The neural bases of multistable perception.

Abstract

Multistable perception is the spontaneous alternation between two or more perceptual states that occurs when sensory information is ambiguous. Multistable phenomena permit dissociation of neural activity related to conscious perception from that related to sensory stimulation, and therefore have been used extensively to study the neural correlates of consciousness. Here, we review recent work on the neural mechanisms underlying multistable perception and how such work has contributed to understanding the neural correlates of consciousness. Particular emphasis is put on the role of high-level brain mechanisms that are involved in actively selecting and interpreting sensory information, and their interactions with lower-level processes that are more directly concerned with the processing of sensory stimulus properties.

Full text

Page 1

The neural bases of multistable
perception
Philipp Sterzer1, Andreas Kleinschmidt2and Geraint Rees3,4
1Department of Psychiatry, Charite ´ Campus Mitte, Charite ´platz 1, D-10117 Berlin, Germany
2INSERM-CEA Cognitive Neuroimaging Unit, CEA/SAC/DSV/DRM/NeuroSpin, Ba ˆt 145, Point Courrier 156, 91191 Gif/Yvette cedex,
France
3Institute of Cognitive Neuroscience, University College London, 17 Queen Square, London WC1N 3AR, UK
4Wellcome Trust Centre for Neuroimaging, University College London, 12 Queen Square, London WC1N 3BG, UK
Multistable perception is the spontaneous alternation
between two or more perceptual states that occurs
when sensory information is ambiguous. Multistable
phenomena permit dissociation of neural activity
related to conscious perception from that related to
sensory stimulation, and therefore have been used
extensively to study the neural correlates of conscious-
ness. Here, we review recent work on the neural mech-
anisms underlying multistable perception and how
such work has contributed to understanding the neural
correlates of consciousness. Particular emphasis is put
on the role of high-level brain mechanisms that are
involved in actively selecting and interpreting sensory
information, and their interactions with lower-level
processes that are more directly concerned with the
processing of sensory stimulus properties.
Introduction
Multistable perception occurs when sensory information is
ambiguous and consistent with two or more mutually
exclusive interpretations. When no additional cues are
available that allow perceptual synthesis to converge on
one unique interpretation, perception alternates spon-
taneously every few seconds between two (‘bistable’) or
more (‘multistable’) interpretations of the same sensory
input. Well-known examples include the Necker cube,
Rubin’s face-vase illusion, bistable apparent motion and
binocular rivalry (Figure 1). Interest in studying multi-
stable perception in human observers has increased with
the advent of modern non-invasive brain imaging tech-
niques such as functional magnetic resonance imaging
(fMRI) because multistable stimuli allow neural activity
related to conscious perception to be distinguished from
that related to physical stimulus properties. Moreover,
multistable perception can help us understand the con-
structive neural processes that generate a unified and
coherent subjective experience of the world even though
the information available is often fragmentary, conflicting
or even ambiguous.
A decade ago, a thought-provoking article on the neural
basis of multistable perception was published in this jour-
nal [1]. Inspired by the first fMRI studies of bistable
perception, (see Refs [1,2] for reviews) and contrasting
with the traditional view that spontaneous perceptual
reversals are a consequence of antagonistic activity within
the visual system (e.g. Refs [3,4]), the authors proposed
that ‘reorganizations of activity throughout the visual
cortex, concurrent with perceptual reversals, are initiated
by higher, largely non-sensory brain centers’ [1]. Since
then, hybrid theoretical proposals have emerged, on the
basis of behavioural evidence, which conceptualise multi-
stableperception as arisingfrominteractionsbetweenlow-
level (sensory) and high-level (cognitive) processes [5]. In
the last decade, the neural mechanisms of binocular riv-
alry, a special case of perceptual multistability, have been
under intense investigation. Similar hybrid models of bin-
ocular rivalry that involve different visual processing
levels have been proposed [6–8]. Finally, converging evi-
dence from several recent lines of empirical neuroscience
suggests a causal role of frontal and parietal cortex in
generating perceptual switches in multistability, as pre-
viously proposed [1].
Here, we review recent findings addressing the neural
mechanisms of different types of multistable perception,
including but not limited to binocular rivalry. We incorpor-
ate these findings into an integrated view of how different
neural processing levels, ranging from early sensory brain
structures to non-sensory associative cortices, might inter-
act to give rise to multistable perception, and how these
processes are related to conscious perception in general. As
mostworkhasstudiedvisualmultistability,wefocusonthe
visual system. However, there are striking behavioural
similarities with comparable phenomena in other sensory
domains(e.g.auditory[9,10]andtactilemultistability[11]).
The same or similar mechanisms involved in visual multi-
stable perception might thus also have a role in other
sensorydomains(Box1).Ourreviewisstructuredaccording
to the traditional hierarchical view of visual information
processing, starting with subcortical and low-level cortical
visual processing. After discussing intermediate processing
stages at the level of extrastriate visual areas, we will focus
on recent insights into higher-order mechanisms involving
frontal and parietal cortices, and their interactions with
lower-level sensory brain regions.
Subcortical and early cortical visual processing
Several early electrophysiological studies concluded that
neural activity at the anatomically earliest post-retinal
visual processing stages, in the lateral geniculate nucleus
(LGN) and primary visual cortex (V1), had little or no
Review
Corresponding author: Sterzer, P. (philipp.sterzer@charite.de)
310
1364-6613/$ – see front matter ? 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.tics.2009.04.006 Available online 18 June 2009

