《认知神经科学》课程教学资源（参考文献）[Squire, L. R.（1992）] Memory and the hippocampus_a synthesis from findings with rats, monkeys, and humans

MEMORY AND THE HIPPOCAMPUS 205 ind othe al that is declare the and with the terms relational and configural.which come from work with rats. n In the Recall.Recognition.and the Feeling of Familiarity ollow,the idea of ory is ore fully derived from What kind of information isac quired as declarative me ork with ex rimental animal is it best asse sed?The assessment of memory ha ry is als with a wo m call and ssed performanc and whichar independent mpanied by asens offamiliarity about the past.Amnes might this Many terms have been us dence ratings tha y attach ela and m also beer (Tulving,1983).This point d (or kinds)that ared ds to the expectation hat rec that erms ther s the he re ce about what the s does and does not do in the lves on either side of a biolog ally neful era particular event ntly or not,but tha e br in is the sh cy distinctions have proved to be prominent in theor processe more quickly and urately than new item kind o 1985 Mandler 1980)The idea ects can ord ory but twith the idea Adeclarative memory when nsiders the ncy to a recent ccurrence of the item.a relat have bee hat re t of es b able to dis expecte terms common with eack MEMORY DECLARATIVE NONDECLARATIVE SEMANTIC EPISODC 器 器警 y for fae of M d Ar ags From H s and Ne.Fro n Primates."p.437 7.by Zol

MEMORY AND THE HIPPOCAMPUS 205 ory for faces, spatial layouts, and other material that is declared by bringing a remembered image to mind rather than by ver￾balizing. The difficulty is that the term declarative is often linked to the notion of conscious memory. In the sections that follow, the idea of declarative memory is developed more fully in an attempt to make contact with similar ideas about memory systems derived from work with experimental animals. Declar￾ative memory is also contrasted with a heterogeneous collection of nondeclarative (implicit) memory abilities, which are ex￾pressed only through performance and which are independent of the structures damaged in amnesia (Figure 5). An additional point about the term declarative memory might be useful at this juncture. Many terms have been used to describe a particular kind of memory (e.g., declarative, explicit, relational, or configural), and many other terms have also been used to describe a kind (or kinds) that are dissociable from the first kind (e.g., nondeclarative, implicit, or habit). However, terms themselves are not the proper focus. If one considers the various biological and purely psychological concepts that have been used, it is striking that they sort themselves out in terms of ideas about what the hippocampus does and does not do in the service of memory. It should not be surprising that these terms place themselves on either side of a biologically meaningful boundary. The brain is the machinery that accomplishes mem￾ory, and history shows that other fundamental psychological distinctions have proved to be prominent in the organization of brain systems (e.g., short-term and long-term memory). Accord￾ingly, the term declarative is used here to describe one kind of memory, but not with the idea that it is especially different from other terms. The more important point is that the terms explicit memory and declarative memory, when one considers the prop￾erties that have been associated with each, describe a biologi￾cally real component of memory that depends on particular structures and connections in the brain. Accordingly, it is to be expected that these terms have much in common with each other and with the terms relational and configural, which come from work with rats. Recall, Recognition, and the Feeling of Familiarity What kind of information is acquired as declarative memory and how is it best assessed? The assessment of memory has relied traditionally on two methods: free recall and recognition. In normal subjects, both recall and recognition are typically accompanied by a sense of familiarity about the past. Amnesic patients perform poorly on tests of recall and recognition, and they have a diminished feeling of familiarity, as reflected in the low confidence ratings that they attach to their recognition choices. Recall and recognition have usually been taken as re￾flections of declarative memory (Tulving, 1983). This point of view leads to the expectation that recall and recognition should be proportionately impaired in amnesic patients and that con￾fidence judgments (which assess awareness about the knowl￾edge being reported) should be commensurate with the reduced level of memory performance. Another point of view is that recognition-memory perfor￾mance benefits not only from the ability to judge consciously whether a particular event has occurred recently or not, but that it also benefits from improved perceptual fluency (i.e., priming, a nonconscious process whereby recently encountered items are processed more quickly and accurately than new items; Gardiner, 1988; Jacoby & Dallas, 1981; Johnston, Dark, & Ja￾coby, 1985; Mandler, 1980). The idea is that subjects can ordi￾narily detect the facility or fluency with which they process a recently encountered test item and can