of the selection task and not so badly at others. We will consider attempts
to explain performance on this task in section 4. But it can hardly be denied
that performance on the basic version of the selection task is very poor.
Given that we are dealing with a general conditional of the form ‘If
anything is anF, then it is aG’ (‘If any card has an A on one side, then it
has a 5 on the other’), this is logically inconsistent with the existence of
something which isFandnot-G. Surely, one might think, people ought to
be capable of realising that it is just as relevant to examine items which are
not-Gas it is to examine items which areF. In the face of this sort of data,
do we have to allow that people sometimes fail to make even very simple
logical inferences? Or are there limits to be set on the degree of irrationality
possible?
3 Philosophical arguments in defence of rationality
Like so many other things we do, reasoning would seem to be something
we can do more or less well. So it should be possible to distinguish between
how one ought to reasonandhow people do in fact reason. How one ought to
reason (presumably, in order to arrive at enough true conclusions and to
avoid false ones) is something studied by logic, an ancient branch of
philosophy. How people actually do reason is something to be studied by
psychology. Of course, it might be that people do reason – by and large – in
the way that they ought to reason. But surely we cannot assume without
investigation that this happy state of rationality prevails?
In general we are suspicious of the sort of conceptual analysis which
would presume to lay down constraints on possible scientiWc enquiry in an
a priorifashion. For the last thing philosophy should seek to do is impose a
conceptual straitjacket on the growth of knowledge. In the end there are
always going to be ways of breaking free from such restraints, such as
devising new concepts which are liberated from the old conceptual neces-
sities. Yet in the present instance there is undeniable plausibility in the
thought that reasoning can only besobad and still bereasoningat all.
Despite the distinction between how people do reason and how they ought
to reason, are there not limits to the extent that these two can diVer? Must
it not be the case that most of the time people reason as they ought to?
After all, not just any old transition between thoughts can qualify as a
piece of reasoning or inference. There must at least be some diVerence
between a piece of reasoning and mere daydreaming! Or so it might seem.
So we need to review arguments which seek to show that rationality is
necessary or indispensable or otherwise guaranteed, and determine what
they do establish.
Philosophical arguments in defence of rationality 111