eyeblink classical conditioning in h.m.: delay and trace paradigms · 2018-03-16 · eyeblink...

Behavioral Neuroscience1993, Vol. 107, No. 6, 911-925

Copyright 1993 by the American Psychological Association, Inc.0735-7044/93/S3.00

Eyeblink Classical Conditioning in H.M.: Delay and Trace Paradigms

Diana S. Woodruff-Pak

H.M., a well-known subject with bilateral removal of medial-temporal-lobe structures andprofound amnesia, performed eyeblink classical conditioning (EBCC) for 21 90-trial sessions in the400-ms delay and 900-ms trace paradigms. A 2nd amnesic subject with temporal lobe lesions and 2normal control subjects (NCSs) were also conditioned. Acquisition occurred in both paradigms forall subjects. Acquisition in the delay paradigm was prolonged in H.M. (perhaps because of hiscerebellar degeneration in the vermis and hemispheres), but not in the 2nd amnesic subject.Amnesic subjects and NCSs showed more rapid acquisition in the trace than in the delay paradigm.Two years after initial EBCC, H.M. attained learning criterion in the trace paradigm in !/H)th asmany trials. No recollection of the experimenters, apparatus, instructions, or procedure wasmanifested by H.M. Results suggest that humans can condition in the 400-ms delay and 900-mstrace EBCC paradigms with the hippocampus radically excised.

Human long-term memory systems have been classified intotwo major forms: explicit (or declarative) memory and implicit(or procedural) memory (Cohen & Squire, 1980; Graf &Schacter, 1985). Explicit memory, which can be conceptualizedas learning with awareness, refers to the acquisition andretention of information about events and facts. It is assessedby accuracy on recall and recognition tests (e.g., word listlearning). Four decades of amnesia research support theconclusion that medial-temporal/diencephalic memory cir-cuits are critical for establishing long-term memory for eventsand facts (Squire, 1992). In contrast, there is considerableevidence that medial-temporal/diencephalic memory circuitsare not critical for several kinds of implicit memory (Schacter,1987; Shimamura, 1986). Implicit memory may be conceptual-ized as learning without awareness, which can be measured bychanges in performance. It consists of multiple, dissociableprocesses, including (a) repetition priming effects and (b) theacquisition and retention of motor, perceptual, or problem-solving skills. Repetition priming is assessed by a decrease in

Diana S. Woodruff-Pak, Department of Psychology, Temple Univer-sity, and Laboratory of Cognitive Neuroscience, Philadelphia Geriat-ric Center, Philadelphia, Pennsylvania.

This research was supported by National Institute on Aging Grant 1RO1 AGO9752-01 and Alzheimer's Association Grants PRG-89-066and IIRG-91-059 to Diana S. Woodruff-Pak. The MassachusettsInstitute of Technology Clinical Center received support from Na-tional Institute of Health Grant RR 00088. These results werepresented in part in November 1991 at the 21st Annual Meeting of theSociety for Neuroscience, New Orleans, Louisiana.

I thank Philip Cho, Joan Coffin, Richard Finkbiner, Mark Map-stone, Hyung W. Pak, Robert Sugiura, and Joy Yucaitis, who assistedwith data collection; Brenda Milner for granting permission toexamine H.M. on repeated occasions; and Suzanne Corkin, whoinitiated this research project and provided invaluable comments andsuggestions on the many drafts of this article. She nurtured theresearch through every stage, and her assistance is gratefully acknowl-edged.

Correspondence concerning this article should be addressed toDiana Woodruff-Pak, Department of Psychology, Temple University,Philadelphia, Pennsylvania 19122. Electronic mail may be sent [email protected].

reaction time or a bias in response to particular words orpatterns as a consequence of prior exposure (during theexperiment) to those words or patterns. Skill learning isassessed by improvement in speed and accuracy across trialson repetitive sensorimotor tasks (e.g., rotary pursuit learning).

Research on skill learning in amnesia was initiated byMilner and Corkin and their colleagues in the 1960s (Corkin,1965, 1968; Milner, 1965; Milner, Corkin, & Teuber, 1968).For example, Corkin trained H.M., the most thoroughlystudied subject with bilateral medial-temporal-lobe resection,on a variety of manual tracking and coordination tasks.Although he had no memory of performing the task, H.M.showed improvement from session to session. He had noexplicit memory of the training sessions, but his performanceimproved in a fashion similar to that of individuals with intactmedial-temporal regions. Amnesic subjects have shown robustlearning of a number of motor skills (e.g., Parkin, 1982; Starr &Phillips, 1970) and perceptual and cognitive skills (e.g., Benz-ing & Squire, 1989; Cohen & Squire, 1980). On repetitionpriming, Warrington and Weiskrantz (1974, 1978) demon-strated that amnesic subjects could show normal retention of alist of familiar words when tested with word-stem or fragmentcues, whereas these same subjects were profoundly impairedon free recall and recognition tests. Similarly, H.M. had severerecall and recognition deficits, but he had word completionpriming scores equivalent to the mean of normal controlsubjects (NCSs). H.M. also showed a pattern of priming tononverbal material (visually presented patterns) equivalent inmagnitude to priming in NCSs (Gabrieli, Milberg, Keane, &Corkin, 1990).

In addition to dissociations between implicit and explicitmemory in amnesics in skill learning and repetition priming,these subjects show dissociations between implicit and explicitmemory in various other situations including eyeblink classicalconditioning (EBCC). Because it can occur even when thesubject is not aware that learning is occurring, EBCC isimplicit. Since the 1940s investigators have documented thatthere is little relationship between subjects' reported aware-ness and their performance in EBCC (Grant, 1973; Kimble,1962; Norris & Grant, 1948; Spence, 1966). Thompson (1989)

911

912 DIANA S. WOODRUFF-PAK

elaborated on the aspects of EBCC that qualify it as aprocedural or explicit task. Along with any discrete behavioralresponse learned by mammals to deal with an aversive event,EBCC is a category of associative learning that appears toinvolve the cerebellum in acquisition and storage of thememory trace.

One advantage in using EBCC as a behavioral measure isthat in rabbits, the neural circuitry underlying classical condi-tioning of the nictitating membrane (NM)/eyeblink responsehas been delineated extensively (Thompson, 1986, 1990). Theipsilateral cerebellum is essential for acquisition and retention(Lavond, Hembree, & Thompson, 1985; McCormick & Thomp-son, 1984a, 1984b; Steinmetz, Lavond, Ivkovich, Logan, &Thompson, 1992). Pyramidal cells in the CA1 region in thehippocampus develop a neural model of the behavioral eye-blink response as rabbits acquire conditioned responses (CRs;Berger, Rinaldi, Weisz, & Thompson, 1983; Berger & Thomp-son, 1978a). Although the hippocampus is engaged duringacquisition in EBCC (Berger, Alger, & Thompson, 1976;Berger & Thompson, 1978b; Thompson et al., 1980), rabbitsacquire CRs normally in the delay paradigm in the absence of ahippocampus (Schmaltz & Theios, 1972; Solomon & Moore,1975). In contrast, lesions of the cerebellar interpositus nucleusin rabbits eliminate hippocampal CA1 pyramidal cell firingwith the behavioral response (Clark, McCormick, Lavond, &Thompson, 1984; Sears & Steinmetz, 1990).

Although the hippocampus is not essential for acquisition inthe delay classical conditioning paradigm in rabbits, hippocam-pal manipulations can increase or decrease the rate of condi-tioning. For this reason the hippocampus is said to play amodulatory role in NM/EBCC (Berger, Berry, & Thompson,1986). Disruption of muscarinic receptors in the septo-hippocampal cholinergic system with scopolamine injectionsimpairs acquisition of the conditioned NM response (Harvey,Gormezano, & Cool-Hauser, 1983; Moore, Goodell, & Solo-mon, 1976; Solomon, Solomon, Vander Schaaf, & Perry, 1983)and eliminates pyramidal cell activity in conjunction with theNM response (Salvatierra & Berry, 1989). Electrical stimula-tion of the preforant path establishing long-term potentiationin the hippocampus causes more rapid classical conditioning ina discrimination paradigm (Berger, 1984).

The standard arrangement of the presentation of stimuli ineyeblink conditioning is the delay paradigm. In the delayparadigm, the conditioned stimulus (CS) is presented first. Inrabbits, optimal conditioning occurs when the interval betweenthe CS onset and the unconditioned stimulus (US) onset isbetween 250 and 500 ms, and the tone and airpuff coterminate(Gormezano, 1966). The optimal CS-US interval in humans iswider, extending from 400 to 1400 ms (Finkbiner & Woodruff-Pak, 1991). In the trace paradigm, the CS is turned off beforethe airpuff is presented. A blank period intervenes betweenthe CS and the US. The term trace is used because the subjectmust form a memory trace of the CS to associate it with thelater arriving US. Two aspects of the experimental conditionsusually differentiate the delay and trace paradigms: (a) Thereis no overlap between the CS and the US in the trace paradigmand (b) the interval between CS onset and US onset isnormally longer in the trace paradigm. Providing older humansubjects with a longer CS-US interval permits them to produce

a higher percentage of CRs (Finkbiner & Woodruff-Pak, 1991;Solomon, Blanchard, Levine, Valezquez, & Groccia-EHison,1991).

