In the first chapter it was mentioned that the scholastic dictum, “Nihil est in intellectu quod nonpriusfuerit in sensu“, was too narrow; it is in fact equivalent to Hume’s criterion that for a word to have meaning it must denote something with instances.
It is now clear exactly why this is too narrow; there is no instance denoted by the word “gravitation”, and gravitation can be in the intellect even though it cannot be sensed. It is perhaps noteworthy that among early philosophers Berkeley, who was much against the use of words without a corresponding idea, concede that there was a legitimate use of words like “gravitation”.
In that chapter it was indicated that for the scientific outlook a concept must be capable of of being related to perception, directly or indirectly; it is now clear what is the precise way in which a concept is indirectly related to perception — it is by the mechanism of testability by deduction. We may also say that Ockham’s razor expresses this: entities that cannot be related to perception even indirectly are unnecessary and not to be introduced.
J. O. Wilson (2013) Foundations of Inference in Natural Science, London: Routledge, pp. 50-1. The original edition was published in 1952. The book outlines views of scientific inference developed since the end of World War I up to the 1950s (see PhilPapers).