NEUROREPORT
LEARNING AND MEMORY
Brain representation of habituation to
repeated complex visual stimulation studied
with PET
HaÊkan Fischer,1,2,CA Tomas Furmark,3 Gustav Wik4 and Mats Fredrikson3
1
Department of Psychiatry, Massachusetts General Hospital East and Harvard Medical School, CNY-9130, Building 149, 13th
Street, Charlestown, MA 02129, USA; 2 Uppsala University PET-centre, University Hospital, Uppsala; 3 Department of
Psychology, Uppsala University, Uppsala; 4 Department of Clinical Neuroscience, Karolinska Institute and Hospital, Stockholm,
Sweden
CA
Corresponding Author
Received 6 October 1999; accepted 25 October 1999
Acknowledgements: We are grateful to the staff of the Uppsala University PET-centre for providing excellent data collection
conditions. We are also thankful to the participating subjects. Supported by grants from the Swedish Brain Foundation (H.F.),
Wenner-Gren Foundations (H.F.) and the Swedish Council for Research in the Humanities and Social Sciences (M.F.).
To investigate CNS habituation (i.e. response decrement due
to stimulus repetition) the present study used positron
emission tomography (PET) to measure regional cerebral blood
¯ow (rCBF) in eight healthy women during two repetitions of
complex visual stimuli. Repeated visual stimulation resulted in
neural habituation bilaterally in the secondary visual cortex and
in the right medial temporal cortex including the amygdala and
the hippocampus. Regional CBF in the left thalamus was
elevated as a function of repeated stimuli presentations. Thus,
repeated presentation of complex visual stimuli result in rCBF
habituation in later stages of the visual processing chain. The
elevated neural activity in the thalamus might be associated
with interruption of further neural transmission related to
suppression of non-meaningful behavior. NeuroReport 11:123±
126 & 2000 Lippincott Williams & Wilkins.
Key words: Amygdala; Emotion; Hippocampus; Human; Occipital cortex; Positron emission tomography; Signal stimulus; Temporal cortex;
Thalamus
INTRODUCTION
Habituation, i.e. response decrement due to stimulus
repetition, occurs throughout the animal kingdom from
aplysia, where peripheral nervous system processes are
deemed important, to humans where central nervous
system (CNS) processes could have a primary role [1].
Numerous studies in rodents, dogs, cats, monkeys and
humans have revealed that autonomic nervous system
processes such as skin conductance responses, ®nger pulse
volume and heart rate undergo response diminution as a
function of stimulus repetition [2]. In humans, Sokolov [1]
proposed that with repeated presentation of a novel
stimulus a neural brain model is created with the hippocampal formation as being critically involved in information processing and storing. Habituation occurs at different
rates depending on stimulus qualities. Repeated exposure
to complex and behaviorally salient stimuli (i.e. signal
stimuli), result in slower autonomic nervous system habituation than simple non-salient stimuli (i.e. non-signal
stimuli) [e.g. 3], possibly re¯ecting a different time course
in the development of the neural representation. To our
0959-4965 & Lippincott Williams & Wilkins
knowledge no previous study on human brain function
using brain imaging techniques with high temporal and
spatial resolution has quantitatively investigated CNS habituation differences to stimuli of different signal value.
Previous studies have indicated that repeatedly presented pictorial snakes results in slower habituation of
autonomic nervous system responses than visual representation of neutral house stimuli [3]. The present study used
PET and a factorial 2 3 2 within-subject analysis of variance
(ANOVA) for repeated measures with stimulus repetition
(®rst vs second presentation) and stimulus (snake vs park
video) as factors to investigate CNS habituation to snake
and park visual stimuli.
MATERIALS AND METHODS
Subjects: Eight women, mean age 25.6 years (range 20±32
years) were selected from a group of healthy volunteers.
Subjects refrained from both tobacco and caffeine for 12 h
prior to the PET investigation. The study was approved by
the Uppsala University Medical Faculty Ethical Review
Board and the Uppsala University Isotope Committee.
Vol 11 No 1 17 January 2000
123
NEUROREPORT
Informed consent was obtained from all participants
according to the declaration of Helsinki.
Procedure: Subjects viewed two videos showing either
park scenes in the wintertime (non-signal stimuli) or
snakes indoors and outdoors (signal stimuli). Each video
was displayed continuously for 2 min during PET-scanning, on an 11 3 8.5 cm TV screen about 30 cm in front of
the subjects eyes. Both videos were novel to the subjects
and each ®lm was subsequently presented twice with
10 min in between presentations. Thus, in the present study
each subject was exposed to a total of four presentations.