There is some disagreement about the role of the hippocam-pus in the trace paradigm in rabbits. Several laboratories havereported that the timing of the CR in the trace (but not thedelay) paradigm with a 500-ms trace interval is disrupted withremoval of the dorsal hippocampus (James, Hardiman, & Yeo,1987; Port, Romano, Steinmetz, Mikhail, & Patterson, 1986;Solomon, Vander Schaaf, Weisz, & Thompson, 1986). Re-searchers in another laboratory argued that more completehippocampal removal prevents acquisition in the trace para-digm with a 500-ms trace but not with a 300-ms trace (Moyer,Deyo, & Disterhoft, 1990). The latter study used a 100-ms CScompared with a 250-ms CS used in the former studies. One ofthe aims of the present study was to compare the role ofhippocampal removal in humans in the 400-ms delay paradigmand the 900-ms trace classical conditioning paradigm with a500-ms trace interval.

Comparisons of the acquisition and retention of NM/EBCCin rabbits and humans reveal strong behavioral parallelsbetween the two species on this learning task (Gormezano,1966). There are also parallel aging effects in rabbits andhumans on NM/EBCC (WoodrufT-Pak, 1988). Because of theclear behavioral parallels, investigators have begun to explorewhether the brain structures engaged in EBCC in humans arethe same structures involved in rabbits.

Studies with humans have examined classical conditioning insubjects with global amnesia, a condition that results frombilateral lesions of the medial-temporal/diencephalic system,and in subjects (presumably without global amnesia) who hadunilateral temporal-lobe lesions. In a study of 2 amnesicsubjects (1 subject with KorsakofFs syndrome and 1 posten-cephalitic [PE] subject), Weiskrantz and Warrington (1979)reported evidence of EBCC in the delay paradigm, withretention at 10 min and 24 hr. However, the subjects' perfor-mance was not compared with that of NCSs. The measures ofconditioning consisted of subjective judgments about theoccurrence of CRs and unconditioned responses (URs) madefrom a videotape. Daum, Channon, and Canavan (1989) repli-cated Weiskrantz and Warrington's (1979) results with 3amnesic subjects (1 PE subject and 2 subjects with epilepsy ofunspecified etiology). This study examined EBCC with a delayparadigm involving a 720-ms CS-US interval. Data on extinc-tion, discrimination conditioning, and discrimination reversalwere also collected. Responses were recorded and analyzed bycomputer. Results demonstrated that acquisition in the delayparadigm was intact. All 3 subjects were impaired in discrimi-nation learning (measured by pairing the airpuff with one freq-uency of tone CS and not with a second different frequency oftone CS) and discrimination reversal (pairing the airpuff withthe second frequency of tone CS and not with the CS fre-quency with which the airpuff was originally paired). The dataon extinction were ambiguous and difficult to interpret. Unfor-tunately, data on NCSs were not provided in this study. In asubsequent study, Daum, Channon, Polkey, and Gray (1991)investigated electrodermal as well as discrimination condition-ing in 17 NCSs and 17 subjects who had undergone unilateralen bloc resection of the right or left temporal lobe for the relief

EYEBLINK CLASSICAL CONDITIONING IN H.M. 913

of intractable epilepsy. Daum et al. (1991) found that theacquisition of CRs was comparable in these unilateral hippo-campectomized subjects to the normal control group butperformance on discrimination conditioning was impaired.Subjects with unilateral hippocampal lesions produced CRs toboth tone CSs, whereas NCSs discriminated by making CRsonly to the CS to which the US was paired.

A recently published article on EBCC in 5 patients with purecerebellar cortical atrophy and 7 patients with olivopontocer-ebellar atrophy demonstrated severe impairment in acquisi-tion in these patients (Topka, Valls-Sole, Massaquoi, &Hallett, 1993). These results supported the role of the cerebel-lum in the expression and timing of the CR. Solomon, Stowe,and Pendlebury (1989) reported impaired delay EBCC relativeto NCSs in a subject with cerebellar dysfunction associatedwith an atrial myxoma (tumor in the heart). The subjectpresumably suffered multiple strokes caused by emboli fromthe tumor. Solomon, Stowe, and Pendlebury (1989) suggestedthat the lesions interrupted afferents to the cerebellum. Lye,O'Boyle, Ramsden, and Schady (1988) described conditioningin the delay paradigm in a subject with a lesion restricted to theright cerebellum. A total of 396 paired presentations of the CSand US to the right eye resulted in very few CRs. When theairpuffUS was switched to the left eye, CRs were elicited on 35of 36 stimulus presentations beginning with the second trial.Results with patients reported by Topka et al. (1993) alongwith the two case studies provide initial suggestive evidencethat the ipsilateral cerebellum is essential for EBCC inhumans, just as it is in rabbits.

The primary purpose of the present study was to determinewhether H.M., with almost complete bilateral removal of thehippocampus, would show normal performance in EBCC. Thisis not the first attempt to determine whether classical condition-ing is normal in H.M. Some years ago, Kimura (reported inMilner et al., 1968) set out to classically condition H.M.,intending to use a neutral stimulus as the CS, with an electricshock US, to elicit a galvanic skin response (GSR) as a UR.However, Kimura had to abort the experiment because H.M.showed no GSR to electric shock, even at intensity levels thatNCSs found painful.

In addition to the large bilateral medial-temporal-lobelesion, H.M. has marked cerebellar atrophy in the vermis andhemispheres. The cerebellar lesion could impair acquisition inboth the delay and trace paradigms. For this reason, a 2ndamnesic subject with bilateral lesions in the medial temporallobe resulting from herpes simplex encephalitis (the PEsubject) was included in the sample. Given that the hippocam-pus is not essential for acquisition of the conditioned NM/eyeblink response in the delay paradigm in rabbits, it waspredicted that acquisition would occur in the delay paradigm,despite H.M.'s almost complete bilateral removal of hippocam-pus. In the trace paradigm, it was anticipated that H.M. andthe PE subject might show some abnormal short-latency CRs.If the hippocampus is crucial for trace but not delay condition-ing, both amnesic subjects should show more impaired perfor-mance in the trace paradigm than in the delay paradigm.Because H.M. has cerebellar atrophy, it was predicted that hisacquisition might be impaired relative to NCSs. In contrast,

the PE subject had neither clinical nor radiologic evidence ofcerebellar abnormality.

The present study expanded on earlier studies in severalways. It incorporated delay and trace paradigms within a singleexperimental protocol, thus permitting a direct comparison oftheir efficacy in amnesia; it examined performance over longretention intervals (5 weeks and 2 years); it included a subject(H.M.) whose amnesia is profound and clearly attributable toextensive and well-documented bilateral medial-temporal-lobe resection (Corkin, 1984); it illustrated the effect of medial-temporal lesions with temporal neocortex intact (H.M.); and itincluded NCSs matched to each amnesic subject for age andsex.

Method

Subjects

The subjects were 2 men with amnesia associated with bilateralmedial-temporal-lobe lesions and 2 NCSs. The NCSs were matched tothe amnesic subjects on the basis of age and sex (see Table 1). The 1stamnesic subject (H.M.) underwent bilateral removal of medial-temporal-lobe structures (including the anterior 8 cm of the hippocam-pus) in 1953 for seizure control. H.M. had marked cerebellar atrophyin the vermis and hemispheres, probably the result of long-termtreatment with Dilantin. H.M. was 64 years old at the time of initialclassical conditioning sessions and had 12 years of education. NCS1was 66 years old and had 16 years of education. The 2nd amnesicsubject had medial-temporal-lobe lesions resulting from herpes sim-plex encephalitis, which severely damages medial-temporal-lobe struc-tures and results in loss of CA1 pyramidal cells (Adams, Corsellis, &Duchen, 1984). A computerized tomography (CT) scan performed 27years after the onset of the disease showed tissue loss in the posterior,inferior aspect of the left temporal lobe, which corresponds to the leftmedial-temporal-lobe region. Given that the PE subject had herpessimplex encephalitis, it was assumed that damage to medial-temporal-lobe structures was bilateral, although left medial-temporal-lobelesions were most evident in the CT scan. No cerebellar degenerationwas apparent. The PE subject was 54 years old and had 16 years ofeducation. NCS2 was 57 years old and had 12 years of education.

Apparatus for EBCC

The classical conditioning apparatus consisted of a portable, auto-mated system with an infrared eyeblink detector and airpufT jetattached to headgear and input to an interface box containing amicroprocessor and a miniature air compressor. The interface box was

Table 1Characteristics of 2 Amnesic and 2 Normal Control Subjects(NCSs) in Eyeblink Classical Conditioning Study

Subject

H.M.

NCS1PE

NCS2

Age(years)

64

6654

57

Education(years)

12

1616

12

Etiology ofmemory loss

Medial temporallobe resection

Herpes simplexencephalitis

Note. All subjects were men. H.M. is identified by his initials,whereas the PE (postencephalitis) subject has been identified by theetiology of his amnesia. Dashes indicate not applicable.


interactive with an IBM 386 computer. The headgear had an adjust-able headband holding the airpuffjet and infrared eyeblink monitor.Filtered air was compressed in the interface box and delivered througha jet, which was set about 2 cm from the cornea.