Page 2

influence on the resolution of binocular rivalry (see Refs
[1,2] for reviews). Subsequently, blood oxygenation level
dependent (BOLD) fMRI studies in humans have repeat-
edly demonstrated strong effects of binocular rivalry on
signals in V1 [12,13], with a close linkage between BOLD
signals and the evolving spatiotemporal dynamics of riv-
alry perception [14]. Moreover, fluctuating perception
during binocular rivalry can be predicted for extended
periods of time from BOLD signals in human V1 alone
[15] when information is accumulated across voxels using
multi-voxel pattern analysis [16,17] (Figure 2a). A recent
magnetoencephalography (MEG) study showed that early
visual cortex activity correlates with the perception not
only in binocular rivalry but also with the figure-ground
reversals underpinning Rubin’s face-vase illusion [18]
(Figure 2b). Transcranial magnetic stimulation over early
visual cortex (V1) can induce perceptual alternations [19],
suggesting a causal role for fluctuations in V1 activity in
Figure 1. Examples of multistable visual phenomena. (a) The Necker cube is a wire-frame drawing of a cube in isometric perspective, which makes which plane is perceived
as being in front ambiguous. This results in perceptual bistability with perception alternating between these two mutually exclusive interpretations of the sensory input. (b)
The ‘spinning wheel illusion’ is an example of bistable apparent motion perception. Two frames showing dashed circles that are offset from each other by the dash length
are shown in rapid alternation. This yields bistable perception of either clockwise or counter-clockwise rotation. (c) Binocular rivalry results from presentation of dissimilar
images to each eye, leading to bistable perceptual alternations between the two images.
Box 1. Auditory multistable perception
Two examples of perceptual multistability in the auditory modality are
auditory stream segregation [79] and the verbal transformation effect
[80]. Auditorystreamsegregation occurswhen twotonesarepresented
alternately or in another repeating temporal pattern (Figure I). These
sequences are perceived either as a single stream of fluctuating sounds
or as two segregated streams, each comprising a single repeating
sound. The temporal dynamics of percept alternations during auditory
streaming are similar to those in bistable visual perception, and
similarly susceptible to volitional control, indicating that auditory and
visual multistable phenomena share common principles [10]. Interest-
ingly, subject-specific biases do not show any correlation across
sensory modalities, suggesting that the neural implementation of
these common principles is at least partly independent for visual and
auditory systems [10]. Neural correlates of auditory streaming have
been observed at different levels of auditory processing. For example,
electrophysiological recordings in monkeys show differential neural
activity primary auditory cortex [81,82] and even in subcortical
structures [83] when comparing the perception of one versus two
auditory streams. In humans, fMRI and magnetoencephalography
show increased activity during the perception of two rather than one
stream in primary auditory cortex and surrounding non-primary
auditory areas [84] and in the intraparietal sulcus [85].
The verbal transformation effect arises when a speech form is
cycled in rapid and continuous repetition [80]. Initially, a percept
matching the original form dominates, whereas at some point
another percept takes over and then alternates with the original
percept. For example, rapid repetitions of the word ‘life’ result in
bistable alternation between the perceived words ‘life’ and ‘fly’. Like
other multistable phenomena, the timing of perceptual transitions for
this verbal transformation effect conforms to gamma distributions [9].
In auditory cortex and prefrontal regions (left inferior frontal cortex
and dorsal anterior cingulate cortex), activity increases during
perceptual transitions. However, only prefrontal activations are
specific for spontaneous changes in speech perception compared to
a control condition where physical stimulus changes were reported.
Although left inferior frontal activations correlate positively with the
number of perceptual transitions, the inverse relationship is found for
cingulate activations, suggesting opposite influences of these two
regions on perceptual alternations. An involvement of left inferior
frontal cortex, in addition to left supramarginal gyrus, in bistable
word perception is also suggested by an increase in gamma activity
(>40 Hz) in these regions before perceptual transitions, as shown
using intracerebral EEG recordings in epileptic patients with diag-
nostic electrode implants [86].
Figure I. Auditory streaming as an example of auditory multistable perception.
Two tones differing in frequency are presented alternately in a repeating
temporal pattern. Listeners perceive the sequence as either one stream with
fluctuating tones or as two segregated streams. The perceptual streams are
indicated by the grey shading.
Review
Trends in Cognitive Sciences Vol.13 No.7
311

