Epistemology,
Bishop and Trout argue, ought to be practical. It ought to actually guide our
reasoning. But standard analytic epistemology (SAE) does not offer any
genuinely practical advice. Bishop & Trout therefore call for its
replacement by a new epistemology; one which diagnoses problems in human
reasoning and allows us to compensate for them.

This new epistemology aims to uncover the normative
assumptions of a branch of science they dub Ameliorative
Psychology. Ameliorative Psychology refers to the diverse branches of study
which focus on human reasoning and its flaws, and which also offer ways to
correct them. In the first half of the book, Bishop and Trout focus on one of
the successes of Ameliorative Psychology, Statistical Prediction Rules (SPRs).
SPRs are (relatively) simple rules that are very successful in predicting a
wide range of outcomes. There are SPRs for predicting whether a past criminal
will reoffend, whether a psychiatric patient is likely to be violent, whether a
candidate for admission to a college or university is likely to graduate,
whether an applicant for a loan is a good risk, and so on. Provided with the
right data (often surprisingly little data), SPRs typically do at least as
well, and often better, than human experts on these problems. In many cases,
SPRs outperform the experts even when the experts are presented with more evidence than the SPR uses, and
even when the expert is also provided with the prediction of the SPR. Human
experts, like the rest of us, tend to believe they have a special insight into
cases and people, and so will selectively depart from the SPRs when they get a
gut feeling that its prediction is wrong. More often than not, it is the expert
who is wrong.

Thus Bishop and Trout’s first recommendation of their
new epistemology made practical: use SPRs, when they exist. Of course, SPRs
have been developed only for a relatively few, specialized and repeated,
decision problems, so they are of no direct use to most of us, most of the
time. Nevertheless, Bishop and Trout believe, they have a lesson for all of us:
we can adapt the principles underlying their success as reasoning strategies to
a wide range of problems. Essentially, the lesson they draw is this: adopt a
cost-benefit approach to epistemology. That is, they recommend that we allocate
cognitive resources so as to maximize, not true believes, but true significant beliefs. Moreover, they
argue, we should do so in a way that maximizes our global success as reasoners. Sometimes, we should choose less
reliable reasoning strategies over more reliable, because they are cheaper, and
therefore allow us to allocate more resources to other areas. By trading off
reliability, cost and significance, we ought to seek a maximally efficient use
of our cognitive resources.

None of this is especially surprising, not, at least,
to those who have read much psychology, or evolutionary biology (in which there
is also a great deal of talk of the relative success of fast and frugal
mechanisms over more elegant but more expensive alternatives). Most of the
strategies that Bishop and Trout recommend in their practical advice section
are also familiar, drawn from the heuristics and biases tradition in psychology
(don’t neglect base rates; think of reasons you might be wrong in order to
reduce overconfidence; don’t be too quick to see causes in correlations). Lots
of the psychological findings they discuss are counterintuitive, but the advice
offered is very familiar. Nevertheless, given the well-documented tendency of
even smart and well-informed people to ignore these findings, it is worth
repeating.

Let me end by considering two related questions: how
much of a challenge to SAE is the alternative offered, and what does it leave
for philosophers to do if it is accepted? The second question first: Bishop and
Trout acknowledge that if we adopt their recommendations, philosophers will
become peripheral players in a new epistemology. I suspect that philosophers
would be even more peripheral than they realize. What is left to do, if we
abandon SAE? Many more SPRs may still be developed and the old ones refined.
Moreover, there is a great deal of work still to be done on identifying the
cognitive flaws in human reasoning. But neither of these are tasks to which
philosophers are particularly well suited. To be sure, philosophers are
intelligent people and I would expect them to be able to make the occasional
contribution, especially to the criticism of proposals advanced by
psychologists and cognitive scientists. But I doubt there would be enough work
going to keep them employed. Not, at least, as philosophers. If Bishop and
Trout are right, I think philosophers would be required to retrain as
psychologists or cognitive scientists, or move to another branch of philosophy.

There is a second possibility. Perhaps SAE could
survive alongside Ameliorative Psychology. Bishop and Trout dismiss this
possibility, but it seems to me that they are wrong to do so. They argue that
SAE is committed to what they call the stasis
requirement, which is the requirement that an acceptable theory of
justification or knowledge leave our epistemic situation more or less
unchanged. They believe that the stasis requirement is built into SAE, because
its very methodology requires it. That methodology utilizes reflective
equilibrium: the attempt to bring our intuitions (gut reactions to particular
cases) and our theories into harmony. Bishop and Trout argue that if we use
this methodology, we are committed to a strong kind of conservatism: if a
theory gives results which conflict with our intuitions, we must reject it. But
that is flat wrong: if we have better reason to accept the theory than the
intuition, reflective equilibrium counsels that it is the intuition that must
go. There is a lively debate about the use of reflective equilibrium in ethical
theory; there it is widely agreed that it can lead us to quite radical
conclusions. Moreover, the kind of reflective equilibrium most of us believe in
is wide, not narrow. Wide reflective
equilibrium requires that our judgments and theories cohere with our best
theories outside the domain we are considering. The lesson is clear: if our
theory of justification implies that beliefs formed by ignoring base rates, or
beliefs of greater strength than is really warranted, is knowledge, than the
theory of justification has to go. Ameliorative psychology can inform and
constrain SAE; it need not replace it.

In the end, I doubt that Bishop and Trout’s proposal
conflicts with SAE. If it is a theory of reasoning excellence, as they claim,
then it has a different subject matter. If it is also a theory of the
justification of belief tokens, then it is itself a version of SAE; one better
informed by current work in psychology. Rather than reading their book as a
proposal for a successor subject to epistemology as it has been practiced, I
think it is better read as a call for philosophers to get their hands dirty in
the empirical literature. So doing will produce better epistemology, but it
will still be epistemology as we know it.

© 2005 Neil Levy

Neil Levy is a research
fellow at the Centre for Applied Philosophy and Public Ethics, University of
Melbourne, Australia and is author of Being
Up-To-Date: Foucault, Sartre, and Postmodernity (Peter Lang, 2001), Moral
Relativism: A Short Introduction (Oneworld, 2002), Sartre
(Oneworld, 2002), and What
Makes Us Moral?: Crossing the Boundaries of Biology (Oneworld, 2004).

Categories: Philosophical