Psychobiology 1991, 19 (1), 1-10 The relationship between faet and souree memory: Findings from amnesie patients and normal subjeets ARTHUR P. SHIMAMURA University of California, Berkeley, California and LARRY R. SQUffiE Veterans Affairs Medical Center, San Diego, California and University of California, San Diego, La Jolla, California Amnesie patients and normal subjects were presented 20 obscure facts to learn, and, after a 2-h delay, were given tests of fact recaIl, fact recognition, and source recall. The subjects were also given a recognition memory test in which they were asked about the learning session itself (event memory test). The amnesie patients exhibited marked impairment on all tests. Impaired source recall (i.e., source amnesia) appeared to be unrelated to impaired fact or event memory. That is, patients who made source errors remembered the facts and the learning session as weIl as did patients who did not make source errors. In a second experiment, normal subjects were tested 1, 3, 6, and 8 weeks after learning. Source recall dropped sharply after a 6- or 8-week reten­ tion interval, demonstrating that source amnesia can occur in normal subjects when they are tested long after learning. For the normal subjects, fact and source memory were correlated. Thus, in normal subjects, source amnesia can reflect the loss of fact memory, event memory, and as­ sociations between them. However, in amnesie patients, source memory impairment was unrelated to the severity of impaired fact or event memory. Source memory impairment may reflect a specific deficit in the association of facts with their contexts. In everyday experience, one often remembers some fac­ of memory-impaired patients. Indeed, it has been argued tual information (e.g., an experimental finding, a news that amnesic patients exhibit a specific impairment of con­ item) but forgets when and where one has encountered textual (Hirst, 1982; Huppert & Piercy, 1976; for review, it. Such instances represent a loss of source or contex­ see Mayes et al., 1985) or episodie memory (Cermak, tual memory. Various theoretical accounts of memory 1984; Kinsbourne & Wood, 1975; for review, see Schacter have distinguished between memory for facts and mem­ & Tulving, 1982). Yet despite these proposals, a number ory tor context (Hirst, 1982; Humphreys, Bain, & Pike, of findings suggest that this kind of distinction cannot ac­ 1989; Jacoby & Dallas, 1981; Schacter, 1987; Tulving, count for the pattern of impairment observed in amnesie 1972, 1983). A common feature ofthese accounts is that patients (Gabrieli, Cohen, & Corkin, 1988; Ostergaard, factual (semantic) and context-specific (episodic) mem­ 1987; Shimamura & Squire, 1987). Thus, in addition to ory are proposed to be distinct and dissociable compo­ their difficulty in learning item and event information en­ nents of memory function (for discussion, see Mayes, countered after the onset of amnesia, amnesie patients are Meudell, & Pickering, 1985; McKoon, Ratcliff, & Dell, also impaired in their ability to acquire new factual in­ 1986; Shimamura, 1990; Shimamura & Squire, 1987; formation (Gabrieli et al., 1988; Shimamura & Squire, Squire, 1987; Tulving, 1986). 1987). Moreover, on tests of premorbid memory, both The relationship between factual and contextual mem­ factual and contextual (i.e., autobiographical) memory are ory has also been considered in neuropsychologieal studies impaired (Albert, Butters, & Levin, 1979; Baddeley & Wilson, 1986; Butters & Cermak, 1986; MacKinnon & Squire, 1989; Shimamura & Squire, 1987; Squire, Haist, & Shimamura, 1989). Thus, it is difficult to argue that This research was supported by the Medical Research Service of the Veterans Administration, NIMH Grant MH24600, the Office of Naval amnesia produces only an impairment of contextual or epi­ Research, the McKnight Foundation, and a Bioscience Grant for Inter­ sodic memory (for review, see Mayes et al., 1985; national Joint Research from the NEDO, Japan. We thank Kim Rivero­ Shimamura, 1989). Frink, 10ni Shutler, and Joyce Zouzounis for their assistance in testing sub­ There is evidence, however, that in some amnesie pa­ jects, and Gus Craik, Frank Haist, Gail Musen, and Stuart Zola-Morgan for helpful comments on earlier drafts. Requests for reprints should be sent tients contextual memory is disproportionately impaired to L. R. Squire, Department ofPsychiatry (V116-A), Veterans Affairs relative to factual memory. For example, memory for Medical Center, 3350 La Jolla Village Dr., San Diego, CA 92161. temporal order information is disproportionately affected 1 Copyright 1991 Psyehonomie Society, Ine. 2 SHIMAMURA AND SQUIRE in patients with Korsakoff's syndrome but not in other order memory. However, several additional questions