Page 3

influencing the outcome of rivalry. Similar to the fMRI
findings in V1, and contrasting with electrophysiological
work (see Ref. [1] for a review), even BOLD signals in the
LGN correlate with the perception of observers during
rivalry [20,21]. Discrepancies between electrophysiological
and fMRI studies can be related to the fact that neuronal
spiking activity (as measured in earlier electrophysiologi-
cal studies) is less closely related to perceptual awareness
than local field potentials [22], which in turn correlate
more strongly with fMRI BOLD signals than spiking
activity [23], thus underlining the usefulness of fMRI for
studying neural correlates of perception. Moreover, con-
text-dependent differences in the coupling between fMRI
and electrophysiological signals in visual cortex could have
a important role [24], in addition to more general differ-
ences in methodologies and species.
fMRI studies in humans have thus established that
signals from anatomically early visual processing stages
including the LGN and V1, reflect the perceptual outcome
ofbinocularrivalry(seeRef.[6]forareview).Togetherwith
earlier electrophysiological findings, they support the
notion that neural activity at these earliest processing
levels, although in itself not being sufficient for generating
a conscious percept, is a prerequisite for visual information
to reach consciousness [25–27]. Yet, although such findings
provide information about which neuronal populations are
involved in resolving binocular rivalry, they are less infor-
mative about how neuronal populations come to prefer just
one perceptual interpretation of a multistable stimulus in
the first place. One interpretative problem arises from the
relatively poor temporal resolution of fMRI. For example,
binocular rivalry itself might evoke different neural mech-
anisms, some primarily related to perceptual conflict, but
others to inter-ocular competition. Changes in perception
can, thus, arise as a by-product of low-level interactions at
monocular processing stages. In this case, perception pas-
sively follows the resolution of ‘inter-ocular’ conflict. Alter-
natively, ‘perceptual’ conflict could drive processes of
inference and interpretation that modulate neural activity
throughout the visual hierarchy down to the lowest levels
(Box 2). For both scenarios, the role of LGN or V1 could be
to gate visual information, either via local competitive
interactions or through modulation by feedback signals
from higher-order areas [27]. Evidence for the latter mech-
anism comes from fMRI studies of bistable apparent
motion [28] and motion-induced blindness [29]. Whenever
apparentmotionisinconsistentwithadditionalimagecues
(e.g. colour), early visual cortex activity is suppressed [28].
In motion-induced blindness, a bistable phenomenon
where a target stimulus surrounded by a field of moving
dotsintermittentlydisappearsfromperceptualawareness,
activity in visual areas V1 to V3 is globally reduced during
target disappearance [29]. Such findings possibly reflect a
gating mechanism whereby feedback from higher-order
areas suppresses those features of input-related activity
in early visual cortex that disagree with the perceptual
outcome.
Extrastriate visual cortex
A fairly congruent picture has emerged of the involvement
of extrastriate visual areas, which lie beyond V1 in the
visual pathway, in multistable perception. Both electro-
physiological binocular rivalry experiments in human and
monkey, plus fMRI experiments in humans reveal strong
correlations between subjective perception and activity in
functionally specialized extrastriate cortex (see Refs
[1,2,30] for reviews). Recent studies have extended our
understanding of the role of extrastriate visual cortex in
conscious perception. The observation that the amplitudes
of percept-related fMRI signal fluctuations during binocu-
lar rivalry in high-level extrastriate visual areas are
similar to those during actual stimulus alternations (see
Refs [1,2] for reviews) initially suggested that the conflict
between incompatible interpretations of visual input had
been fully resolved at this stage of processing, with no
maintained representation of the suppressed stimulus.
However, binocular rivalry is influenced by complex infor-
mation contained in suppressed stimuli [31–34] (see also
Refs [2,6] for reviews), indicating that information from
binocularly suppressed stimuli still processed at suffi-
Figure 2. Neural correlates of multistable perception in human early visual cortex. (a) Functional magnetic resonance imaging BOLD contrast signal patterns from visual
areas V1, V2 and V3 can accurately and blindly predict behavioural reports of perceptual alternations during binocular rivalry between rotating orthogonal gratings [15].
Behavioural reports of alternations between face and vase percepts in the Rubin’s face-vase illusion correlate with activity in early visual cortex measured with
magnetoencephalography (MEG). Specificity of the MEG signals for each percept is achieved by flickering the two components of the bistable image (the faces and the
vase) at different frequencies. These frequency ‘tags’ can be identified in the MEG signal and used to examine neural activity correlating with each the two perceptual
interpretations [18]. Reproduced from Refs [15,18] with permission from Cell Press and National Academy of Sciences, USA.
Review
Trends in Cognitive Sciences Vol.13 No.7
312