then attribute this in￾creased fluency to a recent occurrence of the item. A related idea, which also supposes that recognition performance de￾pends greatly on nonconscious processes, is that subjects might sometimes be able to discriminate successfully between new and old items on a recognition test but be unable to reflect this MEMORY NONDECLARATIVE SKILLS MOTOR PERCEPTUAL COGMTIVE ADAPTATION LEVEL PRIMING PERCEPTUAL SEMANTIC DISPOSITIONS SMPLE CLASSICAL CONOmONMQ NONASSOCIATIVE HABITUATON SEMSmZATON SHUTS M JUDGMENT OPERANT AND PREFERENCE CONDmONMO Figure 5. A memory taxonomy. (Declarative memory includes memory for facts and events and depends on the integrity of the hippocampus and related structures. Nondeclarative memory refers to a heteroge￾neous collection of distinct learning and memory abilities where performance changes but without afford￾ing access to the experience or experiences that caused the change. From "Neuropsychological Investiga￾tions of Memory and Amnesia: Findings From Humans and Nonhuman Primates," p. 437, by S. Zola￾Morgan and L. R. Squire, 1990, in A. Diamond, The Development and Neural Bases of Higher Cognitive Functions, New York: New York Academy of Sciences. Copyright 1990 by the New York Academy of Sciences. Reprinted by permission.)

206 LARRY R.SQUIRE mnesic paticnts (Haist,Shimamura cognilio performanceonforced-cho test bu eyare in fact gues E(W ship an nong recall,rec an d confidence ratings (for the )sh recall scores and confidence ratings also matched The crucial finding was that despite the fact that primingand tive to confidend rating he recc enition iudgn address th recalforced-choice recognition judgm were no bette at several different retention intervals(15s to8 weeks)in both studies with normal subiects have also suggested that recogni RECAL RECAL 2w B 2h h4 ONFIDENCE RATINGS CONFIDENCE RATINGS 10m2 MN lid lin CON the

206 LARRY R. SQUIRE level of performance in verbal reports. For example, it has been suggested that amnesic patients should be able to exhibit suc￾cessful recognition performance on a forced-choice test but would then report that they are in fact guessing (Weiskrantz, 1988). These ideas all lead to the prediction that the relation￾ship among recall, recognition, and confidence ratings (for the recognition choices) should be different in amnesic patients than in normal subjects. For example, to the extent that recogni￾tion performance is based on perceptual fluency, which is in￾tact in amnesia, recognition should be disproportionately spared relative to recall. In addition, to the extent that recogni￾tion performance is governed by nonconscious processes, then recognition choices should also be disproportionately spared relative to confidence ratings. To address these issues, recall, forced-choice recognition, and confidence ratings for the recognition choices were tested at several different retention intervals (15 s to 8 weeks) in both normal subjects and amnesic patients (Haist, Shimamura, & Squire, in press). On all three measures the amnesic patients performed much worse than the normal subjects (Figure 6). Recall, recognition, and confidence ratings were similarly af￾fected. Specifically, when the recognition scores of amnesic pa￾tients and control subjects were matched (the scores of amnesic patients tested from 15 s to 10 min after learning matched con￾trol scores obtained from 1 day to 2 weeks after learning), the free-recall scores and confidence ratings also matched. These results suggest that recall, recognition, and feelings of familiarity are tightly linked functions of declarative memory. The crucial finding was that despite the fact that priming and other nonconscious memory processes are intact in amnesia, the recognition judgments of amnesic patients, and the confi￾dence ratings attached to these judgments, were no better than would have been predicted from the recall scores. Several other studies with normal subjects have also suggested that recogni- 100 90 80 o 70 I 60 0 50 c 0> 40 I 30 20 10 0 A RECALL T p* », ""•~^^T i^ ^^_^— i^ AMN 15s 1m 5m 10m 2h 100 90 80 70 60 50 40 30 20 10 0 -Chanca RECOGNITION AMN 15s In CONFIDENCE RATINGS AMN 15s in Retention Interval (log scale) 100 90 80 S 70 5 60 3 50 £ 40 | 30 20 10 0 B RECALL I "~-^^ ~^~4 ^^^s^4^ T**-^i 100 90 80 8 70 S 60 E 5° § 40 I 30 20 10 0 AMN 1 5s 1 m 5m 1 0m - -Chm* RECOGNITION AMN 15s 1m CONFIDENCE RATINGS AMN 15s 1m val (log scale) Figure 6. A: Free recall, two-choice recognition, and confidence ratings for the recognition choices for amnesic patients (AMN; solid lines) and control subjects (CON; dathed lines) woo wwe given different 20-word lists and tested at each of the indicated retention intervals, ft When the recognition scores of the two groups were matched by redrawing the same data, the scores for free recall and for confidence ratings also matched. Error bars represent standard error of the mean, s - seconds; m - minutes; h •- hours; d * days; w = weeks. (From "On the Relationship,Be*wcett Recall and Recognhion Memory" by F. Haist, A. P. Shimamura, and L. R. Squire, in press, Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, Figures 1 and 2. Copyright 1992 by the American Psychological Association.)

