Upload
lonkes
View
216
Download
0
Embed Size (px)
Citation preview
8/3/2019 Stephen Maren, Stephan G. Anagnostaras and Michael S. Fanselow- The startled seahorse: is the hippocampus nec
1/4
are not known, although it is assumedthat configural strategies normallysupersede elemental strategies.
The lions share of work implicat-ing the hippocampus in contextualconditioning has been derived fromstudies of fear conditioning in rats. Infear conditioning tasks, discrete and/orcontextual CSs that are paired withfootshock come to evoke conditionalfear responses (CRs) such as defecation,potentiated acoustic startle, elevatedblood pressure and freezing. In manylaboratories, contextual fear is assessedby returning the rats to the condition-ing chamber and measuring freezing.That the hippocampus plays an impor-tant role in contextual fear condition-
ing is indicated by studies demonstrat-ing that dorsal hippocampal lesionsattenuate freezing to contextual CSs,but do not alter freezing behavior todiscrete CSs (Refs 2,3). Hippocampal le-sions also impair contextual fear condi-tioning following unsignaled train-ing7,8. While these results suggest thatthe hippocampus is required for con-textual fear conditioning, recentlypublished data challenge this notion.
In a recent study, McNish et al.9 re-examined the role of the hippocampusin contextual fear conditioning by sim-ultaneously measuring two contextualfear responses: freezing and fear-
potentiated startle. Fear-potentiatedstartle to context was characterized byan increase in the amplitude of theacoustic startle response in a shock-associated context compared to startleamplitude in that same context beforefear conditioning. Consistent with ear-lier reports, these authors found thatdorsal hippocampal lesions impairedfreezing in the conditioning context.Surprisingly, however, they did not ob-serve impairments in fear-potentiatedstartle in this same context. Thus, byone measure of fear, context condi-tioning was disrupted by hippocampaldamage, but by another measure of
fear it was intact. In the light of theseresults, McNish et al. suggested thatdeficits in contextual freezing are notthe result of a failure in contextual
learning, but rather are the result ofresponse competition between freez-ing and the motor hyperactivity thattypically accompanies hippocampaldamage10,11. Other groups, includingour own, have reported increasedmotor activity and reduced freezing inrats with hippocampal lesions7,12. Basedon these collective results, McNish et al.argue that the hippocampus is notessential for contextual learning. Itshould be noted, however, that McNishet al. did not measure hyperactivity intheir report. At first glance, these datawould appear to call for a revision ofcurrent thinking regarding the role ofthe hippocampus in contextual learn-ing. However, when the full range of
effects of hippocampal lesions oncontextual fear is taken into account, itbecomes apparent that a responsecompetition explanation of contextualfreezing deficits is indefensible.
Competing responses or symptoms of
a common deficit?
As stated above, we and others havereported elevated motor activity in ratswith dorsal hippocampal lesions thataccompanies reduced contextual freez-ing in these animals8,12. In our study8,rats with lesions in other regions of thehippocampal formation (the fornix orentorhinal cortex) displayed a similar
pattern of results. Across these dif-ferent lesion groups, pre-trainingmotor activity correlated well withpost-training freezing, but within eachgroup this correlation was poor. Onthis basis, we argued that it is unlikelythat motor activity and freezingdeficits are causally related. To bolsterthis claim, we have compiled a corre-lation between freezing and motoractivity from all of the experiments inwhich we measured these parameters.In these experiments, motor activitywas measured during a 3min pre-shockperiod in the same chambers whereconditioning took place. As shown in
Fig. 1, there was no significant corre-lation between motor activity andcontextual freezing in rats with hippo-campal lesions.
M a r e n e t a l . C o n t e x t u a l f e a r c o n d i t i o n i ng UpdateComment
39Copyright 1998, Elsevier Science Ltd. All rights reserved. 1364-6613/98/$19.00 PII: S1364-6613(98)01123-1T r e n d s i n C o g n it i v e S c i e n c e s V o l . 2 , N o . 2 , F e b r u a r y 1 9 9 8
Stephen Maren is at
the Department of
Psychology and
Neuroscience
Program, University
of Michigan, Ann
Arbor, MI 48109-
1109, USA.
tel: +1 313 936 6532
fax: +1 313 763 7480
e-mail: maren@
umich.edu
Stephan G.