Page 4

ciently advanced levels where this information can be
extracted and represented. Indeed, areas outside the ven-
tral visual pathway show responses to binocularly sup-
pressed stimuli: the emotional content of suppressed
stimuli, such as fearful faces, is associated with activity
in the amygdala [35–37] and in the superior temporal
sulcus [35]. Similarly, images of tools evoke activity in
dorsal extrastriate areas, where this object category is
preferentially processed, even during binocular suppres-
sion [38]. But these studies found no differences between
responses to specific object categories (when interocularly
suppressed) in ‘ventral’ extrastriate areas, such as the
fusiform face area (FFA), parahippocampal place area
(PPA) or lateral occipital complex (LOC). However, a
recent high-resolution fMRI study used multi-voxel pat-
tern analysis to show that fine-grained spatial activity
patterns within FFA and PPA still contain information
about the category of binocularly suppressed face and
house stimuli [39], even though there were no significant
differences in the mean response to suppressed faces and
houses in these areas. Thus, the fine-grained spatial pat-
tern of activity measured with fMRI encodes information
about the identity of suppressed stimuli. Similarly, face-
specific electromagnetic responses to binocularly sup-
pressed stimuli are reduced in amplitude but still present
in the human ventral visual pathway [40]. These results
are consistent with more general findings of high-level
processing for stimuli outside awareness in other para-
digms (e.g. see Ref. [30]) and provide a possible neural
basisforhowcomplexstimulusfeaturescouldcontributeto
the resolution of perceptual conflict even when suppressed.
Studies with reversible figures and ambiguous motion
stimuli confirm that activity in extrastriate visual areas
closely reflects conscious perception and, in addition, that
theseareasaredirectlyinvolvedintheresolutionofvisual
ambiguities and conflict. Similar to binocular rivalry,
BOLD signals measured using fMRI from human FFA
are greater during face perception in Rubin’s face-vase
illusion [41,42]. Similarly, viewing of an ambiguous
stimulus whose elements can be perceived as either
grouped or randomly arranged is associated with greater
BOLD activity in the LOC during perception of a coherent
shape [43]. Electrophysiological recordings in macaque
monkeys show percept-specific activity also during bis-
table motion perception in motion-sensitive areas V5/MT
[44], MST and parietal cortex [45]. BOLD contrast fMRI
signals from human motion-sensitive extrastriate areas
such as V5/MT allow accurate prediction of perceptual
states during ambiguous structure-from-motion [46]; and
stimuliambiguouswithrespecttothetypeofmotionorthe
perception of motion versus flicker evoke differential
activity in human motion-sensitive cortex [47–49]. In
contrast to the substantial evidence for an involvement
of V5/MT in bistable motion perception, much less is
known about the role of early extrastriate areas of the
ventral visual stream, for example, V4, in multistable
perception. This is because bistability between features
such as colour is difficult to achieve experimentally. How-
ever, when colour or luminance cues are used to bias
bistable apparent motion perception, activity in V4 is
increased whenever perception is consistent with colour
cues, suggesting that extrastriate colour processing is
sensitivetoperceptualdominance[28].Inmotion-induced
blindness, target visibility is associated with increased
activity in the retinotopic representation of the target in
V4, showing a close relationship between conscious per-
ception and V4 activity [29].
In contrast to neural activity associated with different
perceptual states, another line of research has focused on
neural events associated with perceptual reversals. Rever-
sal-related activity is consistently observed in extrastriate
visual areas and is tuned to the visual feature or attribute
that is perceived to change. Changes involving face or
object percepts are accompanied by activations in object
processing areas of the ventral stream (see Ref. [1] for a
review), whereas perceived changes in motion direction or
Box 2. The ambiguous case of binocular rivalry - a rivalry of rivalries?
Binocular rivalry (Figure 1c in main text) has become a very popular
tool for the investigation of the neural correlates of consciousness
(see Refs [2,6] for reviews), perhaps because it is a particularly
intriguing phenomenon. But more mundane factors such as practical
feasibility and functional anatomy of the visual system can also affect
the choice of experimental paradigm. For example, binocular rivalry is
very flexible with regard to the stimulus material used. In principle,
any stimulus class can be used in binocular rivalry – from simple
grating stimuli optimal for the investigation of low-level processing,
to complex object stimuli that are processed in higher-level visual
areas. The use of object stimuli in binocular rivalry has the additional
advantage that the processing of some object categories is to some
degree spatially segregated in human visual cortex. Responses to two
rivalling object categories (e.g. faces and houses) can therefore easily
be dissociated using fMRI.
However, binocular rivalry potentially differs from other perceptual
ambiguities because there is not only conflict between two interpreta-
tions of one sensory input pattern but also between two different
patterns presented to the two eyes. Thus, in addition to competition
between neural pattern representations, competition between eye-
specific representations at early monocular stages of central visual
processing cancontribute to rivalry.Current modelsofbinocular rivalry
propose a multi-level process involving competitive neural interactions
at both monocular stages (eye rivalry) and binocular stages (pattern
rivalry)ofthevisualprocessinghierarchy[6].Involvementofperceptual
(rather than purely inter-ocular) mechanisms is shown by persistence
of rivalry when the monocular images are rapidly swapped between
eyes, preventing interocular competition, and by the ability of
complementary patchworks of intermingled images presented to each
eye to drive rivalry (see Refs [1,2] for reviews). Neuroimaging studies
have in general used binocular rivalry paradigms that do not explicitly
tap one or other mechanism. Still, binocular rivalry has many
behavioural similarities with other bistable percepts, such as the
temporal characteristics of perceptual alternations. The effects of
intermittent blank periods on rivalry and other forms of bistable
perception are qualitatively identical [54], and the psychophysical
relationships between stimulus strength and alternation dynamics are
also comparable [87], suggesting common computational mechan-
isms.However,therearealsoimportantdifferences.Withothertypesof
multistability, different percepts are mutually exclusive, but this is not
always the case in binocular rivalry. Briefly presented rivalry stimuli
(<500 ms) can be perceived as superimposed on each other; and a
patchy mixture of the two monocular images (’piecemeal rivalry’) can
occur, especially with large stimuli. Furthermore, unlike reversible
figures, it is difficult to wilfully influence perceptual alternations in
binocular rivalry [77]. A possible explanation forsuch differences isthat
the nature of conflict is different between binocular rivalry and other
forms of multistable perception.
Review
Trends in Cognitive Sciences Vol.13 No.7
313