The voltage from the infrared device was amplified and differenti-ated; eyelid movement data were transferred to the computer foranalysis and storage. From a speaker in the interface box, a 1000-Hz,80-dB SPL tone CS was delivered. Air for the corneal airpuff US wascompressed in the interface box, filtered, and delivered at a pressure of5 to 15 psi. US intensity, comparable for all 4 subjects in the initialtesting, was 7 psi. In the 2-year follow-up study with H.M., US intensitywas increased and ranged from 7 to 14 psi during the two sessions. Thetiming of stimulus presentations for the tone CS and airpuff US wascontrolled by the microprocessor in the interface box.

Procedure

Subjects sat in a comfortable chair in a quiet room. Threshold for a1000-Hz tone was determined with a Maico MA25 audiometer(Minneapolis, MN) using the method of limits. All 4 subjects' hearingwas within the normal range, and each stated that he could hear the80-dB tone CS very clearly. While the subject was looking straightahead, the vertical distance between the upper and lower eyelids wasmeasured in millimeters with a ruler to calibrate the eyeblink ampli-tude for data analyses. A silent videotape was played, and the subjectwas told to relax and watch the movie. Eyeblink rate was measured bycounting the number of blinks for a 2-min period. Next, the adjustableheadband holding the airpuff jet and infrared eyeblink monitor wasplaced on the subject's head and adjusted to record eyeblinks. Five to10 US-alone presentations were given to adjust the infrared eyeblinkmonitor and to adjust airpuff intensity (5-10 psi) to yield a completeeyeblink. At this point, the examiner gave neutral instructions for theconditioning session. Subjects were asked to remain alert and attend tothe tones; they were told the following: "Please make yourselfcomfortable and relax. From time to time you will hear some tones andfeel some mild puffs of air in your eye. If you feel like blinking, pleasedo so. Just let your natural reactions take over." At that point theexperimenter answered any questions the subjects had and theninitiated the classical conditioning trials.

The subjects viewed silent videos while the classical conditioningprocedure was carried out. For H.M., three videos were alternatedthroughout the 21-session study, and two of these were the samevideos shown to the PE subject. NCS2 saw two videos, one of whichwas the same for H.M. and the PE subject. A number of videos wereused for NCS1, who was conditioned for 18 sessions. NCSl's interest inthese videos varied, but the other 3 subjects were about equallydistracted by each video. After each conditioning session, subjectswere asked 12 standardized questions assessing their awareness of theCS-US contingency.

For the delay paradigm, CS duration was 500 ms. The US onsetoccurred 400 ms after CS onset; the US coterminated with the CS. Forthe trace paradigm, CS duration was 400 ms followed by a 500-msblank period. This was followed by a 100-ms corneal airpuff US. TheCS-US interval for the trace paradigm was 900 ms, whereas the CS-USinterval for the delay paradigm was 400 ms. In both paradigms theintertrial interval (ITI) was random between 10 and 20 s. Acquisitionsessions consisted of 90 trials and lasted about 45 min. Each sessioncontained 10 blocks of 9 trials. The first trial of every block was aCS-alone trial and the next 8 trials were paired tone-CS and air-puff-US trials. Each 90-trial session consisted of 80 paired CS-US trialsand 10 CS-alone trials.

The size of each subject's eyelid closure distance established thecriterion size for a CR. A CR was scored in the delay paradigm whenan eyeblink '/aith the magnitude of the total blink magnitude (0.5 mmfor an average eye of 10 mm) or greater occurred in the interval

between 100 and 400 ms after the onset of the CS. Eyeblinks with alatency of 0-100ms were not scored as CRs because they were thoughtto be reflexive or "alpha" responses to the tone (see Gormezano, 1966)and therefore not representative of associative learning. In the traceparadigm, a CR was scored for an eyeblink of 'Aith of the total bl inkmagnitude or greater in the interval 100-900 ms after the CS onset. Inaddition to these scoring conventions, responding in the first 100 msafter the CS onset was scored in the trace paradigm because hippocam-pectomized rabbits have shorter latency responses in this paradigm(Port et al., 1986; Solomon et al., 1986).

Experimental Design

The experiment with H.M. and NCS1 took place over a period of 8weeks. In Week 1, H.M. received eight sessions and NCS1 receivedseven sessions of 90-trial, 45-min conditioning sessions with the 400 msCS-US interval delay paradigm. Sessions 1 and 2 for H.M. and Sessions1-3 for NCS1 were given on consecutive days, and the subsequentdelay sessions were given in the morning and afternoon of the sameday. Five weeks after the last delay session, H.M. and NCS1 receivedfive sessions with the trace paradigm. Sessions 1-4 occurred in themorning and afternoon of 2 consecutive days, and Trace Session 5occurred 36 hr later. After a 24-hr interval (5 weeks and 4 days afterthe last session in the delay paradigm), training in the initial 400-msCS-US interval delay paradigm proceeded for six sessions. Sessions9-12 occurred in the morning and afternoon of 2 consecutive days, andSessions 13 and 14 occurred in the morning of the next 2 consecutivedays. Two years after the classical conditioning experiment, H.M. wasconditioned for two 90-trial sessions on 2 consecutive days in the900-ms CS-US interval trace paradigm.

The PE subject and NCS2 received the 400-ms CS-US interval delayparadigm for two 90-trial sessions on 2 consecutive days. Five weekslater they received the 900-ms CS-US trace interval for two 90-trialsessions on 2 consecutive davs.

Results

Delay Paradigm: H.M.

Percentage of CRs. H.M. produced 5% of CRs in pairedCS-US trials in the first session. In the initial sessions in thedelay conditioning paradigm with the percentage of CRs inpaired CS-US trials used as the measure, H.M. gave littleevidence of conditioning (see Figure 1). The mean percentageof CRs in paired trials for normal adult subjects in H.M.'s agerange and older in the first session of the 400-ms CS-USinterval delay paradigm was 32% (SD = 23; Woodruff-Pak &Thompson, 1988) and for subjects ages 60-69 years, about 56%(Solomon, Pomerleau, Bennett, James, & Morse, 1989). It wasnot until Session 6 that H.M. produced a percentage of CRs inpaired trials that was normal for his age (34%). Throughoutthe six sessions of the 5-week delay retest, H.M. producedsufficient CRs in paired trials (16% to 63%) to remain withinone standard deviation of the mean achieved by normal olderadults in one 108-trial session.

When the CS and US are paired, subjects have to produce aresponse within 400 ms for it to count as a CR. However, in theCS-alone trials, in which there is no airpuff US, a response wasscored as a CR in the present study if it fell within the 1,200-msepoch. On CS-alone trials, when H.M. was given a longer timeperiod than 400 ms to respond to the tone CS, he gave muchearlier evidence of classical conditioning, producing almost


100(0<D<ncoD.OT

o:TJo>co

'•6coO>4—

o0)

Dc0)o0)

Q_

90 -

80 •

70

60

50

40

30

20

10

0

o Retest T/C=276

-fr+

Q)9)

m

1 2 3 4 5 6 7 8 9 10 11 12 13 14Sessions (90 Trials Total)

DELAY CLASSICAL CONDITIONING PARADIGM

• CRs in Paired CS-US Trials * CRs in CS-Alone Trials

Figure 1. Percentage of conditioned responses (CRs) in the paired tone conditioned stimulus (CS) andcorneal airpuff unconditioned stimulus (US) trials and in the tone CS-alone trials for H.M. in the 400-msCS-US interval delay paradigm. There were 80 paired CS-US trials and 10 CS-alone trials per session. Thefirst eight sessions were followed by a 5-week period without classical conditioning. Sessions 9-14 wereinitiated 5 weeks and 4 days after Session 8 in the first series of conditioning in the delay paradigm and 24hr after five sessions of training in the trace classical conditioning paradigm. Trials to learning criterion(T/C) were the number of trials required to produce eight CRs in 9 consecutive trials.

60% of CRs by Session 3 and more than 85% of CRs by Session5 (see Figure 1). On the first session of the 5-week delayparadigm retest, H.M.'s CR level on the CS-alone trials was at50%, and CR percentage on the paired CS-US trials onlybegan to approach the level of CRs in the CS-alone trials inSessions 12 and 13.

For normal subjects, there is less of a discrepancy betweenpercentage of CRs in paired CS-US trials and CS-alone trialsthan there was for H.M. Woodruff-Pak and Thompson (1988)reported the percentage of CRs for the CS-alone trials fornormal adult subjects ages 60 to 83 years to be 35% (SD = 28).These are the same subjects who produced 32% of CRs inCS-US trials. Solomon, Pomerleau, Bennett, James, and Morse(1989) did not report the actual CS-alone percentage CRvalues for individual age groups, but these investigators re-ported a correlation of .87 between percentage of CRs forCS-US and CS-alone trials.

Response latency. H.M. is characteristically slow to respond(Corkin, 1968). His difficulty in producing an eyeblink within400 ms apparently delayed his acquisition of CRs in pairedCS-US trials (see Figure 2). In the first eight sessions in thedelay paradigm, H.M. never had a mean eyeblink responseshorter than the CS-US interval of 400 ms. It was not untilSession 12 that he produced an average response faster than

US onset (379.6 ms). He maintained this faster latency forSession 13 (383.9 ms), but in Session 14 his average responseslowed to 423.0 ms.

