For more than two thousand years men have concerned themselves with the question of how ideas become associated or connected to one another. The concern here is primarily with the importance of the temporal order of associating independent elements with the common element and the direction of association between the independent elements and the common element in the verbal transfer situation. Several theories have been proposed which deal with the nature of mediated generalization, and though they differ somewhat as to the exact nature of the mediating process. From the standpoint of the subjects, this experiment appeared to involve a series of paired-associate learning problems. The subjects were required to learn two lists of words to a specified criterion, and were then tested on a third list for a fixed number of trials. Each list consisted of eight pairs of words, and the lists were constructed in such a way that the third list was composed of lour pairs which had a common associate in the first two lists and four pairs which had no such connection. The results of this investigation clearly establish the presence of generalization effects for all paradigms except one of the chaining models. In conclusion, this study has clearly established the presence of generalization effects with three chaining, two acquired stimulus equivalence, and two required response equivalence paradigms, and has strongly suggested the presence of several associative factors and indicated a need for further investigation of others. In addition, individual difference factors, such as task set and strategy, appear as important parameters in generalization and warrant further investigation.

This study investigated the effect of various instructional sets (speed, relaxed time instructions, and "set popular" instructions) on commonality of word associations. Findings were generally supportive of the hypotheses: significantly more common associations were given under speed as compared to relaxed instructions, Ss were able to significantly increase their commonality scores under a set to give popular responses, and need for social approval was differentially related to commonality under relaxed but not under speed instructions. Contrary to a suggestion of Jenkins, social sensitivity (as reflected in the need for approval measure) was not related to the ability to increase one's commonality score under "set popular" instructions.

No abstract

The influence of appropriate contextual knowledge on the encoding of sentences was examined. Subjects read and were later asked to recognize sentences from Bransford and Johnson's (1972) passages. Subjects were informed of the passage title prior to encoding, prior to testing, or not at all. In Experiment 1, title availability during encoding significantly increased both recognition of target sentences and rejection of distractor sentences. Providing the title immediately prior to testing did not improve performance over the no-title condition. In a second experiment, performance in a title-before condition also exceeded performance in a no-title condition, even though distractor sentences were not included in the recognition test. Thus it appears that awareness of context during encoding enables subjects to engage in unique encoding processes which benefit recognition as well as recall memory.Research in human memory over the past 15 years has repeatedly demonstrated that the presence of appropriate contextual information during encoding facilitates sentence memory (Bransford & Johnson, 1972;Dooling & Lachman, 1971;Schustack & Anderson, 1979). The specific role that contextual information plays in aiding memory is not universally agreed on however. One widely held view is that prior contextual knowledge enables subjects to perform encoding operations which would not otherwise be carried out (Bransford & Johnson, 1972). These operations are assumed to lead to a unique and elaborated encoding of the test material which allows it to be more easily retrieved. Alba, Alexander, Hasher, and Caniglia (1981) offer an alternative to this "encoding" explanation. These authors point out that contextual information may aid recall performance simply by providing a retrieval cue. During retrieval contextual information may serve to link one sentence to the next, thereby facilitating recall without playing a major role in determining what has been encoded from each individual sentence.We wish to acknowledge Jani Gabriel Byrne for her help in constructing distractor sentences for Experiment I.Computing services were provided by the Computer Science Center of the University of Maryland.

No abstract