Environmental Complexity Affects Contextual Fear Conditioning
Following Hippocampal Lesions in Rats
Sandra N. Moses,1Gordon Winocur,1–3* Jennifer D. Ryan,1,3,4and Morris Moscovitch1,3,4,5
measure for hippocampal function, even though several studies report
successful acquisition in hippocampal-damaged rodents. The current
study examined whether environmental complexity may account for
these discrepancies. We directly compared single-session contextual
fear conditioning in rats in a simple vs. complex environment. Hippo-
campal lesions led to reduced fear conditioning in both contexts, as
measured by freezing, but the effect was significantly greater in the
complex context. As well, lesions led to generalized fear when the com-
plex context was paired with shock, but not when the simple context
was paired. We suggest that the representation of the simple context
formed by rats with hippocampal lesions was adequate to support asso-
ciative learning, but the representation of the complex context, which
depended to a greater extent on relational learning, was not. The results
were interpreted as consistent with theories of hippocampal function
that emphasize its role in integrating multiple stimulus elements in a
memory trace. V V
Contextual fear conditioning has become a benchmark
C 2007 Wiley-Liss, Inc.
When an animal is administered an aversive foot-shock (US), uncondi-
tioned fear responses become associated with discrete stimuli (CSs) that
are paired with the shock, as well as with the general context in which
the experience occurred. Attempts to understand the mechanisms of
such conditioning have focused on several brain regions, particularly the
hippocampus. Following extensive investigation, there is broad consensus
that the hippocampus is not directly implicated in classical CS-US con-
ditioning (Kim and Fanselow, 1992; Phillips and LeDoux, 1992; but see
Maren et al., 1997), but the evidence with respect to contextual fear-
conditioning (CFC) is mixed. Several investigators have reported that
damage to the hippocampus severely impairs CFC (Kim and Fanselow,
1992; Phillips and LeDoux, 1992; Maren et al., 1997; Rudy et al.,
2002), others have reported no effect (McNish et al., 1997; Frankland
et al., 1998; Good and Honey, 1991; Winocur,
1997), and one group even reported enhanced CFC
under some conditions (Winocur et al., 1987).
Variations in fear conditioning tasks, type of lesion,
and measures of conditioned fear undoubtedly account
for some of the discrepancy among the reports (see
reviews by Phillips and LeDoux, 1994; Holland and
Bouton, 1999; Gewirtz et al., 2000), but one variable
that has received scant attention is the complexity of the
environment in which conditioning occurred. Frequently,
when HPC lesions were found to impair CFC, animals
were trained in a chamber with transparent walls that
allowed visual access to a wide range of background cues
(e.g., Kim and Fanselow, 1992; Phillips and LeDoux,
1992, 1994; Maren et al., 1997). In contrast, in those
studies where HPC lesions did not impair CFC, training
typically was conducted in a more confined environment
where fear responses could be conditioned to a limited
number of salient contextual stimuli (e.g., Winocur
et al., 1987; Good and Honey, 1991; Winocur, 1997).
Interestingly, parallel results of intact contextual learning
in a relatively simple environment, following HPC
lesions, have been obtained in nonaversive tasks. (Hirsh,
1974; Winocur and Olds, 1978; Skinner et al., 1994).
It is widely held that the hippocampus plays a cru-
cial role in integrating and forming relationships
among multiple stimulus elements for purposes of
guiding appropriate behavior (e.g., Cohen and Eichen-
baum, 1993; Rudy and Sutherland, 1995; Nadel and
Moscovitch, 1997; Fanselow, 1999; Rosenbaum et al.,
2001; Moses and Ryan, 2006). By this view, the hip-
pocampus is not considered to be involved in forming
direct associations between specific stimuli and uncon-
ditioned responses, and it follows that HPC lesions
should have limited effects on CFC in a relatively uni-
form environment where discrete background cues can
become directly associated with the fear response. By
contrast, in a complex environment in which an array
of contextual cues is associated with the uncondi-
tioned fear response, HPC lesions are likely to impair
the representation of the context and, hence, have a
detrimental effect on CFC.
If HPC lesions lead to impaired fear conditioning
to a complex context because representation of that
context in memory is impoverished, the shock-
induced fear response should generalize to other envi-
2Department of Psychology, Trent University, Peterborough, Canada;
3Department of Psychology, University of Toronto, Toronto, Canada;
4Department of Psychiatry, University of Toronto, Toronto, Canada;
5Department of Psychology, Baycrest Centre, Toronto, Canada
Grant sponsor: Natural Sciences and Engineering Research Council of
Canada; Grant number: AP8181.
