Belief

2. Two problems for LOTH

First, unless the claim that mentalese exists is trivialized – no matter what neuroscience reveals about how the brain processes information, what it reveals will count as the brain containing mentalese – LOTH involves risky speculation about how our brains work. The theory gives a hostage to fortune. Some are happy to accept this. If neuroscience reveals that there is no mentalese and that we do not process information in a sentential manner, we should say that we do not have beliefs and so embrace eliminativism about belief (see Eliminativism). This is, however, very much a minority view.

Second, LOTH leaves the intimate connection between belief and behaviour obscure. On the face of it predictions about the behaviour of highly complex organisms like ourselves should be enormously difficult. Trees bend in the wind whereas we put on jumpers, go inside houses, lean into the wind, cancel our games of tennis, or whatever. Unlike trees and simple machines, we respond to stimuli in enormously varied ways. Nevertheless we are quite good at predicting human behaviour. We all make many successful predictions of the following kind: someone who has uttered the word ‘Yes’ on hearing the sentence ‘Would you like to come to dinner at 19.30 on the 21st?’ will arrive around 19.30 at the house of the person the sentence came from. What we do, of course, is use hypotheses about what people believe and desire and predict in terms of the rule that subjects will tend to behave in such a way that they achieve what they desire if what they believe is true. Our subject’s ‘Yes’ tells us what they desire, and what we predict – their turning up at the named time – is behaviour that will achieve their desire for dinner.

Now we noted above how LOTH explains the way belief contributes to causing behaviour. In the same general way it explains how belief together with desire explains behaviour. For LOTH treats desires as like beliefs in being internal sentences of mentalese. The difference is that, as it is often put, the desires are stored in the ‘desire’ box, and the beliefs are stored in the ‘belief’ box. The metaphor of different locations marks the fact that beliefs and desires differ in how they relate to the world. Belief is a state that seeks to conform to how things are – the sight of coffee tends to extinguish my belief that there is no coffee near; whereas desire is a state that seeks to conform things to how it is – desire for coffee tends to bring one near coffee. The stored sentences that do the first job count as being in the belief box; the stored sentences of mentalese that do the second job count as being in the desire box. So the way belief and desire combine to produce behaviour is not a problem for the LOTH. The two ‘differently located’ stored sentences get together to produce the behaviour.

The problem, rather, arises from the fact that the connection between behaviour and what subjects believe and desire is most immediately one between behaviour and a rich system of belief and desire. Individual beliefs and desires grossly underdetermine behaviour. There is no behaviour that the belief that there is a mine near the tree, together with the desire to live, as such points to. It is, rather, a rich system of belief – to the effect, say, that there is a mine near the tree, that the mine is likely to be triggered by going near it, that moving one’s legs in such and such a way will not bring one near the tree, that there is not a bigger mine that can only be avoided by going close to the tree, that triggering mines tends to cause death, and so on and so forth, along with the desire to live being greater than the desire to test out the trigger system of the mines – that points to behaviour. When we give little illustrations of connections between subjects’ beliefs and desires and what they do, we take for granted a great deal about what they believe and desire. This is fine. It is by and large common knowledge. But the point remains that only rich systems of belief and desire have the intimate connection with behaviour. The same point could be made with the dinner invitation story. The prediction of our subject’s behaviour assumed a great deal by way of belief and desire. We assumed beliefs about what the words mean, about who uttered them, about which month was intended, …, and we assumed that there were no countervailing desires that outweighed the desire to go to dinner.

The problem for LOTH is that it takes as its starting point individual beliefs and desires. This leaves it seriously unclear what the theory has to say about the connection between a rich story about belief and desire, on the one hand, and behaviour, on the other. There is no behaviour that the individual belief that p and desire that q point to. It is rich systems of both belief and desire that point in some reasonably determinate way to behaviour. The challenge for LOTH is to find some kind of guarantee that the account of individual beliefs and desires it offers is such that if subjects have rich enough sets of these individual beliefs and desires, these rich enough sets of beliefs and desires will cause the reasonably determinate behaviour that tends to satisfy their desires if their beliefs are true.