An argument on motivating reasons

I wrote this a while back, and Errol Lord reminded me of it. I'm not sure if it's sound: thoughts more than welcome.

"I went home because my house was on fire."
"I went home because I believed my house was on fire."
It appears that the first sentence is more apt when it is true that my house was on fire. Conversely, the second sentence is more apt when it is false that my house was on fire.

Most people assume that all actions should be explained the same way: Either by beliefs, or by things believed. Further, most people assume that people cannot act because of things that are not the case (you can't go home because your house is on fire if your house is not, in fact, on fire). From these assumptions, it seems to follow that all actions should be explained by agents' beliefs. That is, motivating reasons are identical to agents' beliefs.

(I'm going to talk only about beliefs, and not their usual partner desire. I suspect the same points could be made with respect to desire if the necessary changes were made.)

This argument is supported by the assumption that one cannot act for something that is not the case. But if agents can be mistaken about their beliefs, then this premise actually supports the claim that beliefs cannot be reasons either. Perhaps one must act because of one’s beliefs about one’s beliefs. But if one can be mistaken about one’s beliefs about one’s beliefs, then... A regress seems to ensue.

Let me explain this more fully with an example. Let’s imagine that I make a bet with a friend that heavier objects fall faster than lighter ones. I state that my reason for such a bet is that heavier objects do, in fact, fall faster than lighter ones. We can imagine an onlooker reasoning: "Alex’s reason for this bet cannot be that heavier objects fall faster than lighter ones. This is because that is not the case. So his reason must be his believing that heavier objects fall faster than lighter ones (this reason does not justify his action: it is a motivating, not a normative, reason)".

But now we complicate the story. When I was at school when I was younger, I learnt that heavier and lighter objects fall at the same rate. When I made the bet, I had forgotten this fact. But, in most other contexts in my life, let’s imagine that I act as though they do fall at the same rate. Indeed, if someone other than my friend – with whom I have a very antagonistic relationship – suggested that objects fall at the same rate regardless of their weight, I would agree with them. In this case, it seems at least somewhat plausible to suppose that I do not, in fact, believe that heavier objects fall faster than lighter ones.

In light of this fact, the onlooker must revise her reasoning. "Alex’s reason for this bet cannot be his belief that heavier objects fall faster than lighter ones. This is because he has no such belief. So his reason must be his believing that he has a belief that heavier objects fall faster than lighter ones".

But this is absurd. First, if I complicated the example further (don't worry, I won't!), then we can imagine the onlooker being forced to conclude that my reason must be my belief that I believe that I believe that heavier objects fall faster than lighter ones, and so on. In fact, there may be some sufficiently complex case where the regress is infinite. Then, assuming that all actions have the same kind of explanation, then all actions should be explained by some infinite regress of beliefs. This cannot be correct. Second, even at this simple stage this is clearly implausible. If it was implausible to suppose that beliefs were motivating reasons, it is even more implausible to suppose that our motivating reasons are our beliefs about our beliefs.

An objector might suggest that where I have gone wrong is in the assumption that a person could be mistaken about their beliefs. But most agree that this is possible for desires, so why not with beliefs?

In summary, if things believed cannot be motivating reasons because they are sometimes, despite what the agent thinks, not the case, then beliefs cannot be reasons either, because they are also sometimes, despite what the agent thinks, not believed.