*Correspondence to: Gordon Winocur, Ph.D., Rotman Research Institute,
Baycrest Centre, 3560 Bathurst Street, Toronto, Ontario, Canada M6A
2E1. E-mail: email@example.com
Accepted for publication 26 January 2007
Published online 5 April 2007 in Wiley InterScience (www.interscience.
HIPPOCAMPUS 17:333–337 (2007)
C2007 WILEY-LISS, INC.
ronments. On the other hand, if, following HPC lesions, a
simple context were represented effectively in memory, thus en-
abling the formation of a strong conditioned fear response, less
generalization of that response would be expected. These pre-
dictions are tested in the present study as part of an examina-
tion of the interactive effects of HPC lesions and environmen-
tal complexity on CFC.
Twenty-eight male Long Evans rats, obtained from the
Charles River Laboratories in St. Constant, Quebec, were
maintained on a 12:12 light/dark cycle. Experimentation was
carried out during the dark phase of the cycle. Fourteen rats
received bilateral NMDA neurotoxic lesions of the hippocam-
pus, and 14 received control surgery (Winocur et al., 2005).
The simple context was a rectangular chamber (50 3 40 3 18 cm3)
with white walls and a floor that consisted of metal rods,
spaced 1.3 cm apart. The top was open for observation. The
chamber was placed on a table, 1.3 m above the floor, against
a wall. The complex context was a chamber (40 3 40 3 25 cm3),
with clear walls, a hinged Plexiglas ceiling that contained sev-
eral holes for ventilation, and a grid floor. This chamber was
placed on a table in the center of a different room. This room
contained visually accessible standard furniture (e.g., desk,
table, bookshelf along one wall, etc.), as well as pictures, light
fixtures, etc. on the walls. In both rooms, illumination was pro-
vided by overhead fluorescent lights under rheostatic control.
Thus, the simple context consisted essentially of a homogene-
ous white surround. By comparison, the complex context con-
sisted of multiple stimulus elements with the potential of form-
ing multiple relations (associations) between them.
One day prior to training rats were preexposed in a counter bal-
anced order to both the simple and complex contexts for 15 min
each to reduce effects of novelty, (Phillips and Ledoux, 1992;
Rudy and O’Reilly, 1999). Training followed over two days. On
day 1 rats were placed in either the simple or complex context and
allowed to explore freely for 5 min. Half the rats in each context
then received a series of 10 foot shocks (1.5 mA; 1 s) with a vari-
able interval (10–120 s) between each shock. After the last shock,
the rat remained in the chamber for 60 s.
On training day 2 rats were placed in the context which
they had not experienced on day 1. The rats that were shocked
on day 1 received no shock on day 2, and the rats that were
not shocked on day 1 received shock on day 2 in an identical
manner as aforementioned.
Testing began 24 h after the second training day and
occurred over two days. On Test day 1, half of the rats were
placed in the simple context and half were placed in the com-
plex context for 8 min. On Test day 2, rats were placed in the
context which they had not experienced on the previous day
for 8 min. The design allowed for each animal to be tested in
the context in which it had been shocked (memory test) and,
in a test of generalization, in the context in which it had never
Freezing was defined, after Anagnostaras et al. (2001), as an
immobilized crouching response in which the only detectable
movement was the rat’s breathing. The amount of time spent
freezing was recorded throughout prexposure, training, and test
HPC lesions varied in size with most affecting 50–95% of
the hippocampus proper. Six of the 14 lesioned rats sustained
damage to 75–95% of the hippocampus, including extensive
damage to all the subfields [CA1-CA3, dentate gyrus (DG)].
Two rats had relatively small lesions that affected 20–30% of
the hippocampus but were included as their performance fell
within the range of their group. Overall, the median value for
HPC destruction was 68%, with the extent and pattern of
damage similar in all conditions. In all cases, extra-HPC dam-
age was minor or nonexistent (Fig. 1).
Table 1 provides the minimum, median, and maximum
times spent freezing by HPC and operated control (OC)
groups in the simple and complex contexts during the 15 min
preexposure sessions, the 60 s period following fear condition-
ing, and the 60 s control period before which no shock was
delivered. The preexposure data, which represent freezing times
averaged over 15, 60-s periods, clearly show that there was no
effect of lesion or context on freezing prior to the administra-
tion of shock. Not shown in the table, but consistent with this
extents of hippocampal lesions. Numbers represent distance in millimeters from bregma.
Coronal reconstructions of minimal (solid) and maximal (cross-thatched)
MOSES ET AL.