Page 5

motion type are associated with activations in motion-
sensitive areas, most notably V5/MT [47,50–52].
Neural activity in functionally specialized extrastriate
areas also has an important role in determining the
perceptual choice at stimulus onset (e.g. during inter-
mittent presentation of ambiguous stimuli [53]). When
an ambiguous visual stimulus is removed for a few
seconds, the percept upon stimulus reappearance tends
to be the one that was dominant when the stimulus
initially disappeared [54,55]. Such perceptual memory
is proposed to reflect a consequence of sub-threshold
elevation in the baseline activation of neurons that
represent the last percept, thereby influencing the per-
ceptual choice at stimulus reappearance [56]. A recent
fMRI study tested whether perceptual memory during
intermittent binocular rivalry is associated with percept-
specific brain activity during temporary removal of a face
(versus grating) binocular rivalry stimulus [53]. Activity
in the FFA was greater during periods that were pre-
ceded by a face percept (and that were thus more likely
to be followed by a face percept at the next stimulus
onset) than during periods preceded by grating percep-
tion (Figure 3a). Thus, the perceptual choice at stimulus
onset is influenced by the last conscious percept before
stimulus removal, and extrastriate visual cortex seems
to be involved in such perceptual memory (for a review
see Ref. [55]). This is consistent with behavioural obser-
vation that perception of rivalrous stimuli can also be
influenced by imagery [57], which enhances content-
related activity in extrastriate visual cortex [58]. It
should be noted that the perceptual choice at the onset
of rivalrous stimuli is also subject to additional influ-
ences, such as low-level spatially localized factors, other
than those governing perceptual alternations during
ongoing rivalry [59].
But neural activity in extrastriate visual cortex can bias
perceptual choice even when there is no measurable mem-
ory effect; that is, when ambiguous stimuli are briefly
presented at long time intervals and perception shows
stochastic behaviour [60]. Activity in the human FFA
measured before presentation of Rubin’s face-vase figure
is higher when observers subsequently report perceiving a
face instead of a vase, suggesting that endogenous vari-
ations in pre-stimulus neuronal activity bias subsequent
perceptual inference (Figure 3b). A corresponding effect
has been associated with visual motion perception, for the
effect of spontaneous variations in pre-stimulus V5/MT
activity on detection of near-threshold coherence in ran-
dom dot motion kinematograms [61]. Such observations of
apparently spontaneous baseline signal fluctuations and
for carry-over between successive trials due to perceptual
memorybothsuggest that
stimulus-driven processes contributes to how perceptual
conflict is resolved by the human brain.
ongoing activity before
Figure 3. The relationship between neuronal population activity in extrastriate visual areas and perceptual report in multistability. (a) When an ambiguous visual stimulus is
removed after a few seconds of stimulation, the percept upon reappearance of the same ambiguous stimulus tends to be the one that was perceptually dominant when the
stimulus initially disappeared. During removal of a face (versus grating) binocular rivalry stimulus (‘delay’), activity in the fusiform face area (FFA), is greater after face than
grating perception [53]. This percept-specific activity might contribute to the maintenance of a perceptual state across periods of stimulus removal. (b) When Rubin’s face-
vase figure is presented briefly, then masked, and repeated again at long and unpredictable intervals (ISI), perception of vase or face behaves stochastically over trials.
Trials on which participants report a face percept are preceded by greater FFA activity than trials yielding the vase percept [60]. This suggests that the perceptual decision is
determined by spontaneous activity fluctuations in extrastriate visual cortex that bias the outcome of perceptual competition between the mutually exclusive interpretations
of the sensory input. Reproduced from Refs [53,60] with permission from MIT Press and National Academy of Sciences, USA.
Review
Trends in Cognitive Sciences Vol.13 No.7
314