Trials to learning criterion. Another means of measuringacquisition is to tabulate the number of conditioning trials ittakes subjects to present a series of CRs (see Table 2). Thecriterion used in many laboratories is 8 CRs in nine consecu-tive trials. Trials to criterion is the number of trials until thefirst trial of a string of 8 CRs in nine consecutive trials.Whereas 70% of normal subjects ages 60-69 attained learningcriterion within the first session (Solomon, Pomerleau, Ben-nett, James, & Morse, 1989), H.M. did not attain this learningcriterion until Session 6. Beginning on Trial 473, H.M. pro-duced 10 consecutive CRs and attained criterion. When he wasretested in the delay paradigm after a 5-week interval, H.M.demonstrated considerable savings. He attained learning crite-rion in 276 trials.

Delay Paradigm: All Subjects

Percentage of CRs. Using the measure percentage of CRsin the paired CS-US trials, H.M.'s acquisition in the 400-msCS-US delay paradigm was slow compared with NCS1, NCS2,and the PE subject (see top of Figure 3). NCS1 produced 50%


1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4

Sessions (80 Paired Trials)


Figure 2. Mean latency of eyeblink response in the paired tone conditioned stimulus (CS) and cornealairpuff unconditioned stimulus (US) trials for H.M. in the 400-ms CS-US interval delay paradigm. Therewere 80 paired trials in each session. Latency was scored as the time between CS onset and a blink of 0.53mm or greater. US onset is indicated at 400 ms.

of CRs in Session 1 in paired trials, thus falling within therange of normal age-matched subjects. NCS2 had the highestpercentage of CRs in Session 1 (76%), thus exceeding thenormal range for his age group. The PE subject was at the lowend of performance of normal subjects for his age range inSession 1 on percentage of CRs in paired CS-US trials (13%),but he improved to 44% of CRs in Session 2, matching theperformance of NCS1. The PE subject's percentage of CRs inCS-alone trials was 33% in Session 1 and 88% in Session 2 (seebottom of Figure 3). His performance on CS-alone trialsactually equaled the above average performance of NCS2 inSession 2. The PE subject's performance on CS-alone trials inSession 1 was comparable to the percentage of CRs in

Table 2Number of Trials to Learning Criterion of Eight ConditionedResponses in Nine Consecutive Trials by 2 Amnesic Subjectsand 2 Normal Control Subjects (NCSs)

Subject

H.M.NCS1PENCS2

Delayparadigm

473315119

15

Delayparadigm

retest

27691

——

Traceparadigm

911

161

Traceparadigm

2-year retest

9———

Note. H.M. is identified by his initials, whereas the PE (postencepha-litis) subject is identified by the etiology of his amnesia. Dashesindicate that no retest was performed.

CS-alone trials of normal adults ages 50-59 years, which was34% (Woodruff-Pak & Thompson, 1988).

Response latency. Mean response latency for H.M. andNCS1, plus and minus one standard deviation of their respec-tive mean response latencies for 14 sessions are shown inFigure 4, along with the means and standard deviations for thePE subject and NCS2 in two sessions. H.M. produced eye-blinks within one standard deviation of NCSl's response.However, NCSl's mean response was, on average, occurringbefore the US onset and was thus a CR. H.M.'s mean responseoccurred after US onset and constituted a UR. H.M.'s meanresponse latency shortened over sessions until it was actuallymore rapid than NCSl's response latency in Sessions 11, 12,and 13. Only on Sessions 12 and 13 did H.M.'s averageresponse attain the latency of a CR.

The PE subject's responses were just slightly slower than onestandard deviation of NCS2's mean response latency. The PEsubject had a mean latency occurring after US onset in the URperiod, whereas NCS2 had an average response latency thatwas shorter than US onset and occurred in the CR period.

Trials to learning criterion. H.M. attained learning criterionin 473 trials. The results for trials to learning criterion aresummarized in Table 2. NCS1 attained a criterion of eight CRsin 9 consecutive trials in Session 4 after 315 trials. The PEsubject and NCS2 attained learning criterion in the delayparadigm much more rapidly than H.M. and NCS1. The PEsubject reached criterion on Trial 119, within the normal rangefor his age. NCS2, making criterion in 15 trials, attained


o0)

o-I-'co

100

80

60

40

20

0

.£

I

2 3 4 5 6 7 8 9 10 11 12 13 14


oo>o"co

0.

100

80

60

40

20

0

•o•«•+

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Sessions of 10 CS—alone Trials

DELAY CLASSICAL CONDITIONING PARADIGM• H.M. O NCS1 • PE a NCS2

Figure 3. Percentage of conditioned responses (CRs) in paired conditioned stimulus-unconditionedstimulus (CS-US) trials (top) and CS-alone trials (bottom) for 2 amnesic subjects (H.M. and thepostencephelatic [PE] subject) and their normal control subjects (NCS1 and NCS2) in the delay paradigm.There were 80 CS-US paired trials and 10 CS-alone trials in each session.

criterion more rapidly than the average 20-year-old subject(Woodruff-Pak & Thompson, 1988).

UR assessment. The UR is a measure of the subject'sreflexive ability to blink. If the UR is abnormal, valid assess-ment of associative learning in the EBCC paradigm cannot bemade. For the 2 amnesic subjects as well as the NCSs, the blinkreflex itself, as assessed by UR amplitude and responselatency, was within the range of normal adult subjects ofcomparable age reported by Solomon, Pomerleau, Bennett,James, and Morse (1989) and Woodruff-Pak and Thompson(1988). The amnesic subjects' blink rates were low: 1 blink perminute for H.M. and 2 blinks per minute for the PE subject.NCS1 had a relatively high blink rate of 25 blinks per minute,and NCS2 had a blink rate of 17 blinks per minute. Solomon,Pomerleau, Bennett, James, and Morse (1989) reported aspontaneous blink rate of about 16 blinks per minute forsubjects in the 50-59-year-old age group and 17 blinks perminute for subjects in the 60-69-year-old age group.

An important control for performance factors in the classi-cal conditioning paradigm is the amplitude of the UR. Ameasure of UR amplitude is computed for every trial on whichthere is an airpuff. UR amplitude is the peak amplitudeoccurring in the UR period 400-1,200 ms after CS onset.

Subjects with low-amplitude URs (less than 3.5 mm), who maynot be alert during the conditioning session, condition signifi-cantly less well than subjects with higher amplitude URs(Woodruff-Pak & Thompson, 1988).

The raw score for UR amplitude was different for eachindividual because the vertical measurement between their topand bottom eyelids ranged from 8 mm to 12 mm: 8 mm forNCS2, 10.5 mm for H.M., 11 mm for the PE subject, and 12mm for NCS1. Figure 5 illustrates UR amplitude for the 4subjects over all classical conditioning sessions. In the delayparadigm, mean UR amplitude did not drop below 3.5 mm forany subject. UR amplitude was generally high for all 4 subjectsand exhibited stability over sessions, at least in the initial delaytraining. In the six sessions of delay retesting, NCS1 exhibitedextreme variability in UR amplitude as compared with H.M.

Trace Paradigm

Percentage CRs. All 4 subjects produced a higher percent-age of CRs in the initial session in the trace paradigm than theyhad in the initial session of the delay paradigm, even thoughthere was a 5-week interval between the last session in delayconditioning and the first session in trace conditioning (see


600 r

8 50°

~, 400

o300

8 20°co

8a:

US Onset

1 2 3 4 5 6 7 8 9 10 11 12 13 14



• H.M. O NCS1 • PE a NCS2

Figure 4. Response latencies and standard deviations of H.M. and Normal Control Subject 1 (NCS1) in14 sessions and the postencephelatic (PE) subject and Normal Control Subject 2 (NCS2) in two sessions inthe delay paradigm. US = unconditioned stimulus.

Figure 6). Averaging percentage of CRs in the paired CS-UStrials over the five sessions in the trace paradigm, H.M.produced about the same percentage of CRs as NCS1 (44%and 49%, respectively). Two years after the initial EBCC,H.M. produced as high a CR percentage in the paired CS-UStrials as the highest percentage produced by NCS1.

In the two sessions in the trace paradigm in which the PEsubject was tested, he produced an average of 46% of CRs.This performance was comparable to the performance ofNCS1, although it was lower than the extraordinary perfor-mance of NCS2 (90%). There are no published data in the900-ms CS-US trace paradigm on subjects in NCS2's agerange, but Finkbiner and Woodruff-Pak (1991) reported on 10subjects in the 18-30-year age range in that paradigm. Youngsubjects produced 62% of CRs (SD = 23). Thus, NCS2'sperformance in the trace paradigm exceeds one standarddeviation of the performance of young adults. The perfor-mance of H.M., NCS1, and the PE subject fall within onestandard deviation of the young adults' performance in thetrace paradigm.

Percentage of CRs in the CS-alone trials were higher than inthe paired CS-US trials in the initial sessions for H.M. (seeFigure 6). However, given the longer CS-US interval, percent-age of CRs in the two types of trials became about equal muchsooner in training in the trace paradigm than in the delayparadigm. At the 2-year follow-up testing, percentage of CRsin the two types of trials was about equal from the first session.As indicated in Figure 6, NCS1 also had better initial perfor-

mance in the CS-alone trials, but the PE subject and NCS2produced about as many CRs in the paired CS-US trials as inthe CS-alone trials.

