Transfer from serial to paired-associate learning with two paired-associate rates

A case for response-produced cues in serial learning.

“…If the transfer task necessitates a partial modification of the functional stimuli acquired during serial learning, namely, abandonment of the response-generated component of the stimulus complex, then anything which would facilitate this adjustment for the subject should be reflected in heightened transfer levels. Two variables which have been shown to increase positive transfer in the paradigm are the use of fully instructed subjects and slower second-list presentation rates (Heaps et al, 1968;Postman & Stark, 1967). A full description of the second list, including the fact that pairs occur in random orders across trials, should orient the subject away from reliance upon the self-generated aspect of the cue complex all the sooner, still leaving him with the benefits of the externally provided source.…”

“…With fixed trials, Postman and Stark (1967, instructed subjects, Cycle 1) found 51%, and Shuell and Keppel (1967, conventional serial learning) obtained 58%. By the most sensitive measure, scores on the first trial, these percentages of maximum possible performance were found: 50 (Postman & Stark, 1967;Shuell & Keppel, 1967), 60 (Heaps, Greene, & Cheney, 1968), and 65 (Stark, 1968). Such levels, representing the highest obtained, nevertheless fall disappointingly short of theoretical expectations.…”