The retention of the acquisition pairs of a mediation and a control paradigm was studied by introducing free-recall or recognition tests before and after the test trial. The 2nd-stage pairs were frequently recalled before the test trial for both paradigms. The Ist-stage pairs, recalled with a lower frequency, were remembered more often for the mediation than for the control paradigm. Retention of the 1st-and 2nd-stage pairs dropped appreciably following the test trial whereas the recall of the test-trial stimulus and the critical response remained high for both paradigms. The B term or mediator was recalled more often for the mediation paradigm when evidence of mediated facilitation was obtained than for the other conditions.The performance on the third or test curs because the repeated term of the stage of a mediational paradigm such mediation paradigm-the mediator, B as B-C, B-A, A-C, frequently has been -chains together the two unique terms superior to performance on the test during either the second or the third stage of a control paradigm, e.g., X-C, stage (see Horton & Kjeldergaard, B-A, A-C. A number of factors 1961; Jenkins, 1963). Horton and might produce the difference in test-Kjeldergaard suggested that with the trial performance and the present ex-B-C, B-A, A-C paradigm, presentation periments investigated some of these of B as the stimulus for the secondfactors, stage pair might elicit C, the response Some mediational theorists have as-learned to B during the first stage, sumed that the facilitation of the test The 5 might actually learn B-C-A durstage of the mediation paradigm oc-j ng tne second stage (the C term may , TH.. , . , . o ... be implicit or explicit). The subse-1 This research was supported by Public v . , \ "' .. , Health Service Research Grant MH 07944