Response latency. The 900-ms CS-US interval in the traceEBCC paradigm provided H.M. with a longer period in whichto produce an eyeblink before the US onset. H.M. achieved aresponse latency shorter than US onset early in the traceconditioning sessions (see Figure 7). His mean responselatency in Session 2 of trace conditioning was 702 ms. In the2-year follow-up conditioning sessions in the trace paradigm,mean response latency was 599 ms in Session 1 and 623 ms inSession 2. H.M. required 12 sessions in the 400-ms CS-USinterval in the delay paradigm to achieve a mean latencyshorter than the 400-ms US onset latency.

Comparing the amnesic subjects' response latency in thetrace paradigm with that of NCSs, both H.M. and the PEsubject responded within one standard deviation of the meanresponse of the NCSs. From Figure 7 it can be seen that themean response latencies of amnesic subjects and their normalcontrols were similar in the trace paradigm.

Trials to learning criterion. Five weeks after being tested inthe delay paradigm, H.M. attained criterion in the traceparadigm in 91 trials. Two years after the first classicalconditioning experiment, H.M. attained criterion in the traceparadigm in 9 trials. His initial acquisition was the slowest ofthe 4 subjects (see Table 2). NCS1 and NCS2 attained criterionin the trace paradigm on the first trial. The PE subject attainedcriterion in 16 trials. Young adults took a mean of 19 trials to


E 14

0> 12

8COcoQ.(0

-o0)

I 2TJ

I °DELAY TRACE

Sessions (80 Paired Trials)DELAY RETEST

• H.M. O NCS1 PE a NCS2

Figure 5. Amplitude of the unconditioned response for each session in the delay and trace paradigm foramnesic subjects (H.M. and the postencephelatic [PE] subject) and Normal Control Subjects 1 and 2(NCS1 and NCS2).

attain criterion in the 900-ms CS-US trace paradigm (Fink-biner & Woodruff-Pak, 1991).

Short latency responses. The number of responses occurring0-99 ms or 100-199 ms after CS onset in each session for all 4

subjects in the trace paradigm were counted. In the first fivesessions in the trace paradigm, H.M. had no responses ineither short latency category. In the 2-year follow-up in thetrace paradigm, he had one response of a 116-ms latency in

0 1 2 3 4 5 6 7

Sessions of 80 Paired Trials

1 2 3 4 5 6 7

Sessions of 10 CS—alone Trials

TRACE CLASSICAL CONDITIONING PARADIGM

H.M. O NCS1 • PE a NCS2

«Q.

Figure 6. Percentage of conditioned responses (CRs) for amnesic subjects (H.M. and the postencephe-latic [PE] subject) and Normal Control Subjects 1 and 2 (NCS1 and NCS2) in the trace classicalconditioning paradigm. Left panel: Percentage of CRs in the paired conditioned stimulus-unconditionedstimulus trials. Right panel: Percentage of CRs in the conditioned stimulus-alone (CS-alone) trials.Sessions 6 and 7 were conducted with H.M. 2 years after the initial classical eyeblink conditioning sessions.


O0>0)

E

1000

900

800

700

ft 600

t> 50°_l

v 400in

O 300o>r£ 200

100

o

US Onset

1 2 3 4 5 6

Sessions of 80 Paired Trials

TRACE CLASSICAL CONDITIONING PARADIGM

• H.M. O NCS1 PE a NCS2

Figure 7. Mean response latency for amnesic patients (H.M. and the postencephelatic [PE] subject) andNormal Control Subjects 1 and 2 (NCS1 and NCS2) in the trace classical conditioning paradigm. Sessions6 and 7 were conducted with H.M. 2 years after the initial classical eyeblink conditioning sessions.Unconditioned stimulus (US) onset was 900 ms as indicated.

Session 6 and no short-latency responses in Session 7. The PEsubject had four responses in the 100-199 ms range in Session1 and none in Session 2. NCS1 had the most short-latencyresponses: In the 0-99 ms category he had 1, 2, 4, 12, and 7 inSessions 1-5; in the 100-199 ms category he had 30, 2, 17, 11,and 39 in Sessions 1-5. NCS2 had no responses in the 0-99 mscategory. In the 100-199 ms category, he had 1 response inSession 1 and none in Session 2.

UR amplitude. In the trace paradigm, UR amplitude forthe 4 subjects was normal and was comparable to UR ampli-tude in the delay paradigm (see Figure 5). All subjectsexceeded a mean UR amplitude of 3.5 mm on each condition-ing day, with the exception of NCS1, who had a mean URamplitude of 3.14 mm on Session 2. On the other sessions,NCSl's mean UR amplitude exceeded 6 mm.

Measures of Explicit Memory

Recall of previous classical conditioning sessions. H.M. didnot recall previous conditioning sessions. In each session, H.M.reported that he had not worn the conditioning apparatusbefore. For example, at the beginning of Session 7 of delaytraining, we asked him whether he remembered the headbandwe had put on his head at each session, and we showed it tohim. His response was, "No, I don't." H.M. had no recollection

of the experimenters, apparatus, instructions, or procedure,and he responded to each session as if it were novel to him.

When the PE subject returned for Session 2 of training inthe delay paradigm, he did not recall the EBCC session the daybefore. When asked whether he remembered what he haddone with the experimenters the day before, he said that he didnot. However, when prompted by an experimenter's mention-ing of tones and airpuffs, he said, "Oh, water in my eye. Ithought it was water." The day before he had repeatedly statedthat we were squirting water in his eye whenever the airpuff hithis eye.

Insight about the CS-US contingency. Spontaneous remarksof the PE subject indicated that he did have insight into thetone-airpuff contingency. This insight was demonstrated earlyin training, on Trial 19 of Session 1. This was a CS-alone trial towhich he responded, "Why is there no water? Usually whenthe buzzer goes off, it shoots water." On Trial 28, which wasalso a CS-alone trial, he said, "Come on, I expect some waterwhen the buzzer goes off."

To assess memory for the contingency between the tone andthe airpuff, the experimenter asked subjects to complete amultiple-choice questionnaire at the end of each testingsession. Examples of subjects' responses to selected questionsasked after conditioning in the initial sessions (seven for H.M.


Table 3Selected Responses from Postconditioning Questionnaire on Awareness of the ConditionedStimulus-Unconditioned Stimulus Contingency by 2 Amnesic Subjects and 2 NormalControl Subjects (NCSs)

Sample question/subject

When the tone came on,what did you do?

H.M.NCS1PENCS2

Which came first: thetone or the airpuff?

H.M.NCS1PENCS2

When did you realize the tonecame before the airpuff?

H.M.NCS1PENCS2

Type and number of responses

Nothing

3

Tonealways"

1622

Beginning"

14

2

Raisedhand

1

Toneusually

11

Halfwaythrough3

111

Closedeyes3

2

2

Airpuffusually

3

Towardend

3

Pressedkey

1

Airpuffalways

2

Never

2

Adlib

7"2c

Adlib

—

Adlib

2d

le

Note. H.M. is identified by his initials, whereas the PE (postencephalitis) subject is identified by theetiology of his amnesia. Data reported are for the first seven sessions in the delay paradigm for H.M. andNCS1 and the first two sessions in the delay paradigm for the PE subject and NCS2. Dashes indicate that aparticular response does not apply for the subject being tested."Indicates an appropriate response. ""'Sometimes I blinked." c"Waited for airpuff." d"From pastsession." c"After 2 hours of movie."

and NCS1, two for the PE subject and NCS2) in the delayparadigm are presented in Table 3. On the basis of thesemeasures, only H.M. lacked explicit memory for the CS-UScontingency. When asked whether the tone or the airpuff camefirst, he most frequently said that the airpuff always came first.He consistently said "No" to the question, "Did you realizethat the tone came before the airpuff?" In contrast to H.M.,the PE subject clearly recognized the CS-US contingency.After Session 1 in the delay paradigm, when the PE subject wasasked what he did when the tone came on, he said, "At firstnothing; then I got used to expecting the puff of air." To thissame question asked at the end of delay Session 2, the PEsubject said, "I waited for the puff." Both NCSs recognizedthat the tone always came on before the airpuff. When askedwhether he made himself blink at the tone when he realizedthe tone was followed by an airpuff, NCS1 said, "Sometimes atthe tone I would blink, but it was more like a reflex." Thissubject said repeatedly that he was not trying to blink to thetone, but a blink just occurred.

