The Cognitive Neuroscience of Human Memory Since H.M.

Download The Cognitive Neuroscience of Human Memory Since H.M.

Post on 08-Dec-2016

214 views

Category:

Documents

2 download

TRANSCRIPT

  • NE34CH12-Squire ARI 13 May 2011 12:29

    The Cognitive Neuroscienceof Human MemorySince H.M.Larry R. Squire1,2,3,4 and John T. Wixted41Veterans Affairs Healthcare System, San Diego, California 921612Department of Psychiatry, 3Department of Neurosciences, and 4Department ofPsychology, University of California, San Diego, La Jolla, California 92093;email: lsquire@ucsd.edu, jwixted@ucsd.edu

    Annu. Rev. Neurosci. 2011. 34:25988

    First published online as a Review in Advance onMarch 29, 2011

    The Annual Review of Neuroscience is online atneuro.annualreviews.org

    This articles doi:10.1146/annurev-neuro-061010-113720

    Copyright c 2011 by Annual Reviews.All rights reserved

    0147-006X/11/0721-0259$20.00

    Keywords

    medial temporal lobe, hippocampus, neocortex, anterograde amnesia,retrograde amnesia

    Abstract

    Work with patient H.M., beginning in the 1950s, established key prin-ciples about the organization ofmemory that inspired decades of experi-mental work. SinceH.M., the study of humanmemory and its disordershas continued to yield new insights and to improve understanding ofthe structure and organization of memory. Here we review this workwith emphasis on the neuroanatomy of medial temporal lobe and dien-cephalic structures important for memory, multiple memory systems,visual perception, immediatememory,memory consolidation, the locusof long-term memory storage, the concepts of recollection and famil-iarity, and the question of how differentmedial temporal lobe structuresmay contribute differently to memory functions.

    259

    Ann

    u. R

    ev. N

    euro

    sci.

    2011

    .34:

    259-

    288.

    Dow

    nloa

    ded

    from

    ww

    w.an

    nual

    revi

    ews.o

    rgby

    Uni

    vers

    ity o

    f Miss

    ouri

    - Col

    umbi

    a on

    04/

    14/1

    3. F

    or p

    erso

    nal u

    se o

    nly.

  • NE34CH12-Squire ARI 13 May 2011 12:29

    Contents

    INTRODUCTION . . . . . . . . . . . . . . . . . . 260ANATOMY OF MEMORY . . . . . . . . . . 261PRINCIPLES OF ORGANIZATION

    SUGGESTED BY H.M.SFINDINGS . . . . . . . . . . . . . . . . . . . . . . . 265

    MULTIPLE MEMORY SYSTEMS . . 266VISUAL PERCEPTION . . . . . . . . . . . . . 267IMMEDIATE MEMORY AND

    WORKING MEMORY . . . . . . . . . . . 268REMOTE MEMORY AND

    MEMORY CONSOLIDATION. . . 271MEMORY IN THE NEOCORTEX. . 273

    Achromatopsia . . . . . . . . . . . . . . . . . . . . . 275Prosopagnosia . . . . . . . . . . . . . . . . . . . . . 275Amusia . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276Knowledge Systems . . . . . . . . . . . . . . . . 277

    RECOLLECTION ANDFAMILIARITY. . . . . . . . . . . . . . . . . . . . 278Recall versus Recognition . . . . . . . . . . 278Remember/Know Procedure . . . . . . . 279Analysis of the Receiver Operating

    Characteristic . . . . . . . . . . . . . . . . . . . 279Newer (Model-Free) Methods . . . . . . 281

    GROUP STUDIES ANDMULTIPLE METHODS . . . . . . . . . 281

    CONCLUSIONS . . . . . . . . . . . . . . . . . . . . 282

    INTRODUCTION

    In the earliest systematic writings about humanmemory, it was already appreciated that thestudy of memory impairment can provide valu-able insights into the structure and organiza-tion of normal function (Ribot 1881, Winslow1861). This tradition of research has contin-ued to prove fruitful and has yielded a broadrange of fundamental information about thestructure and organization of memory. Whatis memory? Is it one thing or many? What arethe concepts and categories that guide our cur-rent understanding of how memory works andthat underlie the classication of its disorders?It is sometimes not appreciated that the con-cepts and categories used in current discussions

    of memory are not xed and were not easily es-tablished. Even the question of which cognitiveoperations reect memory and which dependon other faculties has a long history of empiri-cal work and discussion.