The order of the two videotapes was counterbalanced
between subjects. The subjects were instructed to focus on
the TV screen during each video presentation. Measurements of regional cerebral blood ¯ow (rCBF) and skin
conductance were obtained during each PET scan while
subjective ratings of distress were performed after.
Ratings and autonomic nervous system recordings: Subjective units of distress (SUDs) were rated verbally ranging
from 0 (minimum) to 100 (maximum). Electrodermal activity was recorded by Beckman±Offner Ag/AgCl electrodes,
8 mm in diameter, housed in plastic cups through a
Hagfors-type constant voltage circuit [4]. Skin conductance
level ¯uctuations (NSF) that exceeded 0.05 ìS were estimated and expressed in numbers/min.
Blood ¯ow recordings: An eight-ring brain PET scanner
(GEMS PC2048-15B) with a 10 cm axial ®eld of view was
used [5]. The scanner produces 15 slices with 6.5 mm slice
spacing and an axial and transaxial resolution of 6 mm.
After insertion of a venous catheter, subjects were positioned in the scanner such that the planes were parallel to a
line between the anterior and posterior commissure, where
the most basal slice was below the orbita. Within 30 s after
the start of each video and immediately before each emission
scan, 600±800 MBq of H2 15 O in 3±4 ml water (10 MBq/kg
body weight) was injected i.v. at 12 min intervals. Data were
collected in ®fteen 10 s frames following injection. In the
present study each subject had four scans during two
conditions. Each subject also had four additional scans
following the four scans included in the present study, of no
relevance for this study and therefore not described here.
After all scans an additional injection was performed during
which the scanner couch was automatically translated back
and forth between two positions 10 cm apart. Data were
collected only when in one of these positions and not during
the transfer. This additional scan was performed in order to
obtain a scan with full axial coverage of the brain only to aid
in the stereotactical normalization.
Data from the ®rst 100 s after arrival of bolus to the brain
were summed in a weighted fashion such that data from
the initial uptake phase were given a higher weight [6].
Images were reconstructed from the weighted summation
after being corrected for attenuation, dead time, scatter [7]
using the transmission scan. The additional scan with
couch translations was constructed to a 30 slice image set
with a 20 cm coverage in axial direction using contour
®nding for attenuation correction [8]. All images were
reconstructed to a 128 3 128 matrix with a pixel size of
2 mm using a 15 mm Hanning ®lter.
124
Vol 11 No 1 17 January 2000
H. FISCHER ET AL.
Anatomical standardization: Anatomical normalization
of all individual CBF images into a standard brain shape
[9] was performed automatically [10] by matching the scan
with 20 cm axial coverage to an atlas template. The images
from all 10 cm emission scans were automatically aligned
to the 20 cm scan [11], bringing them into the stereotactic
space and correcting for subject movements between scans.
In order to facilitate comparisons between studies from
different laboratories the Talairach brain [12] has been
mapped into the atlas, allowing results to be communicated in Talairach co-ordinates.
Statistical analysis: Affective and skin conductance
measures of habituation were analyzed using one-tailed
paired t-tests because of a priori hypotheses [1±3]. A 2 3 2
factorial design for repeated measures with stimulus repetition (®rst vs second presentation) and condition (snake vs
park stimulus) as factors was used to analyze subjective
and autonomic nervous system measures.
PET data were normalized for global ¯ow using linear
scaling and analyzed as a 2 3 2 within-subject ANOVA for
repeated measurements with repetition (®rst vs second
presentation) and stimulus (snake vs park stimulus) as
factors, modeled as a one-way blocked ANOVA [13]. The
areas in the brain exhibiting a change in perfusion resulting from the experimental design were excluded from the
estimation of global ¯ow [14], thereby ensuring independence between local and global ¯ow. A ®rst contrast
hereafter referred to as the main effect of stimulus repetition contrast was performed to evaluate rCBF differences
between two subsequent visual presentations averaged
over snake and park videos. A second contrast referred to
as the main effect of stimulus contrast was preformed to
evaluate rCBF differences between snake and park stimulus averaged over the ®rst and second presentation. A
third contrast hereafter referred to as the stimulus repetition by condition contrast was performed to evaluate
differences between two subsequent presentations of snake
and park videos. Each contrast generated a t-map that was
converted to a z-score map using a probability preserving
transform [15]. Local changes were evaluated using the
spatial extent of connected clusters with z-score values
. 1.96.
RESULTS
Main effect of stimulus repetition contrast: A reduction
in subjective units of distress (t(7) ˆ 1.9; p , 0.05) supported behavioral habituation between the ®rst (21.1 28.6,
mean s.d.) and the second visual stimuli presentation
(13.4 18.8) while measures of skin conductance did not
change from the ®rst (1.8 1.2) to the second presentation
(1.9 0.9; t(7) , 1; n.s.).