Discussion

To extend knowledge about the scope of implicit memorycapacities in amnesia, the delay and trace paradigms of EBCCwere administered to 2 amnesic subjects (H.M. and the PEsubject) and to their respective NCSs (NCS1 and NCS2). Theresults add to the growing body of research demonstrating thatthe brain is composed of separable neural systems thatsubserve different components of memory. Previous research

indicated that amnesic subjects with medial-temporal-lobelesions are able to acquire CRs in the delay paradigm (Daumet al., 1989, 1991; Weiskrantz & Warrington, 1979). We havedemonstrated that amnesic subjects with medial-temporal-lobe lesions are able to acquire CRs in the 400-ms delay and900-ms trace paradigms of EBCC. With H.M., it has beendemonstrated that EBCC occurs in these delay and traceparadigms in the almost complete absence of the hippocam-pus. Parametric studies of a variety of delay and traceparadigms might identify some stimulus conditions in whichhippocampectomized subjects would be unsuccessful in acquir-ing conditioned eyeblink responses. However, it is concludedthat the memory system subserving explicit memory is notessential for EBCC, at a minimum in the 400-ms delay and900-ms trace paradigms administered in this study.

Amnesics' Acquisition in the Delay Paradigm

H.M. acquired CRs to the tone CS in the delay paradigm,but his acquisition was slow in the paired trials at the 400-msCS-US interval. In the first five classical conditioning sessions,H.M. produced an average 8% of CRs compared with theaverage of 46% of CRs for NCS1. The PE subject produced anaverage of 28% of CRs in the two sessions in which he wasconditioned in the delay paradigm. The performance of NCS1and the PE subject are within one standard deviation of thepercentage of CRs for their respective age groups reported bySolomon, Pomerleau, Bennett, James, and Morse (1989) andWoodruff-Pak and Thompson (1988). H.M.'s performance in


the first five delay paradigm conditioning sessions was morethan one standard deviation below the mean performance ofolder adults, and NCS2's performance (mean of 95% CRs overtwo sessions) was greater than two standard deviations abovethe performance of older adults in Woodruff-Pak andThompson's (1988) study.

For H.M. to attain learning criterion, 473 trials wererequired compared with the 315 trials to learning criterion forNCS1. The PE subject required 119 trials to reach learningcriterion, and NCS2 needed 15 trials. To our knowledge, thereare no published data on classical conditioning in older adultswith more than one session of conditioning. Thus, it is difficultto estimate the number of trials to learning criterion that aretypical for adults ages 50-69. In 108 trials, 56% of subjects ages50-59 years and 84% of subjects ages 60-83 years did not attaina learning criterion of eight CRs in 9 consecutive trials(Woodruff-Pak & Thompson, 1988). In a recently completeddoctoral dissertation study, 20 normal older adults with a meanage of 74 years were classically conditioned for two 90-trialsessions in the 400-ms CS-US interval delay paradigm (Fink-biner, 1993). Only 25% of those subjects attained the learningcriterion in 180 trials. Because 56% of adults ages 50-59 didnot attain learning criterion in 108 trials in Woodruff-Pak andThompson's study and 75% of adults ages 63-80 did not attainlearning criterion in 180 trials in Finkbiner's study, it appearsthat classical conditioning in the delay paradigm was normal inthe PE subject with bilateral hippocampal lesions, who at-tained criterion in the delay paradigm in 119 trials. Thedependent measures of trials to criterion and percentage ofCRs in the paired trials suggest that conditioning in H.M. inthe delay paradigm was abnormal for his age group.

In the initial conditioning sessions, H.M. required more than400 ms to blink to the tone CS, as evidenced by his perfor-mance in the CS-alone trials. Although H.M.'s mean responselatency was within one standard deviation of the mean re-sponse latency of NCS1, H.M. typically did not respond withinthe CR latency period in advance of the US. This difficulty inproducing an eyeblink within 400 ms apparently obscuredH.M.'s acquisition of CRs in the CS-US paired condition, asmeasured by percentage of CRs occurring within 400 ms.When H.M. was given a longer time period to respond to thetone CS in the CS-alone trials, he showed a higher percentageof CRs than when required to respond rapidly in CS-US pairedtrials. H.M. had a low blink rate of one blink per minute;spontaneous blinks in the 1,200 CS-alone period are thereforenot a likely explanation of his better performance in the longerCS-alone period. Using the dependent measure—percentageof CRs in the CS-alone trials—H.M. achieved a criterion closeto 60% of CRs in Session 3 or within 270 trials. Evaluation ofH.M. with the CS-alone measure yields the interpretation thathis conditioning in the delay paradigm is relatively normal.

Performance of NCS1 was quite variable, especially in latersessions. Variability is apparent in divergent measures of CRpercentage (see Figure 3) and in session-to-session variabilityin UR amplitude (see Figure 4). The seeming decline in theperformance of NCS1 could be attributed to boredom. NCS1found the repetitive classical conditioning testing quite te-dious. Human subjects typically participate in classical condi-tioning studies for one session. Data for NCS1 on the retest

sessions indicate why repeated measures studies of EBCC areseldom carried out. NCS1 was generous with his time, alwayspolite, and willing to comply with all instructions. However, hisdisinterest in a sequence of 18 classical conditioning sessionscontrasts with H.M.'s motivated performance. H.M. came toeach session eager to participate, and he told us on severaloccasions that he was pleased to help with this research.Classical conditioning sessions broke the normal tedium ofH.M.'s routine, whereas they introduced tedium into the moreactive life of NCS1.

The second series of conditioning sessions in the delayparadigm occurred after a 5-week interval and 24 hr after fiveconditioning sessions in the trace classical conditioning para-digm. At that time, H.M. attained learning criterion in 276trials, showing a savings of 197 trials. NCS1 attained learningcriterion in the second delay series in 91 trials, showing asavings of 224 trials. As measured by savings in the delayparadigm, H.M.'s performance was similar to NCSl's. Thus,savings is a second measure of acquisition in the delayparadigm on which H.M. scored in the normal range.

Amnesics'Acquisition in the Trace Paradigm

Using percentage of CRs, response latency, and trials tocriterion to assess performance in the trace classical condition-ing paradigm, H.M. and the PE subject acquired CRs. Bothamnesic subjects learned to make a CR in at least eight out ofnine consecutive trials, thus suggesting that the hippocampusis not critical for trace classical conditioning in humans.Percentage of CRs for the amnesic subjects was within astandard deviation of percentage of CRs for young adultsubjects. Amnesic subjects' response latency was within astandard deviation of the NCSs' response latency. Neithersubject with hippocampal lesions produced many short-latencyresponses. NCS1 produced more short-latency responses thandid the other subjects, and NCS2 produced a greater percent-age of CRs than did normal young adults on this task. Theseresults reflect the range of individual differences in humanperformance on EBCC.

An even more convincing demonstration of acquisition inthe trace paradigm in the almost complete absence of thehippocampus occurred when H.M. attained learning criterionin nine trials 2 years after the initial conditioning experiment.In a 1-year follow-up study with normal elderly subjects, weobserved that reacquisition of EBCC in the delay paradigmoccurred with no evidence of retention (Ferrante & Woodruff-Pak, 1990). Although he had no exposure to EBCC for 2 years,H.M. produced 67% of CRs in Session 1 and 80% of CRs inSession 2 in the 900-ms CS-US interval trace paradigm. Theincrease in US intensity from 7 psi in the initial sessions of thetrace paradigm to 7 to 14 psi in the 2-year follow-up sessionsmay have contributed to more rapid acquisition in the fol-low-up session. US intensity has been shown to accelerate rateof conditioning (Gormezano & Moore, 1962; Prokasy, Grant,& Myers, 1958; Spence & Taylor, 1951). At this higher USintensity, H.M. exhibited rapid conditioning in the traceparadigm, demonstrating that the hippocampus is not essentialfor classical conditioning in the trace paradigm.


The PE subject also showed dramatic acquisition in the traceparadigm and achieved learning criterion on the 16th trial,demonstrating normal performance in the trace paradigm in asubject with bilateral hippocampal lesions.

Explicit Memory

Similar to his performance on measures of skill learning,H.M. demonstrated acquisition of conditioned eyeblink re-sponses, but he had no explicit memory of the training.Postsession interviews with H.M. revealed that he did notremember previous conditioning sessions and never had anyinsight about the relationship between the tone and airpuff. Incontrast, NCS1 demonstrated his understanding of the tone-airpuff contingency after training on Session 1. NCS1 told usthat he blinked to the tone but that he did not think he did soon purpose. This description of responding in the EBCCexperiment characterizes the implicit nature of the task. NCS1expressed the view that he was not aware of thinking aboutblinking to the tone: It just happened. He did not remember toblink, but he found himself blinking when he heard the tone.

NCS1, NCS2, and the PE subject each demonstrated anawareness of the CS-US contingency, but H.M. did not. H.M.also conditioned more poorly than the other subjects. It ispossible that knowledge about the CS-US contingency affectsEBCC, but it is not clear that this awareness was the majorreason for H.M.'s poorer performance. NCS1 articulated thathe was not trying to blink to the tone. On the CS-alone trial,when the PE subject said, "Come on, I expected some waterwhen the buzzer goes off," he did not produce a CR. Althoughhe knew that the airpuff was coming, he did not blink. Theissue of voluntary responding received research attentionduring the late 1950s and into the 1960s, and means ofdetermining subjects who were voluntary responders weredeveloped. On the basis of response latency (see Figures 4 and7), blink rate, and subjects' statements about their perfor-mance, it is concluded that NCS1, NCS2, and the PE subjectwere not purposely blinking to the tone. Rather, they blinkedreflexively to the tone CS after associative learning took place.