Hippocampus DOI 10.1002/hipo
finding, is that in the 5-min pre-shock period on the training
days, HPC and OC groups showed virtually no freezing in ei-
ther context. Similarly, as can be seen in Table 1, the immedi-
ate response to shock by HPC and OC groups was identical in
both contexts. Most rats in both groups froze for the entire
60-s period, in contrast to the corresponding period in the no-
shock condition, where there was virtually no freezing.
Figure 2 shows time spent freezing by each rat in the HPC
and OC groups when tested in the context that was paired
with shock (Fig. 2A) and when tested for generalization of the
fear response to the context that was not paired with shock
(Fig. 2B). As can be seen, the groups exhibited variable
amounts of freezing in the various conditions and this was
reflected in a significant 3-way, group x context-same x context-
different, interaction F1,245 18.40, P < 0.001.
Subsequent analyses of freezing in the context that was paired
with shock (Fig. 2A) confirmed that both groups exhibited more
freezing in the simple context than the complex context, F1,245
13.35, P 5 0.001 and that, overall, rats with HPC lesions exhib-
ited less freezing than controls, F1,24 5 25.83, P < 0.001.
Although the group x context interaction was not statistically sig-
nificant, F1,245 0.52, P 5 0.48, several lines of evidence indicate
that rats with HPC lesions were more impaired in acquiring a
conditioned fear response in the complex context than in the sim-
ple context. First, as can be seen in Figure 2A, whereas there was
no significant difference in freezing behavior between OC groups
tested in the simple and complex contexts, t125 1.77, P 5 0.11,
the corresponding difference between HPC groups was highly
significant, t125 4.48, P 5 0.002, with the HPC group in the
simple context exhibiting considerably more freezing at test. Sec-
ond, a measure of overlap between two groups (U0ranges from 0
to 24 for n 5 7) identified a complete separation of observed
freezing times between HPC and OC groups in the complex con-
text - that is, no OC rat exhibited shorter freezing times than any
of the HPC rats (U05 0) as shown in the left hand panel of Fig-
ure 2. By contrast, there was overlap in the simple context as
reflected by the fact that seven instances of overlap occurred
between OC and HPC rats (U05 7). This indicates that the dis-
Time Spent Freezing by HPC and OC Groups in Simple
and Complex Contexts During Pre-exposure,
Post-shock, No-shock Control Conditions
(0, 0, 2)
(0, 0, 0)
(50, 60, 60)
(0, 0, 3)
(0, 0, 0)
(48, 60, 60)
(0, 1, 3)
(0, 0, 0)
(45, 60, 60)
(0, 1, 2)
(0, 0, 12)
(57, 60, 60)
riod; in the pre-exposure condition, freezing was measured over 15 min and is pre-
sented as an average over 15, 60-s intervals. The numbers in each cell are the mini-
each rat in the simple and complex contexts following fear condi-
tioning in which shock was delivered in the same context. Hippo-
campal lesions impaired contextual fear conditioning in both con-
texts but hippocampal and control groups showed more freezing
in the simple than the complex context. (B) Dot plot showing the
time spent freezing by each rat in tests of generalization. Rats with
(A) Dot plot showing the time spent freezing by
hippocampal lesions exhibited more generalization when shocked
in the complex context than in the simple context. Control rats
showed similar generalization in both conditions. The horizontal
lines represent, the median value across HPC and OC rats in each
contextual condition and provide a reference for between-group
comparisons. The numbers in brackets represent group means.
CONTEXTUAL FEAR CONDITIONING AND ENVIRONMENTAL COMPLEXITY
Hippocampus DOI 10.1002/hipo
tributions of freezing times in the HPC and OC groups are more
similar in the simple than in the complex context.
Additional evidence that HPC lesions had a greater effect on
CFC in the complex context comes from analysis of the general-
ization findings. These data yielded a significant group x context-
different interaction, F1,245 6.17, P 5 0.02, indicating that dif-
ferences between groups in this test depended on the direction of
the context shift. As can be seen from a comparison of Figures
2A,B both OC groups exhibited the same generalization pattern.
That is, relative to their performance in the context that had been
paired with shock, controls in both generalization conditions
exhibited similar decreases in freezing behavior. As can be seen in
Figure 2B, there was no difference in amount of time spent freez-
ing between controls in the two generalization test conditions
t <1. Essentially the same pattern was seen in the HPC group
that was shocked in the simple context and tested for generalization
in the complex context. Their ability to discriminate between the
simple context and the complex context in the generalization test
(Fig. 2B) can be interpreted as further evidence that considerable
CFC occurred in these animals in the simple context. In contrast,
the HPC group that was shocked in the complex context exhib-
ited essentially the same amount of freezing when tested for gen-
eralization to the simple context as did the HPC group that was
tested in the complex context after having been shocked in that
context. This indicates that rats with HPC lesions shocked in the
complex context generalized whatever fear response they had
acquired and that the learning was not strongly associated with
the specific context in which the shock was delivered.