For cerebral blood ¯ow (CBF) there was a signi®cant
difference between the ®rst and second visual stimuli
presentation as re¯ected by an omnibus p , 0.001. Regional
CBF was higher during the ®rst than the second stimulus
presentation in three clusters with the ®rst located in the
right medial temporal cortex (cluster size 802 voxels), the
second in the left secondary visual cortex (cluster size 695
voxels) while the third was located in the right secondary
visual cortex (area 19; cluster size 437 voxels; Table 1).
During the second visual stimulation rCBF was higher in
NEUROREPORT
HABITUATION AND HUMAN BRAIN FUNCTION
Table 1. Regional CBF alterations as a function of complex visual stimulus repetition (®rst vs second presentation) averaged over snake and park
video presentations; brain and Brodmann areas included in each cluster, Talairach coordinates for maximum voxel z-values within each cluster, and
cluster p-values for signi®cant decreases and increases in rCBF
Brain area
Regions with decreased rCBF
Cluster 1
R temporal cortex
R amygdala
R hippocampus
Cluster 2
L occipital cortex
Cluster 3
R occipital cortex
Regions with increased rCBF
Cluster 1
L thalamus
L premotor cortex
Brodmann area
x
y
z
Maximum voxel z-value
15,20,28,34,35,36,38
47
22
22
ÿ26
ÿ6
ÿ12
ÿ17
ÿ17
ÿ21
3.64
3.32
3.15
19
ÿ40
ÿ68
ÿ16
3.40
19
54
ÿ63
25
3.13
6
ÿ18
ÿ35
ÿ24
ÿ24
14
19
3.56
2.56
Cluster p-value
0.002
0.05
0.05
0.03
Bold indicates the location of the maximum voxel value where several areas are given together.
Coordinates correspond to those in the atlas of Talairach and Tournoux [12]. Signi®cance of clusters has been evaluated based on the spatial extent of suprathreshold
clusters with z-scores of > 1.96. R ˆ right hemisphere; L ˆ left hemisphere.
one cluster located in the left thalamus extending into the
premotor cortex (area 6; cluster size 506 voxels; Table 1).
Main effects of stimulus contrast: There were no signi®cant stimulus related differences in either subjective affect
or skin conductance ¯uctuation between snake and park
stimulation.
There was a signi®cant omnibus effect when contrasting
CBF during snake and park stimulation ( p , 0.0001). Locally, rCBF was higher during snake than during park
presentations in one cluster (cluster size 9153 voxels)
located bilaterally in the primary (Brodmann area 17) and
secondary visual cortex (areas 18 and 19) extending into
the left and right parietal cortex (area 39) and the left
posterior temporal cortex (area 37).
Stimulus repetition by condition contrast: No signi®cant
interaction effects between stimulus repetition (®rst vs
second presentation) and condition (snake vs park stimulus) were evident either for subjective distress or skin
conductance. No signi®cant rCBF interaction effects between stimulus repetition and stimulus content [(®rst
minus second snake presentation) minus (®rst minus
second park presentation)] were evident.
DISCUSSION
Behavioral and central neural responses in the right medial
temporal cortex including the amygdala and the hippocampus, and bilaterally in the secondary occipital cortex
habituated similarly to signal and non-signal visual stimuli
while rCBF was higher during the second as compare to
the ®rst presentation in the left thalamus and the left
premotor cortex. These results support Sokolov's [1] proposal that the hippocampal formation is critically involved in
habituation. One explanation why behavioral and central
nervous system responses did not habituate at different
rates to signal and non-signal stimuli could be that the
present study paradigm has too few trials. Another is that
the temporal resolution of 100 sec in the present study is
too low to capture habituation differences between signal
and non-signal stimulus. The present data are in line with
Rajah et al's study [16] that showed rCBF habituation to
motor related visual signal stimuli both in the occipital and
temporal cortices. Because projections to the medial temporal lobe comes from the associative visual cortex rather
than the striatal cortex the input seems to contain behaviorally relevant complex representations rather than simple
perceptual features. Thus, the decreased rCBF in the
entorhinal, perirhinal and hippocampal areas might re¯ect
reduced neural input from higher order visual areas or
more effective stimulus processing as a function of learning. Lesions of both the hippocampus and the amygdala
produce larger de®cits in recognition memory in animals
than separate lesions of just the amygdala or the hippocampus [17]. Thus, a reduced activity in both the hippocampus and the amygdala as observed here might serve to
prevent further processing of cognitively and emotionally
irrelevant information, and this analysis may be mediated
by the thalamus through simple recognition memory. This
is consistent with previous functional imaging studies
showing that the thalamus is involved in recognition
memory [18] and supports an active directing rather than a
passive transmission role for the thalamus.