Neural Substrates of Eyeblink Conditioning

H.M. was not a "pure" case with whom to test the effect ofbilateral hippocampal removal on EBCC because he hadatrophy of the cerebellar vermis and hemispheres. In rabbits,cerebellar cortical lesions ipsilateral to the conditioned eyeincrease the number of conditioning trials required for acquisi-tion by almost seven times (Lavond & Steinmetz, 1989).Cerebellar cortical tissue loss may have increased the numberof conditioning trials required for H.M. to attain learningcriterion.

There was no lesion control subject with bilateral removal ofthe hippocampus and no cerebellar damage. The controlsubject for cerebellar damage was a 54-year-old amnesicsubject who had contracted herpes simplex encephalitis andhad bilateral hippocampal lesions. Herpes simplex encephali-tis damages CA1 pyramidal cells. Hippocampal pyramidal cellsin rabbits are the cells that model the behavioral CR and UR(Berger & Thompson, 1978a; Berger et al., 1983). The PE

subject conditioned more rapidly than H.M. in both the delayand trace paradigms. The difference in conditioning perfor-mance between the 2 amnesic subjects may be the result ofcerebellar cortical lesions in H.M.

Cerebellar cortex may be important for the timing ofmovement, particularly when intervals of less than 0.5 s areinvolved. According to Brooks (1986), the cerebellar cortexhelps to program the relative timing and intensity of the actionof muscles in a response. Keele and Ivry (1990) hypothesizedthat cerebellar cortex provides a critical computation of timingcapability required in the performance of a variety of tasks,including EBCC. From this perspective, the cerebellar cortexprovides the necessary temporal computation for the produc-tion of CRs. Damage to H.M.'s cerebellar cortex may accountfor the poor timing of his responses, especially his difficulty inproducing CRs in the 400-ms CS-US delay interval on thepaired trials.

In the case of the trace conditioning paradigm with its longCS-US interval, it has been observed in hippocampectomizedrabbits that the timing of CRs is abnormal. Rabbits without ahippocampus produce short-latency CRs in a 750-ms CS-USinterval trace paradigm (Port et al., 1986; Solomon et al.,1986). The hippocampectomized rabbits' eyes are open by thetime the US comes on. In this sense the CRs are nonadaptivebecause they do not eliminate any of the force of the airpuff.Moyer et al. (1990) observed a few short-latency CRs, but theysaw fewer CRs of any kind in their hippocampectomizedrabbits in a trace paradigm in which the CS was on for 100 msand in which the US onset was 600 ms after CS onset.

We analyzed the latency of CRs in H.M. and the PE subjectand found that they had fewer short-latency responses thanNCS1 and about the same number as NCS2. Two years afterthe initial EBCC experiment, H.M. had only one short-latencyresponse in Session 1 and no short-latency responses inSession 2 in the trace paradigm. In contrast to hippocampecto-mized rabbits that have short-latency, nonadaptive CRs (Portet al., 1986; Solomon et al., 1986) or very few CRs of any kind(Moyer et al., 1990), these human subjects with bilateralhippocampal lesions had fewer short-latency eyeblinks thandid the NCSs. The human amnesic subjects are different fromhippocampectomized rabbits on many dimensions, one ofwhich is the time lapsing between surgery and classicalconditioning. Rabbits were conditioned 1 to 2 weeks aftersurgery. H.M.'s operation was performed 38 years ago, and thePE subject had his acute episode more than 3 decades ago.Shortly after surgery, before the brain has fully recovered,subjects of any species are more likely to give abnormal re-sponses. On the other hand, in the delay paradigm, rabbitswith bilateral removal of the hippocampus give normal CRswithin several weeks of surgery (Schmaltz & Theios, 1972;Solomon & Moore, 1975). The question of whether EBCC inthe trace paradigm is more vulnerable to hippocampal lesionsthan in the delay paradigm requires additional research. Datafrom H.M. and the PE subject indicate that the hippocampus isnot essential for acquisition and retention in the 900-ms traceparadigm with a 500-ms trace. In different trace paradigms,especially if the trace interval were extended beyond 500 ms,these amnesic subjects might be impaired.


To our knowledge, the present report provides the onlyhuman data on bilateral hippocampectomy in the trace classi-cal conditioning paradigm; these data indicate that acquisitionis possible. Acquisition was also demonstrated in H.M. in thedelay paradigm. The dependent measures of percentage ofCRs in CS-alone trials and savings at the 5-week retestsuggested that H.M.'s EBCC in the delay paradigm wasnormal. However, the dependent measures of CR percentagefor paired CS-US trials, response latency, and trials to learningcriterion suggest that EBCC was abnormal in H.M., apparentlybecause of his slowness to respond. In the longer CS-USinterval trace paradigm, all dependent measures of classicalconditioning were in the normal range for H.M. The PEsubject showed normal EBCC in both the delay and traceparadigms. These results suggest that humans with bilateralmedial-temporal-lobe lesions can acquire the conditionedeyeblink response in the 900-ms trace as well as in the 400-msdelay classical conditioning paradigm. In H.M., this acquisitionoccurred despite almost total bilateral hippocampal removal.

References

Adams, J. H., Corsellis, J. A. N., & Duchen, L. W. (1984). Greenfield'sneuropathology (4th ed). New York: Wiley.

Benzing, W. C, & Squire, L. R. (1989). Preserved learning andmemory in amnesia: Intact adaptation-level effects and learning ofstereoscopic depth. Behavioral Neuroscience, 103, 538-547.

Berger, T. W. (1984). Long-term potentiation of hippocampal synaptictransmission affects rate of behavioral learning. Science, 205, 810-816.

Berger, T. W., Alger, B. E., & Thompson, R. F. (1976). Neuronalsubstrates of classical conditioning in the hippocampus. Science, 192,483-485.

Berger, T. W., Berry, S. D., & Thompson, R. F. (1986). Role of thehippocampus in classical conditioning of aversive and appetitivebehaviors. In R. I. Isaacson & K. H. Pribram (Eds.), The hippocam-pus (Vol. 4, pp. 203-239). New York: Plenum Press.

Berger, T. W., Rinaldi, P., Weisz, D. J., & Thompson, R. F. (1983).Single unit analysis of different hippocampal cell types duringclassical conditioning of the rabbit nictitating membrane response.Journal of Neurophysiology, 59, 1197-1219.

Berger, T. W., & Thompson, R. F. (1978a). Identification of pyramidalcells as the critical elements in hippocampal neuronal plasticityduring learning. Proceedings of the National Academy of Sciences, 75,1572-1576.

Berger, T. W., & Thompson, R. F. (1978b). Neuronal plasticity in thelimbic system during classical conditioning of the rabbit nictitatingmembrane response: I. The hippocampus. Brain Research, 145,323-346.

Brooks, V. B. (1986). The neural basis of motor control. New York:Oxford University Press.

Clark, G. A., McCormick, D. A., Lavond, D. G., & Thompson, R. F.(1984). Effects of lesions of cerebellar nuclei on conditionedbehavioral and hippocampal neuronal responses, Brain Research,291, 125-136.

Cohen, N. J., & Squire, L. R. (1980). Preserved learning and retentionof pattern-analyzing skill in amnesia: Dissociation of "knowinghow" and "knowing that." Science, 210, 207-209.

Corkin, S. (1965). Tactually-guided maze-learning in man: Effects onunilateral cortical excisions and bilateral hippocampal lesions.Neuropsychologia, 3, 339-351.

Corkin, S. (1968). Acquisition of motor skill after bilateral medialtemporal-lobe excision. Neuropsychologia, 6, 255-265.

Corkin, S. (1984). Lasting consequences of bilateral medial temporallobectomy: Clinical course and experimental findings in H.M.Seminars in Neurology, 4, 249-259.

Daum, I., Channon, S., & Canavan, A. G. M. (1989). Classicalconditioning in patients with severe memory problems. Journal ofNeurology, Neurosurgery, and Psychiatry, 52, 47-51.

Daum, I., Channon. S.. Polkey, C. E., & Gray, J. A. (1991). Classicalconditioning after temporal lobe lesions in man: Impairment inconditional discrimination. Behavioral Neuroscience, 105, 396^408.

Ferrante, L. S., & Woodruff-Pak, D. S. (1990). Longitudinal classicalconditioning data in the old-old. Gerontologist, 30, 85A.

Finkbiner, R. G. (1993). Implicit and explicit teaming and memory inhuman aging. Unpublished doctoral dissertation, Temple Univer-sity.

Finkbiner, R. G., & Woodruff-Pak, D. S. (1991). Classical eyeblinkconditioning in adulthood: Effects of age and interstimulus intervalon acquisition in the trace paradigm. Psychology and Aging, 6,109-117.

Gabrieli, J. D. E., Milberg. W., Keane, M. M., & Corkin, S. (1990).Intact priming of patterns despite impaired memory. Neuropsycholo-gia, 28, 417-427.

Gormezano, I. (1966). Classical conditioning. In J. B. Sidowski (Ed.),Experimental methods and instrumentation in psychology (pp. 385-420). New York: McGraw-Hill.

Gormezano, I., & Moore, J. W. (1962). Effects of instructional set andUCS intensity on the latency, percentage, and form of the eyelidresponse. Journal of Experimental Psychology, 63, 487-494.