    One needs only to sample nineteenth-century writings to recognize how differentlymemory was viewed then and now. Forexample, in his classic treatment of memorydisorders, Ribot (1881) considered amnesiasdue to neurological injury together withamnesias due to psychological trauma. Andhe viewed aphasia and agnosia as disordersof memory, wherein (in aphasia, for example)patients have lost their memory for words ormemory for the movements needed to producewords. Today, aphasia is considered a decitof language, and agnosia a decit of visualperception. Memory is affected but only as partof a more fundamental defect in a specic kindof information processing.

    The notion that the study of brain injury canelucidate the organization of memory was itselfa matter for empirical inquiry. If brain regionswere highly interconnected, and the brainsfunctions distributed and integrated one withanother, then damage to any one area wouldproduce a global impairment, blurred acrossmultiple faculties and affecting all ofmental life.But the fact of the matter is different. The brainis highly specialized and modular, with differ-ent regions dedicated to specic operations. Asa result, localized damage can produce strik-ingly specic effects, including a selective andcircumscribed impairment of memory.

    The idea that functions of the nervous sys-tem can be localized was already well acceptedby the end of the nineteenth century. This lo-calizationist view had its roots in the writingsof Gall (1825) and was supported by the exper-imental work of Broca (1861), Ferrier (1876),Fritsch &Hitzig (1870), and others (see Finger1994). Yet, these ideas centered mainly aroundsensory functions, motor control, and languageand did not usefully address the topic of mem-ory. Then, in the early twentieth century, aninuential program of experimental work in ro-dents investigated directly the localization of

    260 Squire Wixted

    Ann

    u. R

    ev. N

    euro

    sci.

    2011

    .34:

    259-

    288.

    Dow

    nloa

    ded

    from

    ww

    w.an

    nual

    revi

    ews.o

    rgby

    Uni

    vers

    ity o

    f Miss

    ouri

    - Col

    umbi

    a on

    04/

    14/1

    3. F

    or p

    erso

    nal u

    se o

    nly.

  • NE34CH12-Squire ARI 13 May 2011 12:29

    memory with the conclusion that memory isdistributed throughout the cortex and that thecontribution tomemory is equivalent across re-gions (Lashley 1929). This idea was stronglychallenged (Hebb 1949, Hunter 1930) by thealternative, and more modern, interpretationthat memory storage is indeed distributed butthat different areas store different features of thewhole. Still, as the midpoint of the twentiethcentury approached, memory functions, whiledistributed, were thought to be well integratedwith perceptual and intellectual functions, andno region of the brain was believed to be dis-proportionately dedicated to memory. All thatwas about to change.

    In 1957, Brenda Milner reported theprofound effect on memory of bilateral medialtemporal lobe resection, carried out to relieveepilepsy in a patient who became known asH.M. (19262008) (Scoville & Milner 1957,Squire 2009) (Figure 1). Remarkably, H.M.exhibited profound forgetfulness but in theabsence of any general intellectual loss orperceptual disorders. He could not form newmemories (anterograde amnesia) and alsocould not access some memories acquiredbefore his surgery (retrograde amnesia). Hisimpairment extended to both verbal and non-verbal material, and it involved informationacquired through all sensory modalities. Thesendings established the fundamental principlethat memory is a distinct cerebral function,separable from other perceptual and cognitiveabilities, and also identied the medial aspectof the temporal lobe as important for memory.The early descriptions of H.M. can be said tohave inaugurated themodern era ofmemory re-search, and the ndings fromH.M. enormouslyinuenced the direction of subsequent work.