Anagnostaras and
Michael S. Fanselow
are at the
Department of
Psychology,
University of
California, Los
Angeles, CA 90095-
1563, USA.
The startled seahorse: isthe hippocampus
necessary for contextualfear conditioning?Stephen Maren, Stephan G. Anagnostaras and
Michael S. Fanselow
It is widely believed that the hip-pocampus plays an important role inlearning and memory in several speciesincluding rats, monkeys and humans.The type of memory served by thisstructure has been described as ex-plicit, configural, declarative, spatial,relational and episodic. In simpler terms,many would agree that the hippo-campus assembles cognitive represen-tations of stimuli and their relation-ships for storage in long-term memory.
Despite its various roles, the gen-eral wisdom is that the hippocampus isnot required for simple forms of asso-ciative learning, such as Pavlovian (clas-sical) conditioning. Nonetheless, sev-eral investigators have suggested that
the hippocampus may have a role inmediating a special case of Pavlovianconditioning, that is, conditioning tocontext1. In the past few years, con-siderable support for this view hasemerged24.
In the learning laboratory, con-text refers to the collection of stimuliassociated with the training environ-ment. Contexts are multimodal (includ-ing olfactory, visual, auditory, tactileand spatial stimuli) and tonic. Contextscan be contrasted with discrete condi-tional stimuli (CSs), such as tones orlights, which are unimodal and phasic.In conditioning experiments, contex-
tual conditioning occurs when trainingconsists of either signaled delivery ofan unconditional stimulus (US) by a CSor unsignaled delivery of a US (thatis, the context signals US delivery). Ineither case, it has been argued that inorder for contextual conditioning toproceed, a configural representationof the many stimulus elements withinthe context must be formed5. It is thisprocess of cognitive representation ofcontext that is thought to require thehippocampus1,6. Notwithstanding, it isconceivable that contexts could also beacquired as individual elemental asso-ciations, and this process would not be
expected to require the hippocampus.The rules governing the selection ofconfigural versus elemental strategiesfor acquiring contextual representations
8/3/2019 Stephen Maren, Stephan G. Anagnostaras and Michael S. Fanselow- The startled seahorse: is the hippocampus nec
2/4
What then is one to make of thebetween-group correlation in thesemeasures? Rather than citing motor
hyperactivity as the cause for contex-tual freezing deficits, we have arguedthat these two consequences of hip-pocampal damage reflect a commonsyndrome8. As others have argued13, in-creased motor activity in rats with hip-pocampal damage may reflect a failureto habituate exploratory activity. Weimagine that this failure to habituateexploration results from an inability ofthe hippocampal rat to encode a con-textual representation of the environ-ment12,14. Thus, we argue that deficitsin contextual freezing and motor hy-peractivity in novel environments inrats with hippocampal damage reflect
a deficit in contextual representation.
Recent versus remote memories
As in humans15 and monkeys16, hippo-campal damage in rats is associatedwith a temporally graded retrogradeamnesia. In other words, recent memo-ries are much more sensitive to hip-pocampal damage than are remotememories. In fear conditioning, for in-stance, dorsal hippocampal lesions im-pair contextual freezing when madeone day following training, but notwhen made 28 days following condi-tioning2. Hippocampal lesions dis-rupted context fear selectively, be-
cause fear conditioning to a tone CSwas not affected by hippocampal le-sions at any time. Importantly, neithercontext fear nor tone fear varied over
the retention intervals in control animals.In a related study, it was found that pre-exposing rats to the conditioning con-
text 28 days before training eliminatedlesion-induced deficits in contextualfreezing17.
Consider the implications for a re-sponse competition view of these data.If deficits in contextual freezing werethe result of motor hyperactivity, thenhyperactivity should have interferedwith contextual freezing at both theone day and 28-day training-to-lesionintervals. Similarly, pre-exposed ani-mals should not have been protectedfrom the hippocampal deficit. Thesewere not the observed results. Freezingwas disrupted when lesions were madeshortly, but not a long time, after train-
ing, and context pre-exposure pro-tected against this deficit. Even thoughthe lesion-to-testing interval was heldconstant in these studies, it is possiblethat one could argue, in each case, thatthe 28-day retention interval somehowprotects against hippocampal lesion-induced increases in motor activity.However, this argument is not tenablein view of recent data showing thattemporally graded context-freezingdeficits can be demonstrated in thesame animal. Anagnostaras et al. (1996Soc. Neurosci. Abstr. 22, 1380) trainedindividual rats on both a recent and aremote context memory. These rats ex-
hibited impaired freezing in the re-cently trained context, but normalfreezing in the remotely trained con-text. The response competition argu-
ment requires that the same rat is bothhyper- and hypoactive!
