Buildup of proactive inhibition (PI) in the Brown-Peterson paradigm is interpreted in terms of the cue-overload principle: The probability of recalling an item declines with the number of items subsumed by its functional retrieval cue. In contrast to a registration interpretation, the cue-overload view predicts that if the effects of initial recall and of differential recency are controlled, performance in a delayed test of all items from successive lists will be independent of their presentation order. This prediction is supported in Experiment 1. A long series of Brown-Peterson trials was presented, with the items in each block of three trials belonging to the same conceptual category and with initial recall tested only occasionally. The final recall of items from initially untested categories was independent of within-category list position. The cue-overload principle gains further support from Experiment 2 which, with a procedure similar to the first experiment, showed that level of final recall varied inversely with the number of lists in the category. The relation of the buildup of PI effect to other memory phenomena is discussed.

No abstract

The modality effect refers to the higher level of recall of the last few items of a list when presentation is auditory as opposed to visual. It is usually attributed to echoic memory. The effect may be sharply reduced by an ostensibly irrelevant auditory item appended to the end of the list. Previous research suggests that this "suffix effect" arises only when the suffix item occurs within 2 sec of the last list item. This finding strengthens the widely held assumption that echoic information decays within 2 sec, and has led to the assumption that if echoic information is to be useful in serial recall it must first be encoded into a more durable modality-independent form. Both assumptions conflict with the research reported here. The first two experiments demonstrate substantial suffix effects with suffix delays of 2 and 4 sec, indicating that echoic information lasts at least 4 sec. This finding implies that echoic information may aid recall directly, an implication that was supported in Experiments 3 and 4. In Experiment 3 serial recall was interrupted with a brief distractor task. The modality effect was smaller when this task was auditory than when it was visual, suggesting that echoic information was still available immediately prior to recency recall. In Experiment 4 list presentation was broken by a 4-sec pause; the modalities of the list halves were combined factorially. Interest focused on the recency positions of the first half. A modality effect was found at these positions when the second half was visual but not when it was auditory. This is contrary to the hypothesis that echoic information is encoded before recall, but is consistent with the hypothesis that echoic information is encoded before recall, but is consistent with the alternative hypothesis that echoic information is used directly at recall. The final two experiments concern the modality effect found when a delay is interpolated between list presentation and recall. Experiment 5 showed that a 20-sec silent copying task interpolated before free recall reduced visual recency more than auditory recency, and so enhanced the modality effect. This suggests that, contrary to prevailing opinion, the modality effect in delayed recall is not the result of a memory that is modality-independent. In Experiment 6 a modality effect found with serial recall after an unfilled interval of 18 sec was unaffected by visual distractor task, but almost eliminated by an auditory distractor task, given just prior to recall. It thus seems that the modality effect in delayed recall is the result of information persisting in echoic form until recall. It is concluded that echoic information can persist for many seconds and is used directly at the time of recall.

Serial position curves for the immediate serial recall of supraspan word lists were investigated as a joint function of input modality and the frequency with which the list words occur in everyday usage. There was a strong effect of frequency at primacy positions but no effect at recency positions. It is concluded from this interaction that, contrary to some views, serial recall is not the product of a simple, unitary process. The modality effect, which is confined to recency positions, was independent of word frequency. The theoretical implications of this finding are discussed, and its incompatibility with one model noted.

Evidence for a representational tactile mernory was obtained from a tactile analogue of the auditory suffix effect. In two experiments, a short sequence of tactile stimuli applied to the fingers was followed by an acoustic (control) or tactile (suffix) recall signal. The serial position curves for the two conditions were cornpared. They revealed similar and strong primacy effects, but recall of the last few stimuli was higher in the control condition, This terminal control advantage Was attributed to the retention of raw tactile inforrnation.