Page 6

Taken together, recent studies have thus clarified how
neural activity in functionally specialized extrastriate
regions relates to conscious perception. It is now firmly
established that neural activity in these areas shows
strong modulations correlated with subjective perception
of ambiguous or rivalrous stimuli. Together with a larger
body of evidence from lesion or microstimulation studies,
this suggests a key role of functionally specialized extra-
striate areas in representing the contents of conscious
visual perception (for a review see Ref. [30]). However, it
should be noted that activity in functionally specialized
areas is not always associated with a conscious percept, as
demonstrated by the presence of stimulus-specific signals
even when stimuli are made completely invisible (e.g. by
binocular suppression). It is an intriguing question for
future research how such unconscious representations of
complex stimulus features can influence the dynamics of
perceptual experience and, ultimately, behaviour.
Parietal and prefrontal cortex
Multistable phenomena can identify neural correlates of
conscious perception by characterizing neuronal popu-
lations that correlate with perception independently of
physical stimulation. Although such studies can, thus,
identify neural processes that reflect the contents of con-
scious perception, they cannot determine how the brain
resolves conflict or ambiguities in the sensory input to
settleononeoftwo(ormore)mutuallyexclusiveperceptual
interpretations. Such an understanding touches upon the
constructive nature of visual perception, which is not a
passive mapping of visual input within the brain but an
active interpretation involving inference. Natural visual
scenes contain many ambiguities and conflicts that usually
go unnoticed because the brain effectively disambiguates
the information received [62]. In such a framework, multi-
stable perception can be conceived of as a frequent re-
evaluation of the current interpretation of the sensory
input, which also occurs during normal vision but becomes
particularly evident when ambiguities are maximised [1].
Such a theoretical account of multistable perception comp-
lements the traditional view of multistability as reflecting
mutual inhibition between different representations of the
sensory input. On the basis of behavioural observations,
such as the stochastic nature of perceptual alternations
and effects of high-level cognitive factors on the dynamics
of multistable perception, it has been argued that an
account solely based on mutual inhibition within sensory
brain areas is unlikely [1]. Investigating neural activity
that is associated with perceptual transitions rather than
perceptual states during multistable perception should
help to understand the inferential processes in visual
perception.
Transient fMRI signal increases associated with per-
ceptual reversals are not only observed in visual cortices
(see earlier) but also in parietal and frontal regions [63]
that have been implicated in visual attention [64] and in
regulating access of sensory information to consciousness
(see Refs. [30,65] for reviews). Although extrastriate areas
are equally engaged by non-rivalrous perceptual changes,
parietal and prefrontal regions show greater activation
during rivalrous perceptual alternations [63]. Such fronto-
parietal activations could reflect top-down processes that
initiate a reorganization of activity in visual cortex during
perceptual reversals [1]. Alternatively, and according with
the traditional view of multistability as a result of neural
activity fluctuations in visual cortex (e.g. Refs [3,4]), fron-
toparietal activations could merely reflect the feed-forward
communication of salient neural events from visual cortex
to higher-order areas, similar to external stimulus
changes. These two scenarios differ in the causal chain
assumed to underlie changes in visual awareness, but it
remains difficult to infer causality from correlative neuro-
physiological measures. Ideally, this question would be
addressed by probing the causal role of frontal and parietal
structuresusingexperimentallesionandmicrostimulation
techniques that have also been used in the study of per-
ceptual decision making (see Ref. [66] for a review). How-
ever, no such studies have been reported to date, although
investigations of multistable perception in patients with
parietal and prefrontal damage after stroke provide sug-
gestive evidence (see later). Alternatively, temporal pre-
cedence is usually considered good evidence in favour of a
putative causal role [1,67].
Chronometric analyses of fMRI signals associated with
spontaneous changes in apparent motion perception show
that activation of right prefrontal cortex precedes that of
V5/MT during bistable perception, relative to externally
induced perceptual changes [68] (Figure 4a). It should be
noted that conclusions from such chronometric analyses of
BOLD contrast fMRI signals – even when appropriately
grounded in demonstrating a region-by-condition inter-
action [68] that removes effects of local variations in neu-
rovascular coupling – are still limited by our incomplete
understanding of the relationship between neural activity
and hemodynamic responses. However, the notion that the
temporal precedence of BOLD activations might suggest a
contributionof prefrontal
dynamics is supported by the finding that perceptual
alternations are slowed in patients with focal damage to
prefrontal [69] and parietal cortex [70] (see Ref. [1] for a
review of earlier lesion studies with similar results). A
similar conclusion was drawn from the electroencephalo-
graphic (EEG) finding that perceptual reversals during
viewing of a Necker cube are preceded by neural activity
in right parietal cortex [71] (Figure 4b), a region that
exhibits stronger BOLD contrast fMRI signals during
perceptual transitions in binocular rivalry compared to
stimulus changes [63].
In addition to being associated with perceptual tran-
sitions during multistability, activity in frontal and par-
ietal cortex can also contribute to percept stabilization. As
mentioned earlier, perception can be stabilized by inter-
mittently removing a multistable stimulus [54,55]. Intri-
guingly, the tendency of an individual observer to stabilize
a percept across such periods of stimulus removal is
strongly correlated with brain activity in frontal and par-
ietal regions [53] previously associated with very different
psychological processes of working memory and atten-
tional selection [72]. Similar neural mechanisms might,
thus, have a role in percept stabilization during multi-
stable perception. The involvement of frontal and parietal
regions in the regulation of percept stability is also
structures toperceptual
Review
Trends in Cognitive Sciences Vol.13 No.7
315