Graf, P., & Schacter, D. L. (1985). Implicit and explicit memory fornew associations in normal and amnesic subjects. Journal of Experi-mental Psychology: Learning, Memory, and Cognition, 11, 45-53.

Grant, D. A. (1973). Cognitive factors in eyelid conditioning. Psycho-physiology, 10, 75-81.

Harvey, J. A., Gormezano, I., & Cool-Hauser, V. A. (1983). Effects ofscopolamine and methylscopolamine on classical conditioning of therabbit nictitating membrane response. Journal of Pharmacology andExperimental Therapeutics, 225, 42-49.

James, G. O., Hardiman, M. J., & Yeo, C. H. (1987). Hippocampallesions and trace conditioning in the rabbit. Behavioural BrainResearch, 23, 109-116.

Keele, S. W., & Ivry, R. (1990). Does the cerebellum provide acommon computation for diverse tasks? A timing hypothesis. In A.Diamond (Ed.), The development and neural bases of higher cognitivefunctions (Vol. 608, pp. 179-211). New York: New York Academy ofSciences.

Kimble, G. A. (1962). Classical conditioning and the problem ofawareness. Journal of Personality, 30, 27-45.

Lavond, D. G., Hembree, T. L., & Thompson, R. F. (1985). Effect ofkainic acid lesions of the cerebellar interpositus nucleus on eyelidconditioning in the rabbit. Brain Research, 326, 179-182.

Lavond, D. G., & Steinmetz, J. E. (1989). Acquisition of classicalconditioning after removal of cerebellar cortex. Behavioural BrainResearch, 33, 113-164.

Lye, R. H., O'Boyle, D. J., Ramsden, R. T., & Schady, W. (1988).Effects of a unilateral cerebellar lesion on the acquisition ofeye-blink conditioning in man. Journal of Physiology, 403, 58P.

McCormick, D. A., & Thompson, R. F. (1984a). Cerebellum: Essentialinvolvement in the classically conditioned eyelid response. Science,223, 296-299.

McCormick, D. A., & Thompson, R. F. (1984b). Neuronal responsesof the rabbit cerebellum during acquisition and performance of aclassically conditioned nictitating membrane-eyelid response. Jour-nal of Neuroscience, 4, 2811-2822.

Milner, B. (1965). Visually-guided maze learning in man: Effects ofbilateral hippocampal, bilateral frontal, and unilateral cerebrallesions. Neuropsychologia, 3, 317-338.


Milner, B., Corkin, S., & Teuber, H.-L. (1968). Further analysis of thehippocampal amnesic syndrome: 14-year follow up study of H. M.Neuropsychologia, 6, 215-234.

Moore, J. W., Goodell, N. A., & Solomon, P. R. (1976). Centralcholinergic blockage by scopolamine and habituation, classicalconditioning, and latent inhibition of the rabbit's nictitating mem-brane response. Physiological Psychology, 4, 395-399.

Moyer, J. R., Deyo, R. A., & Disterhoft, J. F. (1990). Hippocampec-tomy disrupts trace eye-blink conditioning in rabbits. BehavioralNeuroscience, 104, 241-250.

Norris, E. B., & Grant, D. A. (1948). Eyelid conditioning as affected byverbally induced inhibitory set and counter-reinforcement. Ameri-can Journal of Psychology, 61, 37^9.

Parkin, A. J. (1982). Residual learning capability in organic amnesia.Cortex, 18, 417-440.

Port, R. L., Romano, A. G., Steinmetz, J. E., Mikhail, A. A., &Patterson, M. M. (1986). Retention and acquisition of classicaltrace-conditioned responses by rabbits with hippocampal lesions.Behavioral Neuroscience, 100, 745-752.

Prokasy, W. F., Grant, D., & Myers, N. A. (1958). Eyelid conditioningas a function of unconditioned stimulus intensity and intertrialinterval. Journal of Experimental Psychology, 55, 242-246.

Salvatierra, A. T., & Berry, S. D. (1989). Scopolamine disruption ofsepto-hippocampal activity and classical conditioning. BehavioralNeuroscience, 103, 715-721.

Schacter, D. L. (1987). Implicit memory: History and current status.Journal of Experimental Psychology: Learning, Memory, and Cognition,13, 501-518.

Schmaltz, L. W., & Theios, J. (1972). Acquisition and extinction of aclassically conditioned response in hippocampectomized rabbits(Oryctolagus cuniculus). Journal of Comparative and PhysiologicalPsychology, 79, 328-333.

Sears, L. L., & Steinmetz, J. E. (1990). Acquisition of classicallyconditioned-related activity in the hippocampus is affected bylesions of the cerebellar interpositus nucleus. Behavioral Neurosci-ence, 104, 455-463.

Shimamura, A. P. (1986). Priming effects in amnesia: Evidence for adissociable memory function. Quarterly Journal of ExperimentalPsychology, 38A, 619-644.

Solomon, P. R., Blanchard, S., Levine, E., Valezquez, E., & Groccia-Ellison, M. (1991). Attenuation of age-related conditioning deficitsin humans by extension of the interstimulus interval. Psychology andAging, 4, 34-41.

Solomon, P. R., & Moore, J. W. (1975). Latent inhibition and stimulusgeneralization of the classically conditioned nictitating membraneresponse in rabbits (Oryctolagus cuniculus) following dorsal hippo-campal ablations. Journal of Comparative and Physiological Psychol-ogy, 89, 1192-1203.

Solomon, P. R., Pomerleau, D., Bennett, L., James, J., & Morse, D. L.(1989). Acquisition of the classically conditioned eyeblink responsein humans over the life span. Psychology and Aging, 4, 34—41.

Solomon, P. R., Solomon, S. D., Vander Schaaf, E., & Perry, H. E.(1983). Altered activity in the hippocampus is more detrimental toclassical conditioning than removing the structure. Science, 220,329-331.

Solomon, P. R., Stowe, G. T., & Pendlebury, W. W. (1989). Disruptedeyelid conditioning in a patient with damage to the cerebellar\npuls.BehavioralNeuroscience, 103, 898-902.

Solomon, P. R., Vander Schaaf, E. R., Weisz, D. J., & Thompson, R. F.(1986). Hippocampus and trace conditioning of the rabbit's classi-cally conditioned nictitating membrane response. Behavioral Neuro-science, 100, 729-744.

Spence, K. W. (1966). Cognitive and drive factors in the extinction ofthe conditioned eyeblink in human subjects. Psychological Review,73, 445^t58.

Spence, K. W., & Taylor, J. A. (1951). Anxiety and strength of the UCSas determiners of the amount of eyelid conditioning. Journal ofExperimental Psychology, 42, 183-188.

Squire, L. R. (1992). Memory and the hippocampus: A synthesis fromfindings with rats, monkeys, and humans. Psychological Review, 99,195-231.

Starr, A., & Phillips, L. (1970). Verbal and motor memory in theamnesic syndrome. Neuropsychologia, 16, 339-348.

Steinmetz, J. E., Lavond, D. G., Ivkovich, D., Logan, C. G., &Thompson, R. F. (1992). Disruption of classical eyeblink condition-ing after cerebellar lesions: Damage to a memory trace system or asimple performance deficit? Journal of Neuroscience, 12, 4403^426.

Thompson, R. F. (1986). The neurobiology of learning. Science, 233,941-947.

Thompson, R. F. (1989). A model system approach to memory. InP. R. Solomon, J. B. Goethals, C. M. Kelly, & B. R. Stephens (Eds.),Memory: Interdisciplinary approaches (G. Stanley Hall SymposiumSeries, Williams College; pp. 17-32). New York: Springer-Verlag.

Thompson, R. F. (1990). Neural mechanisms of classical conditioningin mammals. Philosophical Transactions of the Royal Society ofLondon B, 329, 161-170.

Thompson, R. F., Berger, T. W., Berry, S. D., Hoehler, F. K., Kettner,R. E., & Weisz, D. J. (1980). Hippocampal substrate of classicalconditioning. Physiological Psychology, 8, 262-269.

Topka, H., Valls-Sole, J., Massaquoi, S. G., & Hallett, M. (1993).Deficit in classical conditioning in patients with cerebellar degenera-tion. Brain, 116, 961-969.

Warrington, E. K., & Weiskrantz, L. (1974). The effect of priorlearning on subsequent retention in amnesic patients. Neuropsycho-logia, 12, 419^28.

Warrington, E. K., & Weiskrantz, L. (1978). Further analysis of theprior learning effect in amnesic patients. Neuropsychologia, 16,169-176.

Weiskrantz, L., & Warrington, E. K. (1979). Conditioning in amnesicpatients. Neuropsychologia, 17, 187-194.

WoodrurT-Pak, D. S. (1988). Aging and classical conditioning: Parallelstudies in rabbits and humans. Neurobiology of Aging, 9, 511-522.

Woodruff-Pak, D. S., & Thompson, R. F. (1988). Classical condition-ing of the eyeblink response in the delay paradigm in adults aged18-83. Psychology and Aging, 3, 219-229.

Received January 22,1993Revision received July 12, 1993

Accepted July 12, 1993

eyeblink classical conditioning in h.m.: delay and trace paradigms · 2018-03-16 · eyeblink...

Documents