David G. Payne

Hypermnesia: the role of repeated testing

The present study was designed to determine whether the increased recall of pictures across repeated tests (hypermnesia) is due to increasing strength of imaginal traces during the retention interval or to increased retrieval practice from prior tests. Subjects studied 60 pictures and then recalled them after various delays that were filled with instructions and, in two cases, reading a passage. Recall on a first test showed no change with retention interval. With retention interval held constant, however, the number of pictures recalled varied directly with the number of prior tests subjects had been given. This finding points up the critical nature of retrieval factors in producing hypermnesia.

Recall criterion does not affect recall level or hypermnesia: A puzzle for generate/recognize theories

Subjects studied a mixed list of 70 words that varied in imagery value and then received three successive tests. Also varied were instructions given to subjects prior to list presentation (imagery or semantic rehearsal) and the type of recall test (standard free recall, an uninhibited-recall procedure in which subjects were encouraged to free associate and to guess while recalling the list, and a forced-recall condition in which they were also told to write a large number of responses to fill the allotted spaces). Recall improved across the three tests in all conditions, but the improvement was greater for high-than for low-imagery words. In addition, hypermnesia (the improved recall across tests) was shown to occur following semantic rehearsal instructions as well as imagery instructions and to occur with low-imagery words, contrary to the imagery hypothesis of the effect. Most importantly, the large variation in recall criterion produced by manipulating instructions at test (as measured by intrusions) did not affect the overall level of correct recall or the magnitude of improvements across tests. Apparently, the assumption of generate-recognize theories that people generate much more information in free recall than they produce (due to a stringent criterion for recognition of the generated material) is false.

Assessing the effectiveness of warnings and the phenomenological characteristics of false memories

The phenomenology of false memories was investigated in three experiments in which participants heard two experimenters read lists of items that were related to critical nonpresented items. In Experiments 1, following a recognition memory test, participants rated the phenomenological characteristics of their memories immediately and after a 48-hour delay. False recognition was prevalent and on several dimensions participants rated their true memories as more vivid than their false memories. In Experiments 2 and 3, following the study phase, participants were warned about the phenomenological differences between true and false memories and were instructed to use this information to avoid reporting nonpresented items. This type of warning was ineffective at reducing false recall (Experiment 2) and false recognition (Experiment 3) relative to unwarned participants. Importantly, the inability of explicit warnings to impact illusory recollections demonstrates that the false memories cannot be attributed simply to a criterion shift.

Hypermnesia as determined by level of recall

The results of three experiments provide converging evidence for the conclusions that hypermnesia (increased recall with repeated testing) does not depend on encoding of material in an imaginal format, but is related to the level of recall across conditions within an experiment. In Experiment 1 subjects performed orthographic, phonological, or semantic operations on words and then recalled them on three successive free recall tests. Orienting tasks affected the level of recall (semantic > phonological > orthographic), and the level of recall was correlated with hypermnesia. In Experiment 2 subjects studied nonsense syllables presented either once or three times and were then given three tests. Recall improved across tests, and the improvement was reliably greater for items studied three times. In both Experiments 1 and 2 subjects who received three tests recalled no more total items than did subjects given a single long test of equivalent duration. In Experiment 3 subjects repeatedly recalled instances of categories from semantic memory. Hypermnesia was observed and was again related to level of recall. The results help delineate the necessary and sufficient conditions for observing hypermnesia on repeated tests, and are in general agreement with the account of the phenomenon provided by J. G. W. Raaijmakers and R. M. Shiffrins (In The psychology of learning and motivation: Advances in research and theory . New York: Academic Press, 1980, Vol. 14.) SAM model.

Source attributions and false memories: a test of the demand characteristics account.

When lists of related words are presented to subjects, they sometimes recall or recognize nonpresented words related to those lists (critical lures). In fact, subjects sometimes claim to remember which of two speakers said the critical lures. We examined whether this finding could be accounted for by demand characteristics. If subjects’ willingness to make source attributions to critical lures reflects experimental demand, one would predict that subjects should be willing to change and should have little confidence in these attributions. Subjects made more attributions, were less likely to change their attributions, and were more confident in their attributions for critical lures than for unrelated distractors. Subjects had even more confidence in the attributions that they made for words that had actually been presented, and they were even less likely to change these attributions. These findings suggest that false memories are quite compelling but that they are also subtly different from true memories.