Page 7

suggested by the correlation of activity in these regions
with the tendency to maintain a percept during intermit-
tent bistable motion perception [73].
Although there is a growing body of evidence for involve-
ment of frontal and parietal cortices, their exact role in
multistable perception and the implications for the under-
standing of conscious perception is still subject to specu-
lation (Box 3). Signals associated with perceptual reversals
are seen in frontal and parietal areas previously implicated
in cognitive processes such as attention and working mem-
ory. Perhaps most striking is the similarity between brain
regions implicated in directing attention to salient events
[64] and those activated by spontaneous perceptual rever-
sals [52,63,68,71]. Recent observations that activations in
frontal[68]andparietal[71]corticesprecedeactivityassoci-
ated with the sensory processing of perceptual switches
suggest that feedback signals from frontoparietal regions
can modulate visual processing even in the absence of
triggering sensory events. These studies investigated spon-
taneouslyalternatingperception,withoutanyinstructionto
wilfully influence perception or to attend to particular
stimulus features. Although it is possible to voluntarily
control multistable perception to some extent, this control
seems to be limited [74–78]. Reversal-related activations
found in neuroimaging studies are thus unlikely to reflect
voluntaryattention,eventhoughtheyoccurinsimilarbrain
regions.Onepossibilityisthatregionsinvolvedinvoluntary
attentionalsogeneratefeedbacksignalsspontaneouslyand
automatically. Such feedback signals, if occurring itera-
tively, could serve the frequent re-evaluation of the current
perceptual interpretation [1]. They could, thus, initiate
perceptual reorganizations whenever the ‘balance of power’
between neuronal populations coding for different multi-
stable percepts in sensory cortices destabilizes due to adap-
tation or mutual inhibition. An alternative possibility,
which would also account for temporal precedence of fron-
toparietal activations, is that such putative destabilization
in sensory cortices might serve as a salient event that
Box 3. Outstanding questions
? Is frontal and parietal activity during spontaneous perceptual
reorganizations directly related to cognitive processes that these
brain regions have been implicated in, such as reorienting of
attention?
? What is the neural basis for the effects of cognitive processes like
attention, working memory or intentions and expectations, on
multistable perception?
? Under which conditions do signals from other senses affect
multistable perception in a single modality [88,89], and how might
such influences be mediated neurally?
? Which neural processes mediate the resolution of local ambi-
guities in visual scenes by contextual information?
? How do emotions, motivation and volition influence the resolu-
tion of perceptual ambiguities?
? How does sensory information that is perceptually suppressed
influence behaviour?
Figure 4. Involvement of frontal and parietal brain regions in multistable perception. (a) Alternating presentation of two frames showing two dots in diagonally opposite
corners of a square results in bistable perception (‘rivalry’) of either horizontal or vertical apparent motion (in addition note that occasionally continuous clockwise or
counter-clockwise motion can be perceived in association with these stimuli). Spontaneous perceptual switches between horizontal and vertical perceived directions of
motion are associated with greater BOLD contrast fMRI signals in bilateral inferior prefrontal cortex (shown in red) compared to similar but stimulus-induced perceptual
changes in a control condition where the perceptual sequence during rivalry is replayed using unambiguous stimuli (‘replay’). Right inferior prefrontal activations
associated with spontaneous perceptual changes occur earlier (shown in green) than those associated with stimulus-induced changes, whereas such a difference is absent
in other brain regions, including visual cortex [68]. (b) During intermittent presentation of ambiguous figures with short stimulus durations (<1 s), perceptual reversals
mostly occur at stimulus onsets. This allows analysis of brain activity time-locked to perceptual reversals. Measurements of brain activity with electroencephalography
(EEG) during intermittent presentation of a complex Necker cube show that perceptual reversals are preceded by greater activity in right inferior parietal cortex, compared
to stable perception [71]. Reproduced from Refs [68,71] with permission from National Academy of Sciences, USA, and Oxford University Press.
Review
Trends in Cognitive Sciences Vol.13 No.7
316