We interpret the substantial fear conditioning that occurred
in HPC rats in the simple context, where the environment con-
sisted essentially of a single explicit stimulus, as being similar to
classical fear conditioning in which a shock-mediated UCR
becomes associated with a discrete CS. As such, this finding is
consistent with numerous demonstrations of normal classical
fear conditioning in rats with HPC lesions (e.g., Kim and Fanse-
low, 1992; Phillips and LeDoux, 1994) as well as with other evi-
dence that HPC lesions have little or no effect on fear condi-
tioning in relatively simple contexts (Winocur et al., 1987;
Winocur, 1997). At the same time, the results showed that rats
with HPC lesions did exhibit less CFC in the simple context
than controls, indicating that, even in relatively uncomplicated
contexts, CFC is not necessarily fully spared following such
lesions (for similar results Chen et al., 1996; Rudy et al., 2002).
The finding that HPC lesions severely impaired CFC in the
complex context confirms numerous reports of this effect (e.g.,
Kim and Fanselow, 1992; Phillips and LeDoux, 1992; Maren
et al., 1997). It is noteworthy that rats with HPC lesions exhib-
ited some modest fear conditioning in the complex context,
which generalized to the simple context; there was no generaliza-
tion, however, from simple to complex context. The asymmetry
between the two generalization conditions likely occurred
because rats were able to create a functional representation of
the simple context, but the representation of the complex con-
text was extremely impoverished. In this situation the simple
context elicited fear probably because it was remembered more
accurately than the complex context. Importantly, this general-
ization effect was absent in control rats who could represent the
simple and complex context in memory equally well.
The ability of rats with HPC lesions to learn a new response
without using contextual cues and the resultant tendency to
generalize that response to another environment that shares
some features with the original environment has been observed
recently in other tasks (e.g., socially-acquired food preference;
Winocur et al., in press).
Overall, the results are consistent with theories of HPC function
that emphasize HPC involvement in learning about relationships
among multiple elements (Cohen and Eichenbaum, 1993; Rudy
and Sutherland, 1995; Nadel and Moscovitch, 1997; Fanselow,
1999; Rosenbaum et al., 2001; Moses and Ryan, 2006). They also
underscore the importance of distinguishing between different kinds
baum et al., 2001). Contextual cues can be linked relationally to
each other to form detailed spatial (O’Keefe and Nadel, 1978) or
nonspatial (Eichenbaum, 1999) representations of an event. Such
representations, when reinstated, support detailed and flexible ep-
isodic memories of the experience. Alternatively, each contextual
cue or combination of cues can be linked independently and
directly to the response that is part of the task at hand. These asso-
ciations give rise to less complex, rigid memories that are depend-
ent on the presence of specific background cues or landmarks.
Following the distinction between relational and associative learn-
ing, the two types of contextual representations have been charac-
terized as expressions of relational and associative contextual learn-
ing (Rosenbaum et al., 2001), with the hippocampus implicated
in the former but not the latter. Efficient fear conditioning in the
complex context of the present study required that rats form rep-
resentations of the environment in relational terms, thereby plac-
ing rats with HPC lesions at a disadvantage, and promoting gen-
eralization to other contexts. By contrast, in the simple context,
the highly salient background cue could be associated readily with
the UCR through extra-HPC structures, and little generalization
to other contexts would be evident. While normally less efficient
than relational learning, under such conditions, associative con-
textual learning in rats with HPC lesions can be expected to ap-
proximate that of normal rats.
The authors thank Andrea Goode and Marie-E`ve Couture
for their excellent technical support, and Malcolm Binns for
his help in analyzing the data. The research was funded by
grants to G.W. and M.M. from The Natural Sciences and En-
gineering Research Council of Canada.
Anagnostaras SG, Gale GD, Fanselow MS. 2001. Hippocampus and
contextual fear conditioning: Recent controversies and advances.
Chen C, Kim JJ, Thompson RF, Tonegawa S. 1996. Hippocampal
lesions impair contextual fear conditioning in two strains of mice.
Behav Neurosci 110:1177–1180.
MOSES ET AL.
