There has been unprecedented interest in recent years in questions pertaining to accuracy and distortion in memory. This interest, catalyzed in part by real-life problems, marks a significant departure from the quantity-oriented approach that has characterized much of traditional memory research. We outline a correspondence metaphor of memory underlying accuracy-oriented research, and show how the features of this metaphor are manifested across the disparate bodies of research reviewed here. These include work in the Gestalt tradition, spatial memory, memory for gist, schema theory, source monitoring, fluency misattributions, false recall and recognition, postevent misinformation, false memories, eyewitness research, and autobiographical memory. In examining the dynamics of memory accuracy, we highlight the importance of metacognitive monitoring and control processes. We end by discussing some of the methodological, theoretical, and metatheoretical issues inherent in accuracy-oriented research, attempting to prepare the groundwork for a more coherent psychology of memory accuracy.

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.Seldom do we appreciate the paramount importance of numbers in our cognitive milieu. A number stands for our height, another for our IQ, a third indicates the balance in our checking account, a fourth our telephoneand the list is virtually inexhaustible. The ease with which we use numbers masks the fact that complex cognitive processes are required for the mere recognition of numerical stimuli or the execution of the simplest numerical comparisons and calculations. Tapping those processes poses a challenge for students of numerical cognition. In the present study we propose to use selective attention as a tool to uncover the nature of numerical perception. Simultaneously, we will use the results to reflect on models of selective attention itself. The classic measure of selective attention (indeed, its failure), of course, is that based on the seminal work of Stroop (1935), and Strooplike effects have been reported for numerical dimensions. However, recent research has shown Stroop interference to be malleable, its presence depending on several heretofore neglected experimental factors. Therefore, despite a substantial amount of empirical study, we still lack a wide theoretical understanding of the Stroop effect, and we have notoriously little insight into processes of'numerical cognition.This research was supported by Israel Science Foundation Grant 58-93 to D.A. We thank Helena Kadlec, Robert Melara, Michael Masson, and Lawrence Marks for many insightful comments that improved the manuscript. Bob Melara inspired D.A.'s interest in selective attention when they both enjoyed the hospitality and scholarship of Larry Marks in the Pierce Laboratory at Yale University. We appreciate their continuing interest and support. Thanks are also due to Yoav Arieh, Joseph Glickson, and Lilach Shalev for detailed ...

In 7 experiments, participants selected the larger member of pairs of digits that differed in numerical magnitude as well as in physical size. Selective attention to the relevant dimension (number or size) was gauged by Garner and Stroop interference, both of which varied considerably as a function of the number and relative discriminability of values along the constituent dimensions. When the to-be-ignored dimension was more discriminable, sizable Garner and Stroop effects affected performance on the relevant dimension. When it was less discriminable, Garner and Stroop effects were considerably smaller regardless of whether the relevant dimension was numerical or physical size. The sensitivity of Stroop interference to manipulations of discriminability is accounted for by the allocation of attention to the constituent dimensions. The demonstrated malleability of the Stroop effect is incompatible with claims of strong automaticity in numerical processing.

No abstract

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.