First, we assume that all the person's beliefs are similarly disconnected from their ordinary roles. Second, we assume that the same is true for all members of her population, and under normal conditions.
According to SLS, none of that matters. As long as the person has a feeling of conviction that her husband has been replaced, she believes that her husband has been replaced.
If this argument works, it would have a lot of interesting consequences. It would show that for agents with suitable feelings of conviction, belief content supervenes on these feelings. Beliefs wouldn't just be "in the head", but "in the feelings".
I don't think the argument works. In fact, it seems to me that the way in which the argument fails illustrates that there must be a priori constraints of the kind that SLS reject.
Return to the first stage of the argument. We are considering a Capgras patient whose feeling of conviction is disconnected from everything that would normally come with such a feeling. In our patient, unlike in normal Capgras patients, seeing her husband does not cause any sensation of unfamiliarity or unease. When our patient she sees her husband, she has an immediate sensation of recognition. She is (let's say) happy to see him, remembers all sorts of things they did together, and so on. She behaves to him in a completely normal way. She has almost completely normal beliefs and feelings about him. She just has a feeling that her husband has been abducted and replaced by an imposter.
Her other feelings and beliefs are normal. She does not, for example, feel or believe that anyone has ever been abducted and replaced by an imposter. (This hypothesis is entailed by the hypothesis of which the patient feels convinced, but we assume that her feeling is completely isolated from other states that normally, and rationally, go with it.) Nor does she feel or believe that the person in her living room is someone else than the person she married 18 years ago. And so on.
The case is really bizarre. But I think it's conceivable. Intuitively, our patient has a concept of her husband, of what it is to be abducted, etc., and she can put these concepts together to entertain the thought that her husband has been abducted. When she does entertain this thought, she has a certain feeling: it "seems to her" that the thought is true.
But does she believe that her husband has been abducted? I'd say no. Let's slightly weaken the isolated character of the problematic feeling. Let's assume the person can become introspectively aware of the feeling. She will recognize a strong tension between that feeling and all her other attitudes. `Weird', she ought to think, `why do I have this feeling of conviction when I wonder whether my husband has been abducted, even though the assumption that he has been abducted makes no sense in light of my evidence, even though it clashes with everything else I believe, and even though it doesn't at all affect my actions towards my husband, nor what I say?'. She should not think, `Oh, I believe that my husband has been abducted.'
Things get worse if we move to the later stages of the argument where we assume that the person is thoroughly mad, that she satisfies none of the usual rationality conditions, even though she is a normal member of her population under normal circumstances. This makes it worse because now I don't even see how the person could have a concept for her husband, or a concept for being abducted, etc. The agent might still have a certain feeling when entertaining a mentalese sentence, `my husband has been abducted'. But how does that mentalese sentence get to mean that her husband has been abducted?
SLS gesture towards proposals to define mental content in phenomenal terms. I can vaguely see how, say, the concept of red could have its content in virtue of its association with certain phenomenal experiences. I struggle to see how we could get to the concept of my husband. Even if we could, this isn't enough. SLS can't help themselves to any connection between the person's feeling of conviction and perceptual experience. All such connections are broken. A single, isolated feeling of conviction must be enough to give itself a content. And we don't just need any kind of "content". The mad patient is supposed to feel convinced that her husband has been abducted. I don't understand how this could be a correct description of her feeling.
Let's set this problem aside. Let's assume that we could somehow ensure that the person's feeling is associated with the proposition that her husband has been abducted and replaced. The other problem would remain. Should we say that she believes that proposition – even though she doesn't have any sensations of unfamiliarity around her husband, even though she doesn't believe anything that's entailed by the proposition, even though she denies believing it, and so on? I'd say no. The agent may have a feeling that normally accompanies the relevant belief, but she doesn't have the belief.
Perhaps SLS should have aimed at a weaker conclusion. They could have argued that there are no constitutive links between beliefs and behaviour, or between beliefs and other non-mental states or events. This would be enough to clash with views like Lewis's. But it would allow them to accept constitutive links between beliefs and other mental states. We wouldn't have to consider agents whose feeling towards the husband proposition is completely disconnected from her other feelings, from her perceptual experiences, from her sensations of familiarity, etc.
It is not obviously crazy to think that phenomenal experiences have a kind of built-in representational content – a content that is not defined in terms of normal causal connections between the experience and physical events. If there is some such phenomenal content, then one can perhaps make sense of a notion of "phenomenal belief". And then one can plausibly imagine cases in which an agent's "phenomenal beliefs" come apart from the "functional beliefs" that explain her behaviour. We might then wonder whether our ordinary concept of belief is the phenomenal or the functional concept, or a combination of the two.
Here's one way to motivate the idea of phenomenal belief.
I can't rule out that I'm radically mistaken about the external world. It might turn out that I don't have hands, or that the sun won't rise tomorrow. But I know what I believe about the world. I know for sure that I believe that I have hands. I also believe that I'm disposed to behave in a way that would be reasonable if I had hands. But this could turn out to be false. It could also turn out that I'm an unusual member of a population in which my inner state normally causes behaviour that would be reasonable if one has wings. In general, any supposed causal connections between my state and the external world could turn out to be absent. But it couldn't turn out that I don't believe that I have hands. So my belief can't be analysed in physical/functional terms.
I'm not convinced by this line of thought, but as an argument against views like Lewis's, I find it more persuasive than SLS's Capgras argument.