MEMORY AND THE HIPPOCAMPUS 207 ion r ory need not benefit from ming:that is.recogni oair the ability of rats to learn odor discriminations(Eichen tion can be a chan that other infl 1988 timing tasks (Mec in t-Wilso 1980 visual stimul d,1989.nm the 985) imple-obis scrim natio and concu 19892 esia (H that recall ca of th elps,Risse,&Volpe,19;Hirst.Joh son,Phelps,& findings in rats and monkeys ent with and they nnot learn or m on rath spatia bu they also imental design (Haist,Shima spatial memory d sibility that affe following hipp dam 9).In that stu sions were u nable tion by impaired search strategies and imp red ability to orga vere than of re nco ng D or ts( ecall and re nition was arr gedby stud ying amnesic pat os.lfci differ from eachother n important ways.qu n a cons rely or e aled in the uld be dis red ir mne a relative t s and must a ate the ohiect to the reogntionse d.By nd 6 Th arity judgments appea be tightly inked fur ofdecla ble that h one dam obicct-place task call i agrcethatrecognaitio more difficult an recognition,not because the object-place 988.Jo ton.Hawle and Elliott (19)concluded that pe The possibility that hir pal les ns might impair sp ia nally more tha een t te inc iousmemory hus,in a rea with damage to the hipp ocampus (Cave i8mamplfetmcoioaprtomancenaydawlimlebenei call the objects.r ects took th Spatial Memory als (from 、th One view about thes ective role of the hipp ory tas I be ea with at of the patient Th for spa mation ( the two tests was matched to the 978 int this the hippo f ndependent)space.This des ibes the unction of al me ry was pro mpal lesions ar tasks of spatial memory it is to the ct recogn

MEMORY AND THE HIPPOCAMPUS 207 tion memory need not benefit from priming; that is, recogni￾tion can be at chance levels despite the fact that other influ￾ences of stimuli on behavior can be detected (Bonnano & Stil￾lings, 1986; Kunst-Wilson & Zajonc, 1980; Mandler, Nakamura, & Van Zandt, 1987; Seamon, Brody, & Kauff, 1983; Seamon, Marsh, & Brody, 1984; Squire, Shimamura, & Graf, 1985). Two previously published reports concluded that recall can be disproportionately impaired in amnesia (Hirst, Johnson, Phelps, Risse, & Volpe, 1986; Hirst, Johnson, Phelps, & Volpe, 1988). In these studies, recall and recognition were compared at only a single point, and recall and recognition tests were given sequentially in the same session rather than in separate ses￾sions. However, we were unable to replicate this finding using the same experimental design (Haist, Shimamura, & Squire, in press). The different findings did not reflect differences in the severity of amnesia. While it remains unclear what factors do account for the different findings, one possibility is that varia￾tions in the locus of pathology are important. For example, damage to the frontal lobes could be expected to affect recall more than recognition (Jetter, Poser, Freeman, & Markowitsch, 1986), presumably because recall is affected more than recogni￾tion by impaired search strategies and impaired ability to orga￾nize incoming information. In summary, a test of alternative views about recall and recog￾nition was arranged by studying amnesic patients. If either rec￾ognition judgments or confidence ratings (i.e., feelings of famil￾iarity) were significantly supported by processes that are intact in amnesia (e.g., nonconscious memory processes that rely on increased facility of perceptual processing), then recall scores should be disproportionately impaired in amnesia relative to either recognition scores or confidence ratings. However, this effect was not observed. Instead, recall, recognition, and famil￾iarity judgments appear to be tightly linked functions of declar￾ative memory and similarly dependent on the brain systems damaged in amnesia. Other recent studies of normal subjects agree that recognition memory need not benefit from percep￾tual fluency (Hayman & Tulving, 1989; Watkins & Gibson, 1988). Johnston, Hawley, and Elliott (1991) concluded that per￾ceptual fluency may sometimes contribute to recognition per￾formance but that its contribution is small when explicit, conscious memory is readily available. Thus, in a real-world situation where material is relatively familiar (i.e., recognition