View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by Elsevier - Publisher Connector Neuron, Vol. 44, 149–160, September 30, 2004, Copyright 2004 by Cell Press The Cognitive Neuroscience Review of Memory Distortion Daniel L. Schacter* and Scott D. Slotnick distortions: misattribution, suggestibility, and bias. Mis- Department of Psychology attribution occurs when retrieved information is as- Harvard University signed to the wrong source (e.g., mistaking a previously Cambridge, Massachusetts 02138 imagined event for a real one); suggestibility refers to the incorporation of inaccurate information from external sources, such as misleading questions, into one’s own Memory distortion occurs in the laboratory and in ev- memories; and bias involves the distorting influences of eryday life. This article focuses on false recognition, present knowledge, beliefs, and feelings on recollection a common type of memory distortion in which individu- of previous experience. als incorrectly claim to have encountered a novel ob- Cognitive psychologists have long been interested in ject or event. By considering evidence from neuropsy- each of the distortion-related sins and have produced chology, neuroimaging, and electrophysiology, we much research concerning their properties and impli- address three questions. (1) Are there patterns of neu- cations (for reviews, see Johnson et al., 1993; Koriat ral activity that can distinguish between true and false and Goldsmith, 1996; Roediger and McDermott, 2000; recognition? (2) Which brain regions contribute to false Schacter, 2001; Schacter et al., 1998a). Such distortions recognition? (3) Which brain regions play a role in mon- have interested psychologists because they can provide itoring or reducing false recognition? Neuroimaging insight into the constructive nature of memory, revealing and electrophysiological studies suggest that sensory how bits of information are patched together to form activity is greater for true recognition compared to memories with varying degrees of accuracy. Cognitive false recognition. Neuropsychological and neuro- neuroscientists have, until recently, shown little interest imaging results indicate that the hippocampus and in examining memory distortions. During the past de- several cortical regions contribute to false recogni- cade, however, there has been steadily increasing re- tion. Evidence from neuropsychology, neuroimaging, search on the topic, and a cognitive neuroscience of and electrophysiology implicates the prefrontal cortex memory distortion is beginning to emerge. Much of this in retrieval monitoring that can limit the rate of false work has focused on the misattribution error called false recognition. recognition, which occurs when subjects incorrectly claim that a novel item has been encountered earlier in Introduction an experiment. False recognition is typically inferred The scientific analysis of memory would not be possible from “old” responses to novel items that are conceptu- if memory were perfect. Experimental and theoretical ally or perceptually related to previously studied items— analyses of memory typically depend on variations in that is, false alarms above and beyond the “baseline” the proportions of items or responses recalled, recog- level of false alarms to unrelated novel items (cf. Roe- nized, or otherwise produced across different condi- diger and McDermott, 1995; Underwood, 1965). tions. Such variations, which provide the basis for draw- Cognitive neuroscience research during the past de- ing inferences about the processes that support memory cade has used three different approaches to analyze function, necessarily involve some degree of forgetting. false recognition and related forms of memory distor- tion: neuroimaging, electrophysiology, and neuropsy- Beyond this ubiquitous sense in which imperfect reten- chology. Several studies have used functional neuro- tion is necessary to study memory, observations con- imaging techniques—positron emission tomography (PET) cerning the nature of forgetting have been conceptually and functional magnetic resonance imaging (fMRI)—in central to the cognitive neuroscience of memory. For an attempt to identify specific brain regions associated example, observations of pathological forgetting after with true and false recognition. Electrophysiological ac- damage to the medial temporal lobe (MTL) in human tivity associated with true and false recognition has been amnesic patients have greatly influenced our current examined using event-related potentials (ERPs), which understanding of the role of the MTL in various aspects measure changes in voltage topography on the scalp of memory (Squire and Schacter, 2002). over time. Neuropsychological analyses have focused Memory’s imperfections, however, are not restricted on brain-damaged patients with lesions to MTL and pre- to forgetting. Distortions and illusions of memory also frontal regions that have been implicated previously in occur—that is, there are instances in which some form accurate remembering. of memory is present, but incorrect. Memory distortions We consider evidence from each of these approaches can be considered in the context of the proposal that in an attempt to address three main issues that have memory’s imperfections can be divided into seven basic emerged in the literature during the past decade. First, categories, or “sins.” (Schacter, 1999, 2001). Three of we examine the evidence that brain activity can distin- the sins involve different types of forgetting (called tran- guish between true and false memories. From the sub- sience, absent-mindedness, and blocking), and one re- jective standpoint of an individual engaged in the act of fers to intrusive, unwanted memories of arousing or trau- remembering, true and false recognition are associated matic events (persistence). Most relevant to the topic with an identical mnemonic outcome: in both cases, of the present paper, three of the sins concern memory people believe they are remembering an event that oc- curred in their past. Therefore, it has been of consider- *Correspondence: [email protected] able interest to determine whether, and in what sense, Neuron 150 brain activity differs during true and false recognition. Neuroimaging of True and False Recognition We focus in particular on the hypothesis, tested by sev- In the first neuroimaging study to compare true and false eral investigators, that true memory, more than false recognition, PET scans were carried out while subjects memory, is preferentially associated with the reactiva- performed a recognition test that followed study of vari- tion of sensory/perceptual processing that occurred ous DRM associate lists (Schacter et al., 1996a). After during encoding. Second, we examine evidence that auditory presentation of lists containing 20 words that addresses the issue of what kind of brain activity is were associates of nonpresented theme words, subjects associated with, or responsible for, the generation of were scanned during separate test blocks in which they false memories. A number of studies have been particu- responded separately to true targets (words that had larly concerned with the role of MTL structures in the been studied previously), false targets (nonstudied se- generation of false memories, but recent research has mantic associates of previously studied items), or target also begun to examine the role of other regions as well. controls (nonpresented words that were unrelated to Third, we will consider research that has examined what previously studied words). Compared to a common brain regions are involved in monitoring or reducing sus- baseline fixation condition, both true and false recogni- ceptibility to false memories. Much of this research has tion were associated with blood flow increases in vari- focused on the analysis of regions within prefrontal ous regions that are commonly activated by episodic cortex. retrieval tasks, including dorsolateral/anterior prefron- tal, medial parietal, and medial temporal regions. Direct Distinguishing between True and False Memories: comparisons indicated greater activation during true The Sensory Reactivation Hypothesis than false recognition in a left temporoparietal region Although subjective responses are comparable during previously associated with auditory processing and re- true and false recognition, cognitive studies have pro- tention. Because subjects had heard true targets—but vided some evidence for differences in the nature of the not false targets—during the auditory study phase of representations that support the two forms of memory. the experiment, this finding supports the possibility that In particular, several behavioral studies indicate that true recognition is preferentially associated with an audi- true memories are accompanied by retrieval of greater tory sensory signature that reflects memory for auditory/ sensory/perceptual details than are false memories phonological aspects of previously studied words. (e.g., Mather et al., 1997; Norman and Schacter, 1997; In a related study, Schacter et al. (1997a) performed a similar experiment using the then newly developed Schooler et al., 1986). Such findings make intuitive sense event-related fMRI, which allows stimuli from different because true memories engage perceptual encoding conditions to be mixed together (as opposed to pre- processes that are presumably not involved in the cre- sented in separate test blocks according to condition, ation of false memories.