42 THE FACTORS OF THE MIND

measure a physical or objective phenomenon, as a regression
coefficient or a covariance may under certain conditions claim to do.1

(c) The Appeal to a priori Postulates.—But, however many
tables we collect, we still cannot legitimately extrapolate the
results, unless some further assumption is made about the
total population or universe which our samples are pre-
sumed to represent. This holds of all empirical prediction.
And here once again we observe how the peculiar fallacies
that invalidate so many factorial generalizations are, in fact,
but special instances of the difficulties that surround every
attempt at reasoning by induction. No inductive inference
can be justified on formal grounds alone : certain material
postulates, generally obscure and in most cases unexpressed,
are essential to carry logical conviction. As we have already
seen, any effort to reach probabilities by inverse reasoning
implies antecedent or a priori probabilities, as well as the
explicit or a posteriori probabilities supplied by the em-
pirical research ; and, however much the mode of argument
is recast, it seems wholly impossible to escape such initial
assumptions.

In factorial work they usually take the form of certain
general notions about the structure of the mind or the
physiological working of the nervous system. Assumptions
of this kind are avowedly introduced alike by Thomson and
by Spearman. • It is true that, when made explicit, the
enthusiastic advocate of factorial statistics will often reject
these non-statistical postulates as irrelevant to the procedure
itself: Stephenson, for example, has more than once pro-
tested against " attempts to drag physiology into the pic-
ture." Other critics declare that {he very search for factors
common to groups of traits or tests is equivalent to
invoking the discredited doctrine of faculties common to
various mental processes : while others again protest that
the mere notion that the mind can be dissected by a quan-

11 have ventured to criticize the exclusive reliance placed by psychologists
on the method of correlation in one or two earlier papers (cf, [93], p. 247,
[121], p. 170 f.). " Some people have been misled into the belief that corre-
lation is the key to all the secrets of nature. In reality, its utility as a statis-
tical method is narrowly limited; furthermore, it is one of the most difficult
of statistics to explain" (Snedecor, Statistical Metbods9 1937, p.