UNIVERSITY OF CALIFORNIA
When theoretical assumptions and experimental methods prove useful in the laboratory, they begin to be taken for granted. Periodic challenges to such firmly entrenched conceptions are to be welcomed. If the attack is successful, it prevents the perpetuation of an error; if it fails, it may nevertheless perform a useful function by forcing a reexamination and clarification of prevailing views. New experimental and analytic techniques developed in the course of the controversy may prove of lasting value regardless of the outcome of the debate. The current attack on the traditional conception of associative strength by exponents of one- trial learning ( Estes, 1960; Rock, 1957) may have such beneficial effects in the long run. However, as we shall try to show, the facts and arguments which have been put forth thus far have failed to inflict decisive damage on the theory of incremental growth or to put the all-or-none hypothesis on a firm empirical footing.
Acquisition . The question at issue is whether associative strength increases by successive increments as a function of practice or changes in all- or-none fashion from zero to maximum. The fact that the number of correct responses increases with trials is consistent with both positions. On the incremental assumption each trial adds to the strength of the associations. According to the all-or-none hypothesis, each repetition provides an opportunity for new associations to be established in a single trial. The finding that items within a series are acquired at different rates also presents no basic difficulty for either interpretation. The rate at which habit strength grows may be expected to vary from item to item. Alternatively, the organism's input capacity restricts the number of different associations which can be established on a single trial. Whatever makes for differences in difficulty among items can be expected to influence gradual and all-or-none learning alike.