amnesic patients (Huppert & Piercy, 1976; Meudell, about source amnesia remain. First, how can an amnesic Mayes, Ostergaard, & Piekering, 1985; Shimamura, patient remember a fact that was presented during a learn­ Janowsky, & Squire, 1990; Squire, 1982; Squire, Nadel, ing session but not remember the session itself? One & Slater, 1981). The severity of impairment in temporal possibility is that the source itself has actually been for­ order memory can differ between two groups of amnesic gotten; alternatively, the association between acquired in­ patients despite the fact that the two groups are equiva­ formation and the leaming event may have been lost. One lently impaired in recall and recognition memory of facts way to address this issue is to determine directly how well or items (Shimamura et al., 1990; Squire, 1982). More­ amnesic patients who commit source errors actually re­ over, amnesic patients can exhibit a greater impairment member the learning event (e.g., recognizing aspects of in temporal order memory than control subjects do, even the study session: the kind of material, the format of when the control subjects are tested after a long reten­ presentation, the type of response). Second, what struc­ tion interval in order to match their fact or item mem­ tures and functions must be damaged to produce source ory scores to those of the amnesic patients (Meudell amnesia? Can frontal lobe dysfunction explain which pa­ et al., 1985). tients exhibit source errors and which do not? Third, what Disproportionate deficits in temporal order memory ap­ is the relationship between source memory and fact mem­ pear to be related to frontal lobe pathology. First, impaired ory in normal subjects? Normal subjects tend not to ex­ temporal order memory was observed in patients with hibit source errors when the retention interval is 1 week frontal lobe lesions, even though fact and event memory or less (Schacter et al. , 1984; Shimamura & Squire, were preserved (Milner, 1971; Shimamura et al., 1990). 1987). Do they exhibit source errors at longer intervals? Second, impaired temporal order memory in patients with If so, are fact memory and source memory correlated, Korsakoff' s syndrome was correlated with impairment on or are they unrelated? neuropsychological tests sensitive to frontal lobe damage In two experiments, we assessed the relationship be­ (Squire, 1982). Third, recent quantitative analysis of com­ tween fact memory and source memory. In Experiment 1, puted tomography (CT) scans in patients with Korsakoff's amnesie patients and control subjects were presented facts syndrome revealed frontal lobe atrophy in addition to to learn (e.g., "The official flower of British Columbia midline diencephalic damage (Shimamura, Jernigan, & is the dogwood"), and after a 2-h retention interval they Squire, 1988; see also Jacobson & Lishman, 1987; Kopel­ were asked to recall both the facts and the last time that man, 1989). they had encountered the facts (i.e., source recall). In ad­ Another example of a contextual or episodic memory dition, subjects took a recognition memory test for the impairment that seems to occur out of proportion to im­ facts as weH as a recognition memory test for the learn­ paired factual memory is the phenomenon of source amne­ ing event (e.g., "Was the material presented on sheets sia. Some amnesie patients, after successfuHy recalling of paper or index cards?"). In Experiment 2, we evalu­ recently presented facts, fail to identify the learning ses­ ated forgetting of fact memory, source memory, and event sion as the source of the facts (Schacter, Harbluk, & memory in normal subjects across an 8-week period. Sub­ McLachlan, 1984; Shimamura & Squire, 1987). Source jects were given the same study-test procedure that was memory impairment is also prominent in older adults, used in Experiment 1, except that the retention interval which suggests that aging affects contextual memory more was 1, 3, 6, or 8 weeks. In this way, we deterrnined the than fact memory (Janowsky, Shimamura, & Squire, relationship between fact and source memory at several 1989; McIntyre & Craik, 1987). points during forgetting. Interestingly, patients who exhibit source amnesia for facts and those who do not can be equivalently impaired EXPERIMENT 1 on recall and recognition tests for the facts themselves (Shimamura & Squire, 1987). Also, control subjects tested Method at a longer (l-week) retention interval in order to match Subjects their recall and recognition memory for facts to that of the Amnesie patients. We tested 12 amnesie patients: 7 patients with alcoholie Korsakoff's syndrome, 3 with amnesia due to an episode amnesic patients do not exhibit source amnesia (Schacter of anoxia or isehemia, 1 with amnesia due to a bilateral