In the previous discussion of mediated and non-mediated generalization (i), it was suggested that the former type of generalization might occur along a number of different gradients in addition to those involving synonyms. One of the possible other gradients suggested was that involving antonyms. The present paper reports an experiment designed to determine experimentally whether generalization does occur to antonyms.In the previous paper, we described the mechanism which, it is assumed, mediates generalization in the case of synonymous words (i, Fig. i; see also 3, pp. 27-28). In the case of antonyms it is possible that the mediating mechanism is somewhat more complex than is the case with synonyms. To explore this theoretical possibility would require a great deal of discussion, and we merely mention the point here, reserving until a later time the examination of this question.A previous study by Riess (4) showed that generalization to antonyms does occur. He used the galvanic skin reaction to a bell as the response to be conditioned. The present experiment employed a larger number of antonyms than Riess used, and the method was essentially the same as that used in a previous investigation of homophone and synonym generalization (2). METHOD 1. Materials.-A list of 10 unrelated familiar words (presentation or P-list) was assembled such that from each word an antonym could be derived. Other complicating semantic or homophonic relationships were minimized. According to our hypothesis, the repetition (reinforcement) of these antonyms should increase the reaction potentials of the words in the original (P) list more than would the repetition of an unrelated list (control), and hence recall of the P-list should reflect this difference. Two reinforcement lists were accordingly constructed: (1) the A-R list was composed of xo words each one of which was an antonym of one word in the P-list; (2) the C-Ri list was composed of 10 words which were not related to any words in the P-list in any evident or formal way, directly or indirectly. These two lists, along with the

No abstract

No abstract

No abstract