performance is well above chance levels) and decision time is uncontrolled, recognition performance may draw little benefit from implicit memory. Spatial Memory One view about the selective role of the hippocampus in memory, derived especially from studies of rats, is that it is involved in memory for spatial information (OTCeefe & Nadel, 1978). According to this view, the hippocampus is a memory system that stores information about nonegocentric (viewpoint￾independent) space. This view describes the function of the hippocampus too narrowly. Although many of the tasks sensi￾tive to hippocampal lesions are tasks of spatial memory, it is also clear that hippocampal lesions impair nonspatial memory. For example, lesions of hippocampus or related structures im￾pair the ability of rats to learn odor discriminations (Eichen￾baum, Fagan, Mathews, & Cohen, 1988), timing tasks (Meek, Church, & Olton, 1984), and configural discriminations that involve unique combinations of auditory or visual stimuli (Rudy & Sutherland, 1989). In monkeys, lesions of the hippo￾campal formation impair recognition memory for visual ob￾jects, simple-object-discrimination tasks, and concurrent-dis￾crimination learning for objects (Zola-Morgan et al., 1989a). None of these tasks has an obvious spatial component. These findings in rats and monkeys are in agreement with findings from amnesic patients with hippocampal damage. The patients do lose their way, and they cannot learn or re￾member spatial layouts, but they also forget prose passages, tactual impressions, odors, faces, and melodies. In amnesia, spatial memory impairment is just one aspect of a broad im￾pairment in (declarative) memory (Squire, 1979). A recent study in monkeys raised the possibility that, whereas memory is broadly affected following hippocampal damage, spatial memory ability might be disproportionally im￾paired (Parkinson, Murray, & Mishkin, 1988). In that study, monkeys with hippocampal formation lesions were unable to learn object-place associations. The impairment was much more severe than was observed in an earlier study of recogni￾tion memory for visual objects (Mishkin, 1978). However, these two tasks (object-place memory and visual recognition mem￾ory) differ from each other in important ways, quite apart from the fact that one of the tasks is spatial and the other is not. For example, in the object-place task the location of an object must be recalled in the absence of external cues. The monkey is confronted with two identical objects, placed in two familiar locations, and must associate the object to the spatial location that was recently rewarded. By contrast, the visual object task is a task of recognition. The monkey is confronted with two dif￾ferent objects (a novel one and a familiar one) and must recog￾nize which one was recently presented. Accordingly, it is possi￾ble that monkeys failed the object-place task because recall is more difficult than recognition, not because the object-place task requires spatial memory. The possibility that hippocampal lesions might impair spa￾tial memory disproportionally more than nonspatial memory has been tested directly in amnesic patients, including patients with confirmed damage to the hippocampus (Cave & Squire, 1991). Fourteen amnesic patients inspected an array of 16 toy objects (cf. Smith & Milner, 1981) and were subsequently asked to recall the objects, recognize their names on a multiple-choice test, and then reconstruct the array by placing the objects in their original locations. Normal subjects took the same tests, at one of several different retention intervals (from 5 min to 5 weeks after learning), so that their performance on the object￾memory tasks could be equated with that of the patients. The results were that, when performance of the amnesic patients on the two object-memory tests was matched to the object-me￾mory performance of control subjects, spatial memory perfor￾mance was also equivalent for amnesic patients and control subjects. That is, the impairment in spatial memory was pro￾portional to the impairment in object recall and object recogni￾tion. This result might seem in conflict with earlier studies in rats with hippocampal lesions, which have reported that spatial