132 How to Win Every ArgumentAlthough two events might be consecutive, we cannot simply
assume that the one would not have occurred without the other.
The second might have happened anyway. The two events
might both be linked by a factor common to both. Increased
prosperity might influence our propensity to consume canned
peas, and also to engage in activities which increase the rate of
illegitimacy. Small children at gaming machines provide vivid
illustrations of the post hoc fallacy. They may often be seen with
crossed fingers, eyes closed, hopping on one leg, or in whatever
physical contortion once preceded a win. They link their random
preparations with the outcome of their luck; and in this they
differ in no wise from more adult gamblers, whose concealed
rabbits' feet and clenched-teeth incantations betray the same
supposition. If it worked once, it can work again.
Unfortunately for our predictive ability, every event is pre-
ceded by an infinite number of other events. Before we can
assign the idea of cause, we need rather more than simple suc-
cession in time. The philosopher David Hume pointed to reg-
ularity as the chief requirement, with some contiguity in time
and space. We are more likely to describe a germ as the cause of
an infection in a man if its presence has regularly preceded the
infection, and if it is found in the body which is infected.
The charm of the post hoc fallacy emerges when we leave
behind the everyday idea of cause and effect. Although we
suppose we understand the mechanisms by which one event
leads to another, Hume showed that it boils down to our
expectation of regularity. The candle flame on the finger and the
subsequent pain are called cause and effect because we expect
the one to follow on regularly from the other. Of course, we
concoct all kinds of explanations as invisible threads to link the
two, but they come down to interposing unseen events between
our first and second. How do we know that these unseen events
really are the cause? Easy. They always follow from one another.
This gap in our knowledge provides a vacant lot in which