Of course, a response competition
view also predicts that freezing to toneCSs should be disrupted by motor hyper-activity. However, tone fear is typicallyimmune to the effects of hippocampallesions2,3. Nevertheless, the argumenthas been made that tone conditioningmay be less susceptible to hippocampaldamage than is context conditioningbecause it is a stronger memory. In theKim and Fanselow2 study, however,context-conditioning deficits in hippo-campal rats were found at levels of con-text fear that either equaled or exceededlevels of tone fear. A response com-petition view cannot possibly explainthis pattern of results.
Neurons or axons?
The work discussed thus far has madeuse of electrolytic hippocampal lesions.These lesions destroy not only neurons inthe dorsal hippocampus, but also axonalprojections that merely pass throughthe hippocampus. Because axonal pro-
jections running through the dorsalhippocampus have been implicated inhippocampal lesion-induced increases inmotor activity18, we recently made neuro-toxic lesions of the dorsal hippocampusthat selectively destroyed neurons in thehippocampus, leaving the fibers intact6.As expected, only the electrolytic lesions
increased motor activity.We then examined contextual fear
conditioning in these rats. Consistentwith earlier results, we found that
UpdateComment
M a r e n e t a l . C o n t e x t u a l f e a r c o n d i t i o n i n g
40T r e n d s i n C o g n it i v e S c i e n c e s V o l . 2 , N o . 2 , F e b r u a r y 1 9 9 8
A Pre-shock activity B Contextual fear C Activity suppression D Activityfreezing correlation
0
5
10
15
20
0
20
40
60
80
0
40
20
60
80
100
0
0 10 20 30 40
0.1
0.2
0.3
0.4
Cross-ov
ers
Cross-overs
Freezing(%
time)
Freezing(%
time)
Suppressionratio
Sham Hipp Sham Hipp Sham Hipp
Sham
Hipp
Fig. 1 Motor activity and freezing following hippocampal lesions. The data shown are pooled from five separate experiments conducted under similar and
balanced conditions. In all cases, hippocampal lesions were made before training so that an assessment of both hyperactivity and contextual freezing could be made
in the same rats. (A) Pre-shock activity. On the conditioning day, horizontal cage cross-overs were assessed during the 3 min period before the first shock. Rats that
received electrolytic lesions of the dorsal hippocampus (n = 48) made more cross-overs than sham animals (n = 49), which had lesions of other hippocampal regions,
during this period [F(1,95) = 24.3; p
8/3/2019 Stephen Maren, Stephan G. Anagnostaras and Michael S. Fanselow- The startled seahorse: is the hippocampus nec
3/4
hippocampal lesions made shortly aftertraining severely disrupted contextualfreezing, and lesions made a long time
after training yielded small deficits infreezing. Interestingly, we also foundthat neurotoxic hippocampal lesionsmade before training did not affectthe acquisition of contextual fear con-ditioning. In other words, rats withpre-training neurotoxic lesions condi-tioned to context normally. To explainthese results, we argued that intactrats use a configural strategy to ac-quire contextual fear (which is sensitiveto post-training hippocampal lesions),but that rats with neurotoxic hip-pocampal lesions use an elementalstrategy to acquire context fear (with-out a hippocampus a configural strat-
egy is not available). In either case, itcannot be argued that motor hyperac-tivity is responsible for the pattern ofdeficits that we observed, because neu-rotoxic hippocampal lesions did notyield enhanced motor activity. Andeven if motor hyperactivity were re-cruited as an explanation, it cannot ex-plain temporally graded retrogradeamnesia or the selectivity of the deficitfor post-training as compared to pre-training lesions. Again, it seems that aresponse competition argument failsto explain the existing data.
It should also be pointed out that re-versible pharmacological manipulations
of the hippocampus prevent contextualconditioning17. In these experiments,intra-hippocampal infusion of an N-methyl-D-aspartate receptor antagonist
during training attenuated contextualfreezing measured during a drug-freetest. A response competition account is
unable to account for these data.
