These principles may seem reasonable enough, until one realises that
compliance with them would be – to say the least – very time-consuming
indeed! These are not in fact principles whichWnite beings could ever
employ.
In eVect, we need to incorporate the maxim ‘Oughtimpliescan’ into
epistemology. The way in which weoughtto reason is constrained by the
way in which wecanreason, and is relative to the powers and capacities of
the human brain. Even where there is complete agreement concerning
valid logical systems, it remains an open question how weought(explicitly)
to reason. But this is a question which has rarely been asked, let alone
answered. So alongside the logician’s project – and partially independent
of it, at least – is the project ofpractical epistemology,orpractical ration-
ality. In fact there is not one project here but many. One task is to describe
what principles of reasoning we should employ, given what is known or
reasonably believed about human cognitive powers and functions, but
assuming that our only goals are to obtain truth and avoid falsehood. But
then there are a number of more complex subsidiary tasks which arise
when we factor in the various other goals which people may legitimately
have when reasoning, such as the desire to reach a decision within a given
space of time. These tasks are largely unexplored, and cry out for further
attention from epistemologists and philosophers of psychology (but see
Cherniak, 1986; Stich, 1990; and Stein, 1996 for some discussion).
It is also important to remember that the development of normative
standards for reasoning (the equivalent of the ‘logician’s task’) is in some
areas still very much work in progress. In philosophy of science, for
example, the dominant tradition has, until recently, been the positivist
one. This tradition – which is standardly bothinstrumentalistin relation
to theories which go beyond what is observational, andfalsiWcationistin
its methodology – has provided the normative standards assumed by
most psychologists investigating subjects’ abilities to reason scientiWcally.
But there have also been other, more realist, approaches to norms of
scientiWc reasoning, derived from reXection on actual scientiWc practice as
well as from armchair philosophical reasoning (Lakatos, 1970; Boyd,
1973, 1983). Naturally, as scientiWc realists ourselves, we favour this latter
approach.
These disputes matter, because Koslowski (1996) has shown convin-
cingly that a whole tradition of experiments which apparently reveal
weaknesses in subjects’ abilities to reason scientiWcally, may actually reXect
a weakness in the positivistic philosophy of science upon which the ex-
perimental designs are based, rather than in the subjects’ ability to evaluate
hypotheses. So what, in a scientiWc context, may look like examples of
unwarranted ‘ConWrmation Bias’ or ‘Belief Perseverance’, may turn out to
Practical rationality 129