Hippocampus DOI 10.1002/hipo
Cohen NJ, Eichenbaum H. 1993. Memory, amnesia, and the hippo- Download full-text
campal system. Massachusetts: The MIT Press.
Eichenbaum H. 1999. The hippocampus and mechanisms of declara-
tive memory. Behav Brain Res 103:123–133.
Fanselow MS. 1999. Learning theory and neuropsychology: Configur-
ing their disparate elements in the hippocampus. J Exp Psychol:
Anim Behav Proc 3:275–283.
Frankland PW, Cestari V, Filipkowski RK, McDonald RJ, Silva AJ.
1998. The dorsal hippocampus in essential for context discrimina-
tion but not for contextual conditioning. Behav Neurosci 70:44–
Gewirtz JC, McNish KA, Davis M. 2000. Is the hippocampus neces-
sary for contextual fear conditioning? Behav Brain Res 110:
Good M, Honey RC. 1991. Conditioning and contextual retrieval in
hippocampal rats. Behav Neurosci 105:499–509.
Hirsh R. 1974. The hippocampus and contextual retrieval of informa-
tion from memory: a theory. Behav Biol 12:421–444.
Holland PC, Bouton ME. 1999. Hippocampus and context in classical
conditioning. Curr Opin Neurobiol 9:195–202.
Kim JJ, Fanselow MS. 1992. Modality-specific retrograde amnesia of
fear. Science 256:675–677.
Maren S, Aharonov G, Fanselow MS. 1997. Neurotoxic lesions of the
dorsal hippocampus and Pavlovian fear conditioning in rats. Behav
Brain Res 88:261–274.
McNish KA, Gewirtz JC, Davis M. 1997. Evidence of contextual fear
after lesions of the hippocampus: a disruption of freezing but not
fear-potentiated startle. J Neurosci 17:9353–9360.
Moses SN, Ryan JD. 2006. A comparison and evaluation of the pre-
dictions of relational and conjunctive accounts of hippocampal
function. Hippocampus 16:43–65.
Nadel L, Moscovitch M. 1997. Memory consolidation, retrograde amne-
sia and the hippocampal complex. Curr Opin Neurobiol 7:217–
O’Keefe J, Nadel L. 1978. The Hippocampus as a Cognitive Map.
Oxford, UK: Oxford University Press.
Phillips RG, LeDoux JE. 1992. Differential contribution of amygdala
and hippocampus to cued and contextual fear conditioning. Behav
Phillips RG, LeDoux JE. 1994. Lesions of the dorsal hippocampal for-
mation interfere with background but not foreground contextual
fear conditioning. Learn Mem 1:34–44.
Rosenbaum RS, Winocur G, Moscovitch M. 2001. New views on old
memories: Re-evaluating the role of the hippocampal complex.
Behav Brain Res 127:183–197.
Rudy JW, O’Reilly RC. 1999. Contextual fear conditioning, conjunc-
tive representations, pattern completion, and the hippocampus.
Behav Neurosci 113:867–880.
Rudy JW, Sutherland RJ. 1995. Configural association theory and the
hippocampal formation: An appraisal and reconfiguration. Hippo-
Rudy JW, Barrientos RM, O’Reilly RC. 2002. Hippocampal forma-
tion supports conditioning to memory of a context. Behav Neuro-
Skinner DM, Martin GM, Harley C, Kolb B, Pridgar A, Bechara A,
van der Kooy D. 1994. Acquisition of conditional discriminations
in hippocampal lesioned and decorticated rats: Evidence for learn-
ing that is separate from both simple and classical conditioning
and configural learning. Behav Neurosci 108:911–926.
Winocur G. 1997. Hippocampal lesions alter conditioning to condi-
tional and contextual stimuli. Behav Brain Res 88:219–229.
Winocur G, Olds J. 1978. Effects of context manipulation on memory
and reversal learning in rats with hippocampal lesions. J Comp
Physiol Psychol 92:312–321.
Winocur G, Rawlins JNP, Gray JA. 1987. The hippocampus and con-
ditioning to contextual cues. Behav Neurosci 101:617–625.
Winocur G, Moscovitch M, Caruana DA, Binns M. 2005. Retrograde
amnesia in rats with lesions to the hippocampus on a test of spatial
memory. Neuropsychologia 43:1580–1590.
Winocur G, Moscovitch M, Sekeres M. (in press). Memory consoli-
dation or transformation: Context manipulation and hippocampal
representations of memory. Nature Neuroscience.
CONTEXTUAL FEAR CONDITIONING AND ENVIRONMENTAL COMPLEXITY
Hippocampus DOI 10.1002/hipo