Startle versus freezing
The foregoing arguments make astrong case that deficits in contextualfreezing in hippocampal rats are notmerely the outcome of motor hyperac-tivity. But what accounts for the pat-tern of results reported by McNish etal.? Why do hippocampal lesions affectcontextual fear as measured by freez-ing but not by potentiated startle? Onehint at an explanation can be found bya close inspection of their freezingdata. Typically, in studies of contextualconditioning, freezing is eliminated
completely by changes in context cuesand by post-training lesions of the hip-pocampus2,19. By comparison, McNishet al. reported rather small deficits withboth of these manipulations. Perhapsthe residual freezing following eithercontext shift or hippocampal lesionswas mediated by conditioning to somesalient element of the context. This el-ement could have been provided byMcNish and colleagues use of small,confining chambers. If this is true, thena few simple assumptions would allowus to simulate the McNish et al. data,namely: (1) hippocampal lesions selec-tively eliminate the use of configural
cues (an assumption that is consistentwith most views of hippocampal func-tion), (2) freezing has a lower responsethreshold than fear-potentiated startle
(this assumption is quoted from theconclusions of McNish et al.), and (3) po-tentiated startle is more sensitive toelemental than to configural cues. Thislast assumption may be the most con-troversial of the three, but there iscertainly substantial evidence in thePavlovian conditioning literature that dif-ferent aspects of the CS control differ-ent CRs, even though all are related toa single CSUS association20. These threeassumptions yield a model that predictsthe entire pattern of results reported byMcNish et al., that is: (1) hippocampallesions cause a partial reduction in freez-ing, (2) hippocampal lesions do not af-fect startle, and (3) context shifts par-tially reduce freezing and completelyeliminate startle. Figure 2 illustratesthe results of this simulation. Obviously,additional research must test the verac-ity of this model, but we feel confidentthat whatever the eventual solution isit will not be an explanation that relies
on response competition.
Conclusions
The work by McNish et al. has providednew insights into the neural mecha-nisms of contextual fear conditioning.Clearly, models that invoke a role forthe hippocampus in the mediation ofcontextual learning are impelled to ex-plain the selective deficits in freezingthat McNish et al. have described. Fromour perspective, it is insufficient to in-voke response competition as an expla-nation for this pattern of results.Although the exact cause of this pat-tern of deficits is not yet known, we
feel that it is premature to concludethat the hippocampus is not essentialfor contextual fear conditioning.Further parametric examination offear-potentiated startle in rats withhippocampal lesions is required.
References
1 Nadel, L. and Willner, J. (1980) Context and
conditioning: a place for space Physiol.
Psychol. 8, 218228
2 Kim, J.J. and Fanselow, M.S. (1992) Modality-
specific retrograde amnesia of fear Science
256, 675677
3 Phillips, R.G. and LeDoux, J.E. (1992)Differential contribution of amygdala and
hippocampus to cued and contextual fear
conditioning Behav. Neurosci. 106, 274285
4 Selden, N.R. et al. (1991) Complementary
roles for the amygdala and hippocampus in
aversive conditioning to explicit and
contextual cues Neuroscience 42, 335350
5 Fanselow, M.S. (1986) Associative vs
topographical accounts of the immediate
shock freezing deficit in rats: implications
for the response selection rules governing
species specific defensive reactions Learn.
Motiv. 17, 1639
6 Maren, S., Aharonov, G. and Fanselow, M.S.
(1997) Neurotoxic lesions of the dorsalhippocampus and Pavlovian fear condition-
ing in rats Behav. Brain Res. 88, 261274
7 Kim, J.J., Rison, R.A. and Fanselow, M.S.
M a r e n e t a l . C o n t e x t u a l f e a r c o n d i t i o n i ng UpdateComment
41T r e n d s i n C o g n it i v e S c i e n c e s V o l . 2 , N o . 2 , F e b r u a r y 1 9 9 8
Freezing Startle Freezing Startle0
20
40
60
80
100McNish et al.
