The basic-level convergence effect in memory distortions

Abstract— Whereas most previous findings suggest that memory may become more abstract over time, so that memory for gist outlasts verbatim memory, there are findings suggesting that abstract information may sometimes be instantiated in more specific terms. In this study, we examined the hypothesis that retained information tends to converge at an intermediate level of abstractness—the basic level. In two experiments, we found bidirectional, symmetrical shifts in the memory for story material: Participants presented with either subordinate terms (e.g., sports car) or superordinate terms (e.g., vehicle) tended to falsely report basic-level terms (e.g., car) instead. This pattern emerged for both recall and recognition memory tests, at both immediate and delayed testing, and under free and forced reporting. The results suggest that the basic level, which has been considered cognitively optimal for perception, categorization, and communication, is also the preferred level for retaining episodic information in memory.     

Interactions between spontaneous instantiations to the basic level and post-event suggestions

Extensive research shows that post-event suggestions can distort the memory for a target event. In this study we examined the effect of such suggestions as they interact with the products of a spontaneous memory process: instantiation of abstract information to an intermediate level of abstractness, the basic level (Pansky & Koriat, 2004 Pansky, A. and Koriat, A. 2004. The basic-level convergence effect in memory distortions. Psychological Science, 15: 52–59. [Crossref] , [Google Scholar]). Participants read a narrative containing items presented at the superordinate level (e.g., FRUIT), were exposed to suggestions that referred to these items at the basic level (e.g., APPLE), and were finally asked to recall the original items. We found that the tendency to instantiate spontaneously in the control (non-misleading) condition, particularly over time, increased following exposure to suggestions that were likely to coincide with those instantiations. Exposure to such suggestions, either immediately or following a 24-hour delay, reduced subsequent correct recall of the original items only if the suggested information coincided with the information one tends to instantiate spontaneously in a given context. Suggestibility, in this case, was particularly pronounced and phenomenologically compelling in terms of remember/know judgements. The findings are taken to imply that effects of post-event suggestions can be understood in terms of the constructive processes that set the stage for their occurrence.     

The perception of number from the separability of the stimulus: The Stroop effect revisited

The literature on numerical perception is reviewed from the standpoint of research on selective attention, and predictions are made concerning the dimensional interaction between physical and numerical size of numerals. We manipulated stimulus differences to make the classification of numerical value slightly better (Experiment 1), substantially better (Experiment 2), or worse (Experiments 3–4) than classification of physical size. Garner, Stroop, and redundancy effects were used to gauge the degree of interactive processing. For nearly matched discriminability, both number and size appeared separable when the dimensions were varied orthogonally, but showed Stroop interference and redundancy gain when the dimensions were varied in a correlative fashion. When mismatched, asymmetric Garner and Stroop effects emerged in orthogonal contexts along with Stroop and redundancy gains in correlative contexts. These findings define a unique relation: Numerical value and physical size were optionally separable dimensions. We conclude that a magnitude representation is not mandatory for the perception of numerals. Our conclusions offer a new perspective for understanding both numerical perception and the Stroop phenomenon itself.     

Inoculating against eyewitness suggestibility via interpolated verbatim vs. gist testing

In real-life situations, eyewitnesses often have control over the level of generality in which they choose to report event information. In the present study, we adopted an early-intervention approach to investigate to what extent eyewitness memory may be inoculated against suggestibility, following two different levels of interpolated reporting: verbatim and gist. After viewing a target event, participants responded to interpolated questions that required reporting of target details at either the verbatim or the gist level. After 48 hr, both groups of participants were misled about half of the target details and were finally tested for verbatim memory of all the details. The findings were consistent with our predictions: Whereas verbatim testing was successful in completely inoculating against suggestibility, gist testing did not reduce it whatsoever. These findings are particularly interesting in light of the comparable testing effects found for these two modes of interpolated testing.     

Enhancing the quantity and accuracy of eyewitness memory via initial memory testing

Previous studies have demonstrated the benefits of initial memory testing in terms of “inoculating” eyewitness memory against forgetting. The aim of the present study was to determine to what extent and under which conditions such testing may also enhance the accuracy of subsequent retrieval. Two aspects of interpolated testing were manipulated: the mode of interpolated testing (forced verbatim vs. free level) and its timing (immediate vs. delayed). After witnessing a target event, participants were questioned about event details either immediately or after a 48-h delay, and were either required to respond at the verbatim level or were given control over the grain size of their responses. Verbatim memory for event details was finally tested 72 h after the event under both standard forced-report conditions and free-report conditions. Immediate interpolated testing was found to improve both memory quantity and memory accuracy on the final test, whereas delayed interpolated testing improved only memory quantity (and to a lesser extent). Although the mode of interpolated testing affected performance on the initial test, it had no effect on either memory quantity or memory accuracy on the final test. Practical implications with regard to eyewitness questioning are discussed.     

Influence of electroencephalography neurofeedback training on episodic memory: A randomized,sham-controlled,double-blind study

The relationships between memory processes and oscillatory electroencephalography (EEG) are well established. Neurofeedback training (NFT) may cause participants to better regulate their brain EEG oscillations. The present study is a double-blind sham-controlled design investigating the effect of NFT on memory. NFT included up-training upper alpha (UA) band, up-training sensory-motor rhythm (SMR) band and sham protocol. Thirty healthy adult volunteers were randomly divided into three treatment groups. NFT sessions (30 min each) took place twice weekly for a total of 10 sessions while memory testing took place pre- and post-training. The results indicate dissociation between SMR and UA NFT and different memory processes. While the SMR protocol resulted in improving automatic, item-specific and familiarity-based processes in memory, the UA protocol resulted in improved strategic and controlled recollection. The implications of the results are discussed.     

Interactions between spontaneous instantiations to the basic level and post-event suggestions

Extensive research shows that post-event suggestions can distort the memory for a target event. In this study we examined the effect of such suggestions as they interact with the products of a spontaneous memory process: instantiation of abstract information to an intermediate level of abstractness, the basic level (Pansky & Koriat, 2004 ). Participants read a narrative containing items presented at the superordinate level (e.g., FRUIT), were exposed to suggestions that referred to these items at the basic level (e.g., APPLE), and were finally asked to recall the original items. We found that the tendency to instantiate spontaneously in the control (non-misleading) condition, particularly over time, increased following exposure to suggestions that were likely to coincide with those instantiations. Exposure to such suggestions, either immediately or following a 24-hour delay, reduced subsequent correct recall of the original items only if the suggested information coincided with the information one tends to instantiate spontaneously in a given context. Suggestibility, in this case, was particularly pronounced and phenomenologically compelling in terms of remember/know judgements. The findings are taken to imply that effects of post-event suggestions can be understood in terms of the constructive processes that set the stage for their occurrence.     

Comparative judgment of numerosity and numerical magnitude: attention preempts automaticity

It is commonly believed that humans are unable to ignore the meanings of numerical symbols, even when these meanings are irrelevant to the task at hand. In 5 experiments, the authors tested the notion of automatic activation of numerical magnitude by asking participants to compare, while timed, pairs of numerical arrays on either numerosity or numerical value. Garner and Stroop effects were used to gauge the degree of interactive processing. The results showed that both effects were sensitive to the discriminability of values along the constituent dimensions, to the number of stimulus values used, and to practice and motivation. Notably, Stroop and Garner effects were eliminated under several conditions. These findings are incompatible with claims of obligatory activation of meaning in numerical processing, and they cast doubt on theories positing automatic processing of semantic information for alphanumerical symbols.