VARIOUS USES OF FACTORS 65

their turn nothing but algebraic devices for expressing complex
qualitative wholes in quantitative form, in order to render our
reasoning more rigorous. •

We shall see in a moment that, in examining the assumptions
made by certain logicians to account for inductive generalization
in the physical sciences, Broad has already observed that it should
be possible " to eliminate the hypothetical generating factors "
(postulated for this purpose) " and to state the case wholly in terms
of observable characteristics and their relations." His formal
proof,1 expressed in terms of attributes rather than of variables,
could, I believe, be elaborated to prove that the same conclusion
holds good in psychological factor-analysis, But what seems to
have been overlooked by the critics of factors is this—a point
strongly emphasized by Broad himself : " even if we confined our
efforts to establishing eductions, and gave up efforts to establish
generalizations inductively, we should still be pre-supposing the
existence of universal laws."

(Ill) Causal Explanation.—The view which we have
reached is very different from the view which most factorists
hold—or at any rate from the view which most readers
derive from the phraseology that f actorists employ. Whereas
we have concluded that mental factors have less objective
importance than the actions from which they are inferred,
most factorists apparently regard them as more real and
more objective. Factors are accorded a superior pre-
dictive power, not only because they are tacitly assumed to
possess a more concrete and more permanent nature than
overt actions or behaviour, but also because they are held
to be the true producers of the performances we observe
and of the correlations between them. The latter are but

criterion" (he. cit. sup., pp. 114, 307, and refs.); his arguments tempt
me to ask whether the * factors' could not also be suppressed in his
theoretical deductions as well. It will be remembered that for Thomson
the mental factors which our tests excite or sample are in the last analysis
c numerous small components,' But with the revised form of his * sampling
theory,' does not the postulate of organization into ' sub-pools' become far
more important than the postulate of c atomic elements' ? It would seem,
indeed, that this suggestion is not far removed from Thomson's own view :
for, as he himself observes, " the only reason for using the word * elements *
is that it is difficult to speak of the different parts of the mind without
assuming some * itemsJ in terms of which to think."
1 Arist. Soc. ProŁt> he. cit., pp. 38-9. See below, p. 224.