Simulation
Hippocampal lesionContext shift
Proportionofnorma
lbehavior(%)
Fig. 2 The actual data of McNish et al. and the data from our simulation presented
as a proportion of normal behavior displayed. Our model assumes that there are 100
shock-associated units, half of which are configural; the other half are elemental. Contextshifts eliminate half of each type of unit. Hippocampal lesions eliminate all of the con-
figural units but none of the elemental units. A subtractive-response elicitation threshold
of 25 units is set for startle, and a threshold of 10 is set for freezing. Freezing is determined
by both the available configural and elemental units, while startle is determined only
by the elemental units. Therefore, freezing = (elemental + configural) 10 and startle =
elemental 25. In the figure, context shift represents the measure of fear (either freezing
or potentiation of startle) in the novel test chamber divided by the measure of fear ob-
tained in the training chamber. For hippocampal lesions, the hippocampal data are divided
by the sham data. The model represents the first simulation that was conducted and no
data-fitting techniques were used.
8/3/2019 Stephen Maren, Stephan G. Anagnostaras and Michael S. Fanselow- The startled seahorse: is the hippocampus nec
4/4
(1993) Effects of amygdala, hippocampus,
and periaqueductal gray lesions on short-
and long-term contextual fear Behav.
Neurosci. 107, 10931098
8 Maren, S. and Fanselow, M.S. (1997)
Electrolytic lesions of the dorsal hippo-
campus, fimbria/fornix, or entorhinal cortex
produce anterograde deficits in contextual
fear conditioning in rats Neurobiol. Learn.Mem. 67, 142149
9 McNish, K.A., Gewiritz, J.C. and Davis, M.
(1997) Evidence of contextual fear after
lesions of the hippocampus: a disruption of
freezing but not fear potentiated startle
J. Neurosci. 17, 93539360
10 Douglas, R.J. and Isaacson, R.L. (1964)
Hippocampal lesions and activity Psycho-
nomic Sci. 1, 187188
11 Blanchard, D.C. et al. (1977) Movement arrest
and the hippocampus Physiol. Psychol. 5,
331335
12 Good, M. and Honey, R.C. (1997) Dissociable
effects of selective lesions to the
hippocampal subsystems on exploratory
behavior, contextual learning, and spatial
learning Behav. Neurosci. 111, 487493
13 OKeefe, J. and Nadel, L. (1978) The
Hippocampus as a Cognitive Map, Clarendon
Press
14 Fanselow, M.S. (1996) Modality-specificmemory of fear: differential involvement of
the amygdala and hippocampal formation in
Pavlovian fear conditioning, in Perception,
Memory and Emotion: Frontiers in Neuro-
science (Ono, T. et al., eds), pp. 499512,
Pergamon Press
15 Squire, L.R. and Alvarez, P. (1995) Retrograde
amnesia and memory consolidation: a neuro-
biological perspective Curr. Opin. Neurobiol.
5, 169177
16 Zola-Morgan, S. and Squire, L.R. (1990) The
primate hippocampal formation: evidence
for a time-limited role in memory storage
Science 250, 288290
17 Young, S.L., Bohenek, D. and Fanselow, M.S.
(1994) NMDA processes mediate anterograde
amnesia of contextual fear conditioning
induced by hippocampal damage:
immunization against amnesia by context
preexposure Behav. Neurosci. 108, 1929
18 Whishaw, I.Q. and Jarrard, L.E. (1995)Similarities vs differences in place learning
and circadian activity in rats after
fimbria-fornix section or ibotenate removal
of hippocampal cells Hippocampus 5,
595604
19 Fanselow, M.S. (1990) Factors governing one
trial contextual conditioning Anim. Learn.
Behav. 18, 264270
20 Holland, P.C. (1984) Origins of behavior in
Pavlovian conditioning, in The Psychology
of Learning and Motivation (Bower, G., ed.),
pp. 129174, PrenticeHall
UpdateComment
M a r e n e t a l . C o n t e x t u a l f e a r c o n d i t i o n i n g
42Copyright 1998, Elsevier Science Ltd. All rights reserved. 1364-6613/98/$19.00 PII: S1364-6613(98)01124-3
T r e n d s i n C o g n it i v e S c i e n c e s V o l . 2 , N o . 2 , F e b r u a r y 1 9 9 8
Response from McNish,Gewirtz and Davis
In our recent article examining therole of the hippocampus in contextualfear conditioning, we developed aparadigm which produced contextualfreezing and fear-potentiated startlethat was specific to a context previ-ously paired with shock1. Lesions ofthe central nucleus of the amygdala
blocked both freezing and fear-poten-tiated startle, consistent with the no-tion that this structure is critically in-volved in mediating conditioned fearresponses. In contrast, lesions of thedorsal hippocampus disrupted contex-tual freezing, but had no effect onfear-potentiated startle. Based onthese results, we concluded that de-spite a disruption of freezing, fear tothe context was preserved in animalswith hippocampal lesions.