A direct comparison of recognition failure rates for recallable names in episodic and semanticmemory tests

In an extension of Muter’s (1978) research, subjects studied pairs of lowercase cues and uppercase targets consisting of famous names (e.g., betsy ROSS), nonfamous names (e.g., edwin CONWAY), weakly related words (e.g., grasp BABY), and unrelated words (e.g., art GO). Following recognition tests in which surname and word targets were tested in the absence of their cues, cued recall tests for the surname and word targets were given. In semantic recognition and recall tests, the response to a surname was to be made solely on the basis of its fame, regardless of whether or not it had appeared in the study list. In episodic memory tests, the response to a surname was to be made solely on the basis of whether or not it had appeared in the study list, regardless of its fame. In all tests, the response to a nonname was to be made solely on the basis of whether or not it had appeared in the study list. The Tulving-Wiseman (1975) function accurately predicted recognition failure rates for famous surnames, whether or not they were from the study list and whether the test was episodic or semantic, and for targets from the weakly relatedword pairs. However, recognition failure rates were lower than the Tulving-Wiseman function predicted for nonfamous surnames in the episodic memory test and for targets from unrelated word pairs. Discussion focused on these results’ implications for the nature of the Tulving-Wiseman function and the psychological reality of the episodic-semantic memory distinction.

Effective warnings in the Deese-Roediger-McDermott false-memory paradigm: The role of identifiability

These experiments document that warnings can substantially reduce false memories in the Deese-Roediger-McDermott (DRM) paradigm when the critical items are easily identifiable. Participants in a norming study identified the critical item after hearing a list of words. The lists with critical items that could be identified by the largest proportion of participants (high identifiable [HI] lists) and the smallest proportion of participants (low identifiable [LI] lists) were used in the experiment. Participants heard lists of words (e.g., bed, rest, doze) related to a critical item (e.g., sleep) and were warned about the nature of the lists before the study phase. The results indicated that warnings reduced false recognition of critical items for HI lists but not LI lists.

Comprehension and retention of nonlinear text : Considerations of working memory and material-appropriate processing

The specific processes that support comprehension and retention of hypertext information and the manner in which those processes differ from those involved in the comprehension and retention of traditional linear text were examined in 4 experiments. Experiments 1 and 2 demonstrate that although processing hypertext is not more demanding overall than processing linear text, processing hypertext does make more demands for use of relational information. Experiments 3 and 4 provide tests of the operational assumptions and methods used in the initial two experiments and provide converging evidence to support the notion that hypertext requires a different balance of processing resources than does linear text. Results and directions for research with new information technologies are discussed using the constructs of a limited-capacity working memory and material-appropriate processing.

The generation effect: Further tests of the lexical activation hypothesis

Three experiments compared recognition memory for word versus nonword responses when they had been either read or generated using a rhyme rule and either a word or nonword stimulus. That is, either the wordshop or the nonwordthop was generated from either the wordchop or the nonwordphop. In Experiment 1, the lexicality of the stimulus and response terms was manipulated orthogonally between subjects; in Experiments 2 and 3, it was manipulated orthogonally within subjects. In Experiment 3, subjects also made a lexical (word-nonword) decision about each response term after it had been read or generated. In all three experiments, memory performance was better for generated than read responses. This generation effect occurred only if the response term was a word, regardless of whether the stimulus term was a word or a nonword. These results are discussed in terms of the roles that lexical memory and response unitization play in the generation effect.

Recognition performance level and the magnitude of the misinformation effect in eyewitness memory

The issue of whether misleading postevent information affects performance on the modified recognition test introduced by McCloskey and Zaragoza (1985) was examined in a meta-analysis. Results indicated that a misinformation effect can be obtained with the modified test. The meta-analysis also revealed that recognition hit rates are higher in studies that yield a misinformation effect than in studies in which the misinformation effect is not significant. The data from the meta-analysis were also used to assess whether the misinformation effect is related to the length of the retention interval. Results showed that a misinformation effect is more likely to be obtained with long retention intervals, although in the available data there is a confound between the length of the retention interval and the recognition level obtained.