of rationality suggests that people can be rational and – more saliently –
fail to be rational in a number of diVerent ways. Above all, a major
division in forms of irrationality follows the distinction betweenbeliefsand
desires.
While much of interest might be said about the rationality of desires, we
can only acknowledge how much we are neglecting in this area. For
example, it is debated whether there is an issue ofrationalityas such
concerning what people may or may not desire. If one moves from desires
which are merely unconventional through the whole spectrum of possible
eccentricities, to desires which are strange and ‘unnatural’, does one arrive
at a point where desires become genuinely irrational rather than just
extremely unusual and weird? Or should we accept the view famously
expressed by Hume (1751, Appendix I), that rationality applies to means,
not ends? This is a topic we shall not even touch on here.
Where desires conXict, there can certainly be a problem of sorts over the
rationality of an agent who acknowledges one desire as more important,
but in practice, in the heat of the moment, gives in to a contrary temptation
- the notorious and familiar problem ofweakness of the will. And even if
desires cannot be rational or irrational when taken on their own, one can
state certain formal principles about the rational structure and ordering of
sets of desires. For example, preferences should be transitively ordered:
subjects who prefer outcome A to outcome B and also prefer outcome B to
outcome C, thenought toprefer outcome A to outcome C.
We will leave questions about the rationality of desire aside, however,
and concentrate on belief, and irrationality as displayed in reasoning and
inference – a topic more than large enough for the present chapter. But
even when restricted to matters of belief, the question, ‘Are humans
rational?’ makes doubtful sense. Rather like the question, ‘Are humans
tall?’, it is radically incomplete without some speciWcation of either a
standard or a purpose. Moreover, as soon as one reXects, it becomes
obvious that there aremanydiVerent standards of rational belief-for-
mation in diVerent domains and diVerent areas of enquiry. The norms
governing reasoning with conditionals are diVerent from those governing
probabilities, and these are diVerent again from the standards governing
inference to the best explanation, such as might be involved in scientiWc
enquiry – to name but a few. The question, ‘Are humans rational?’ needs
either to be relativised to a particular domain, or to be understood as a
generalisation across domains.
It is also worth noting, and brieXy exploring the connections between, a
number of diVerent notions of epistemic rationality. We should distinguish
betweencreature rationality(applicable to the person as a whole),mental
state rationality(where it is a belief or other epistemic attitude which is
106 Reasoning and irrationality