Our interpretation of the effectsof hippocampal lesions on contextualfreezing challenges the notion thatcontext conditioning, like spatial learn-
ing, is a hippocampal-dependent task.This notion was encouraged by demon-strations that lesions of the hippocam-pus disrupted freezing to contextualcues, but had no effect on freezing toexplicit cues2,3. One interpretation ofthese findings is that the hippocampusis critically involved in forming com-plex, polymodal associations, as wouldbe required in forming a represen-tation of context but not in unimodalor elemental associations4. An alterna-tive interpretation is that hippocampallesions enhance motor activity, whichpreferentially disrupts weak condi-tioned freezing responses5. Given that
contextual fear is likely to be less strongthan fear to explicit cues6, one mightexpect the lesions to have a greater im-pact on freezing to contextual cues.
The commentary by Maren et al. at-tempted to rule out the response com-petition hypothesis. Furthermore, theypropose a model that appears to simu-late our data while preserving the cen-tral role of the hippocampus in contex-tual fear conditioning. Below, we willoutline why response competition,
coupled with a strength of condition-ing argument, is a reasonable alterna-tive explanation for the effects of hippo-campal lesions on freezing. We will alsohighlight several problems inherent inthe model proposed by Maren et al.
Response competition
Hippocampal lesions increase motor
activity
It has frequently been reported thathippocampal lesions increase motor ac-tivity. Recently, Maren and Fanselow7
have reported that across-groups in-creases in motor activity produced bylesions of the dorsal hippocampus, en-
torhinal cortex and fimbria-fornix werehighly correlated with the disruptionof freezing. They have argued thatthese effects are not causal but reflecta common underlying syndrome, be-cause within a given group the corre-lations between activity and freezingdeficits are poor. However, the lack ofsignificant within-group correlationsdoes not discount a causal relationship.Because the lesions significantly en-hanced motor activity, there is a nar-rower distribution of activity levelswithin a group than across groups, de-creasing the likelihood of finding a sig-nificant correlation within a group.
The important point is that becausetheir experimental manipulation wasat the group level, it is the significantbetween-groups correlation that is rel-
evant, not the non-significant within-group correlations.
Interestingly, it has recently beenreported that excitotoxic dorsal hip-pocampal lesions produced increases inactivity, deficits in freezing and impair-ments in spatial learning5. In contrast,entorhinal cortex lesions also disruptedspatial learning, but had no effect oneither activity or freezing. This sug-gests that there is a closer relationshipbetween motor activity and freezingthan between freezing and spatiallearning. If the freezing deficits trulyreflected a disruption of contextual
fear conditioning, one would have ex-pected them to go hand-in-hand withdeficits in spatial learning.
In an attempt to rule out a responsecompetition account, Maren et al. citea study showing that local infusion ofthe N-methyl-D-aspartate (NMDA) an-tagonist DL-2-amino-5-phosphonovaler-ate (APV) into the dorsal hippocampusduring contextual fear conditioningdisrupted freezing measured the nextday8. However, the dose of APV in-fused into the hippocampus (10 g)was twice the maximal dose given in-traventricularly (5 g) to block contex-tual fear conditioning9. Because lower,
rather than higher, doses of APV givenlocally would be expected to block con-ditioning, these data do not rule outthe possibility of spread to extra-hippocampal structures or the ventri-cles. Hence, further studies are re-quired to demonstrate the importanceof NMDA receptors in the hippocam-pus in contextual fear conditioning.
Hippocampal lesions disrupt freezing
to explicit cues
An important foundation of Maren etal.s thesis is that: dorsal hippocampallesions attenuate freezing to con-textual conditioned stimuli (CSs) but do
not alter freezing behavior to discreteCSs. However, they have recentlyreported that chemical lesions of thedorsal hippocampus disrupted freezing
K.A. McNish,
J.C. Gewirtz and
M. Davis are at the
Departments of
Psychiatry and
Psychology, Yale
University School of
Medicine,
Connecticut Mental
Health Center, New
Haven, CT 06508,
USA.
tel: +1 203 789 7448fax: +1 203 562 7079
e-mail: michael.davis
@yale.edu