    ANATOMY OF MEMORY

    The work with H.M. is sometimes cited incor-rectly as evidence of the importance of the hip-pocampus formemory, but this particular pointcould not of course be established by a large le-sion that included not only the hippocampusbut also the amygdala together with the adja-

    cent parahippocampal gyrus. Which structureswithinH.M.s lesion are important for memorybecame understood only gradually during the1980s following the successful development ofan animal model of human amnesia in the non-human primate (Mishkin 1978). Cumulativestudies in the monkey (Murray 1992, Squire &Zola-Morgan 1991, Zola-Morgan et al. 1994)considerably claried this issue. The impor-tant structures proved to be the hippocam-pus and the adjacent entorhinal, perirhinal,and parahippocampal cortices, which make upmuch of the parahippocampal gyrus (Figure 2).

    One particularly instructive case of humanmemory impairment became available duringthis same time period (Zola-Morgan et al.1986). R.B. developed a moderately severe,enduring impairment following an ischemicepisode in 1978. During the ve years until hisdeath, his memory decit was well documentedwith formal tests. Detailed histological exami-nation of his brain revealed a circumscribed bi-lateral lesion involving the entire CA1 eld ofthe hippocampus. Note that a lesion connedto the CA1 eld must substantially disrupt hip-pocampal function because the CA1 eld is abottleneck in the unidirectional chain of pro-cessing that begins at the dentate gyrus and endsin the subiculum and entorhinal cortex. R.B.was the rst case of memory impairment fol-lowing a lesion limited to the hippocampus thatwas supported by extensive neuropsychologicaltesting as well as neuropathological analysis.

    Thendings fromR.B., considered togetherwith the much more severe impairment inH.M., made two useful points. First, damage tothe hippocampus itself is sufcient to producea clinically signicant and readily detectablememory impairment. Second, additional dam-age to the adjacent cortical regions along theparahippocampal gyrus (as in H.M.) greatly ex-acerbates thememory impairment. These sameconclusions about the neuroanatomy of mod-est and severe memory impairment were alsoestablished in the monkey (Zola-Morgan et al.1994).

    Another case was subsequently described(patient G.D.) with a histologically conrmed

    www.annualreviews.org Neuroscience of Human Memory Since H.M. 261

    Ann

    u. R

    ev. N

    euro

    sci.

    2011

    .34:

    259-

    288.

    Dow

    nloa

    ded

    from

    ww

    w.an

    nual

    revi

    ews.o

    rgby

    Uni

    vers

    ity o

    f Miss

    ouri

    - Col

    umbi

    a on

    04/

    14/1

    3. F

    or p

    erso

    nal u

    se o

    nly.

  • NE34CH12-Squire ARI 13 May 2011 12:29

    bilateral lesion conned to the CA1 eld andwith a memory impairment very similar to R.B.(Rempel-Clower et al. 1996). Two other pa-tients were also of interest. L.M. andW.H. had

    somewhat more severe memory impairmentthan didR.B. andG.D., but the impairmentwasstill moderate in comparison toH.M. (Rempel-Clower et al. 1996). Histological examination

    A

    A

    H

    H

    cscs

    cscs

    PRPRECEC

    ECEC

    MMNMMN

    V

    V

    V

    cscs

    PRPRECEC

    A

    A

    H

    H

    cs

    cs

    PREC

    EC

    MMN

    V

    V

    V

    cs

    PREC

    a

    Patient H.M. Healthy 66-year-old male

    Rostral

    Caudal

    b

    c

    AH

    MMN

    cs

    Amygdaloid complexHippocampal region

    Medial mammillary nuclei

    Collateral sulcusV Lateral ventricle

    PREC

    Perirhinal cortexEntorhinal cortex

    262 Squire Wixted

    Ann

    u. R

    ev. N

    euro

    sci.

    2011

    .34:

    259-

    288.

    Dow

    nloa

    ded

    from

    ww

    w.an

    nual

    revi

    ews.o

    rgby

    Uni

    vers

    ity o

    f Miss

    ouri

    - Col

    umbi

    a on

    04/

    14/1

    3. F

    or p

    erso

    nal u

    se o

    nly.

