Sammdsy.-Difficulry of learning a list of 12 CVC pairs increased monotonically as a function of number of list items ( 1 , 3, 6, and 1 2 ) grouped together for simultaneous exposure. Contrary to expectation, the difficulty was not affected by locus of grouping, i.e., whether grouping occurred on the study part, test part, or on both puts of each trial. Subsequent free recall also failed to show significant differences between conditions. The over-all grouping decrement was interpreted in terms of a strengthening of multiple associations among list members.Previous research by Brown and Brown (1965) showed the commonly employed single or successive method of pair presentation to be vastly superior to a simzdtaneous method, whe~ein subsets of two or four pairs are presented together for learning. The superiozity was invariant with respect to size of subset, formal intralist similarity, and whether the pairs appeared always in the same or in different subsets on each trial.As suggested in the previous study, the inferiority of the simultaneous method may have derived from 3 . procedural difference on the study (St) and test (T) parts of each learning trial. Specifically, grouping (of pairs) for simultaneous exposure occurred only on the St part of the trial: the sr~mulus (S) terms were presented individually on &e T part of the trial. Such a procedural change, by eliminating the pair-grouping context of St, also may reduce the number of available S cues on T and thereby lead to slower learning (McGeoch & Irion, 1952, pp. 448-45 1 ) .T o obtain more general information on the, effects of locus and size of grouping, the present study grouped varying numbers of list items together on St, T, and on both St and T. It was hypothesized that use of the same pair context during St and T woulc reduce interference from pairs contained in different subsets and perhaps from ?airs contained within the same subset. Thus a relative facilitation of learning was predicted for grouping on both St and T as compared to either locils alone.
METHODThe 160 paid college students learned an identical 12-pair list of CVCs of 75 to 85% Archer (1960) association value to a criterion of one errorless trial.The list was exactly the low formal similarity list used in the earlier sn~dy and described in detail elsewhere (Battig & Brackett, 1963).