The increased thalamic rCBF together with the decreased
rCBF in the occipital and temporal cortices suggest that
recognition memory in the context of repeated complex
visual stimulation result in higher subcortical and lower
cortical neural activation in brain regions of relevance for
visual stimuli processing. Hence, the present data might
suggest that recognition memory related to habituation to
visual stimuli is subcortically rather than cortically represented. It could be argued that, at least on a short-term
basis, the growth of the neural model is located in the
thalamus rather than in the hippocampus, as proposed by
Sokolov [1]. It is likely that the function of habituation is to
suppress allocation of limited attention resources to behaviorally non-meaningful stimulation. Thus, the increased
activity in the thalamus could be related to behavioral
inhibition or suppression of the neural input at this stage
either alone or through back-projections to the visual
cortex.
Neural responses in the amygdala undergo habituation
Vol 11 No 1 17 January 2000
125
NEUROREPORT
not only to static pictorial visual stimuli, as previous
imaging studies have shown [e.g. 19], but also to complex
moving visual stimuli. The right rather than left localized
neural alteration possibly re¯ect the pictorial non-verbal
nature of the stimuli [20] or that attentive processes in
general seem right rather than left lateralized [21]. That
rCBF in the amygdala decreases as a function of repeated
visual stimulation might explain why amygdala activation
has not been found in previous functional imaging studies
of normal [22] and pathological fear and anxiety [23] that
use paradigms where visual stimuli are repeatedly presented over time.
In conclusion, repeated presentation of both visual
signal and non signal stimuli resulted in neural habituation in the amygdala, the hippocampus as well as in
secondary and tertiary but not primary visual processing
brain areas. Increased neural activity was evident in the
left thalamus and premotor cortex. The lowered subcortical and cortical, and the enhanced thalamic neural activity
as a function of repeated visual stimulation might re¯ect
suppression of non relevant sensory information at the
thalamic level.
REFERENCES
1.
2.
3.
4.
Sokolov EN. Pavlov J Biol Sci 25, 142±150 (1990).
Thompson RF and Spencer WA. Psychol Rev 73, 16±43 (1966).
Ohman A, Eriksson A, Fredrikson M et al. Biol Psychol 2, 85±93 (1974).
Venables PH and Christie MJ. Mechanisms, instrumentation, recording
126
Vol 11 No 1 17 January 2000
H. FISCHER ET AL.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
techniques and quanti®cation of responses. In: Prokasy WF and Raskin
DC, eds. Electrodermal Activity in Psychological Research. New York:
Academic Press, 1973: 1±124.
Holte S, Eriksson L and Dahlbom M. Eur J Nucl Med 15, 719±721 (1989).
Andersson JLR and Schneider H. J Nucl Med 38, 334±340 (1995).
BergstroÈm M, Eriksson L, Bohm C et al. J Comput Assist Tomogr 7, 42±50
(1983).
BergstroÈm M, Litton J, Eriksson L et al. J Comput Assist Tomogr 6, 365±
372 (1982).
Greitz T, Bohm C, Holte S and Eriksson L. J Comp Assist Tomogr 15, 26±
38 (1991).
Andersson JLR and Thurfjell L. J Comput Assist Tomogr 21, 136±144
(1997).
Andersson JLR. J Nucl Med 36, 657±669 (1995).
Talairach J and Tournoux P. A Co-planar Stereotactic Atlas of the Human
Brain. Stuttgart: George Thieme Verlag; 1988.
Friston KJ, Holmes AP, Worsley KJ et al. Human Brain Mapp 2, 189±210
(1995).
Andersson JLR. Neuroimage 6, 237±244 (1997).
Friston KJ, Frith CD, Liddle PF and Frackowiak RSJ. J Cerebr Blood Flow
Metab 11, 690±699 (1991).
Rajah MN, Hussley D, Houle S et al. Neuroimage 7, 314±325 (1998).
Saunders RC, Murray EA and Mishkin M. Neuropsychologia 22, 785±796
(1984).
Elliott R and Dolan RJ. Neuroimage 7, 14±22 (1998).
Whalen PJ, Rauch SL, Etcoff NL et al. J Neurosci 18, 411±418 (1998).
Kelley WM, Miezin FM, McDermott KB et al. Neuron 20, 927±936 (1998).
Posner MI and Raichle ME. Images of Mind. New York: Scienti®c
American Library, 1994.
Fredrikson M, Wik G, Fischer H and Andersson J. Neuroreport 6, 797±
101 (1995).
Fredrikson M, Fischer H and Wik G. J Clin Psychiat 58(suppl 16), 16±21
(1997).