  • NE34CH12-Squire ARI 13 May 2011 12:29

    revealed extensive bilateral lesions of the hip-pocampal region, involving all theCAelds andthe dentate gyrus. There was also some cell lossin entorhinal cortex and, for W.H., cell loss inthe subiculum, as well. The more severe mem-ory impairment in these two cases, in compar-ison to R.B. and G.D., could be due to the ad-ditional damage within the hippocampus or tothe cell loss in entorhinal cortex.

    There are only a small number of caseswhere detailed neuropsychological testingand thorough neurohistological analysis havecombined to demonstrate memory impairmentafter limited hippocampal damage or largermedial temporal lobe lesions (see also Victor& Agamanolis 1990). Yet, neuroanatomicalinformation is essential because it lays thegroundwork for classifying memory disorders,for understanding qualitative and quantitativedifferences between patients, and for address-ing questions about how specic structuresmaycontribute differently to memory functions.Nonetheless, in the absence of histologicaldata, valuable information can be obtainedfrom structural imaging. Methods for high-resolution imaging of hippocampal damagewere developed some time ago (Press et al.1989), and quantitative data can now be ob-tained that provide reliable estimates of tissuevolume (Gold & Squire 2005). These estimatesare based on guidelines dened histologicallyand use landmarks in the medial temporallobe that are visible on MRI (Insausti et al.1998a,b).

    An interesting observation has emergedfrom calculations of hippocampal volume inmemory-impaired patients, usually patients

    who have sustained an anoxic episode. Acrossa number of reports, hippocampal volume (orarea in the coronal plane) is typically reduced by40% [41%, n = 10 (Isaacs et al. 2003); 44%,n = 5 (Shrager et al. 2008); 43%, n = 4 (Squireet al. 1990); 45%, n = 1 (Cipolotti et al. 2001);46%, n = 1 (Mayes et al. 2002)]. Neurohis-tological data from two of these patients (L.M.andW.H.) suggest an explanation for this strik-ing consistency. As described above, these twopatients had extensive cell loss in the hippocam-pus as well as in the dentate gyrus. Accordingly,a reduction in hippocampal volume of 40%, asestimated by MRI, may indicate a nearly com-plete loss of hippocampal neurons. The tissuecollapses, but it does not disappear entirely. Avolume loss in the hippocampus of 40% mayrepresent a maximum value for some etiologiesof memory impairment.

    While medial temporal lobe structures havereceived the most attention in studies of mem-ory and memory impairment, it is notable thatdamage to the diencephalicmidline also impairsmemory. The decit has essentially the samefeatures as in medial temporal lobe amnesia.The best-known cause of diencephalic amnesiais alcoholic Korsakoff s syndrome. Here,damage to the medial dorsal thalamic nucleus(alone or perhaps in combination with damageto the mammillary nuclei) has been associatedwith memory impairment (Victor et al. 1989).Another survey of Korsakoff s syndrome docu-mented damage to these two structures and, inaddition, identied a role for the anterior thala-mic nuclei (Harding et al. 2000). Six cases thatwere studied both neuropsychologically andneurohistologically (Gold & Squire 2006, Mair

    Figure 1Left column. Magnetic resonance images arranged from rostral (a) to caudal (c) through the temporal lobe of patient H.M. (in 1993 atage 67) and a 66-year-old healthy male (right). The comparison brain illustrates the structures that appear to have been removed duringH.M.s surgery in 1953. The lesion was bilaterally symmetrical, extending caudally 5.4 cm on the left side and 5.1 cm on the right. Thefull caudal extent of abnormal tissue is not illustrated. The damage included medial temporal polar cortex, most of the amygdaloidcomplex, virtually all the entorhinal cortex, and approximately the rostral half of the hippocampal region (dentate gyrus, hippocampus,and subicular complex). The perirhinal cortex was substantiall...

Recommended

View more >