Page 8

activates frontoparietal regions in a feed-forward manner.
Frontoparietalregionscouldinturnworktore-directatten-
tion to the sensory input and initiate a re-evaluation of its
current interpretation, eventually leading to a perceptual
reversal.Whetherfrontoparietalcorticesareinvolvedspon-
taneously or as a consequence of sensory destabilization, it
seemslikelythattheirfunctionalroleisnotrestrictedtothe
detection of salient external events, but rather that it
extends into a constructive process of evaluating sensory
input for its perceptual significance. These regions could
thus take an active part in inferential processes that give
rise to the subjectively unitary and unambiguous conscious
perceptionofthesensoryinput.Tomorecloselycharacterize
the causal relationshipsbetweendifferentprocessinglevels
in the brain and how they interact to generate conscious
awareness is an intriguing challenge for future research.
Possible experimental approaches include methods with
greater temporal resolution and experimental lesion or
microstimulation techniques in the context of multistable
perception.
Conclusions
Multistable phenomena have not only helped to elucidate
the neural processes underlying perceptual awareness of
sensory stimuli but also to shape our view on the construc-
tive nature of visual processing that provides us with
unitary perceptual experiences despite many inconsisten-
cies and ambiguities in the sensory input. In contrast to
previous models that explicitly contrast ‘low-level’ (e.g.
Refs [3,4]) or ‘high-level’ mechanisms [1], we and many
others (e.g. Refs [1,5]) now consider multistable perception
to be the product of continuous interactions between ‘low-
level’ (sensory) and ‘high-level’ (frontal and parietal) brain
regions. There is now unequivocal evidence that fluctu-
ations in neuronalpopulation activity at both anatomically
early and later stages of visual processing are strongly
correlated with perception. Such activity fluctuations can
arise from different sources, including top-down modu-
lation, mnemonic processes, adaptation and spontaneous
fluctuations. Conversely, there is an increasing body of
evidence supporting the hypothesis that high-level fronto-
parietal processes continuously re-evaluate the current
interpretation of the sensory input and initiate changes
in subjective perception [1]. Such a causal role of higher-
order processes in initiating perceptual reorganizations is
not irreconcilable with the notion that activity fluctuations
in sensory brain areas have a role in determining percep-
tion. Rather, perceptual alternations could be determined
by the joint effect of local processes embedded into a more
global process. That is, whenever local processes act to
destabilize activity that underpins the currently dominant
percept, higher-order evaluative processes can take effect
and initiate a perceptual reorganization. Future research
using multistable perception and related phenomena
should seek to provide a more detailed account of how
these processes contribute to human consciousness.
Acknowledgements
A.K. receives support from the Agence Nationale de la Recherche
(SPONTACT grant), G.R. from the Wellcome Trust, and P.S. from the
German Research Foundation (Emmy-Noether Programme).
References
1 Leopold, D.A. and Logothetis, N.K. (1999) Multistable phenomena:
changing views in perception. Trends Cogn. Sci. 3, 254–264
2 Blake, R. and Logothetis, N.K. (2002) Visual competition. Nat. Rev.
Neurosci. 3, 13–21
3 Attneave, F. (1971) Multistability in perception. Sci. Am. 225, 63–71
4 Blake, R. (1989) A neural theory of binocular rivalry. Psychol. Rev. 96,
145–167
5 Long, G.M. and Toppino, T.C. (2004) Enduring interest in perceptual
ambiguity: alternating views of reversible figures. Psychol. Bull. 130,
748–768
6 Tong, F. et al. (2006) Neural bases of binocular rivalry. Trends Cogn.
Sci. 10, 502–511
7 Freeman, A.W. (2005) Multistage model for binocular rivalry. J.
Neurophysiol. 94, 4412–4420
8 Wilson, H.R. (2003) Computational evidence for a rivalry hierarchy in
vision. Proc. Natl. Acad. Sci. U. S. A. 100, 14499–14503
9 Kondo, H.M. and Kashino, M. (2007) Neural mechanisms of auditory
awareness underlying verbal transformations. Neuroimage 36, 123–
130
10 Pressnitzer, D. and Hupe, J.M. (2006) Temporal dynamics of auditory
and visual bistability reveal common principles of perceptual
organization. Curr. Biol. 16, 1351–1357
11 Carter,O.etal.(2008)Tactilerivalry demonstratedwithanambiguous
apparent-motion quartet. Curr. Biol. 18, 1050–1054
12 Polonsky, A. et al. (2000) Neuronal activity in human primary visual
cortex correlates with perception during binocular rivalry. Nat.
Neurosci. 3, 1153–1159
13 Tong, F. and Engel, S.A. (2001) Interocular rivalry revealed in the
human cortical blind-spot representation. Nature 411, 195–199
14 Lee, S.H. et al. (2005) Traveling waves of activity in primary visual
cortex during binocular rivalry. Nat. Neurosci. 8, 22–23
15 Haynes, J.D. and Rees, G. (2005) Predicting the stream of
consciousness from activity in human visual cortex. Curr. Biol. 15,
1301–1307
16 Haynes, J.D. and Rees, G. (2006) Decoding mental states from brain
activity in humans. Nat. Rev. Neurosci. 7, 523–534
17 Norman, K.A. et al. (2006) Beyond mind-reading: multi-voxel pattern
analysis of fMRI data. Trends Cogn. Sci. 10, 424–430
18 Parkkonen, L. et al. (2008) Early visual brain areas reflect the percept
of anambiguousscene.Proc. Natl.Acad. Sci.U. S. A. 105,20500–20504
19 Pearson, J. et al. (2007) The effects of transcranial magnetic
stimulation on visual rivalry. J. Vis. 7, 1–11
20 Haynes, J.D. et al. (2005) Eye-specific effects of binocular rivalry in the
human lateral geniculate nucleus. Nature 438, 496–499
21 Wunderlich,K.etal.(2005)Neuralcorrelatesofbinocularrivalryinthe
human lateral geniculate nucleus. Nat. Neurosci. 8, 1595–1602
22 Wilke, M. et al. (2006) Local field potential reflects perceptual
suppression in monkey visual cortex. Proc. Natl. Acad. Sci. U. S. A.
103, 17507–17512
23 Logothetis, N.K. et al. (2001) Neurophysiological investigation of the
basis of the fMRI signal. Nature 412, 150–157
24 Maier, A. et al. (2008) Divergence of fMRI and neural signals in V1
duringperceptualsuppressionintheawakemonkey.Nat.Neurosci.11,
1193–1200
25 Lamme, V.A. and Roelfsema, P.R. (2000) The distinct modes of vision
offered by feedforward and recurrent processing. Trends Neurosci. 23,
571–579
26 Pollen, D.A. (2003) Explicit neural representations, recursive neural
networks and conscious visual perception. Cereb. Cortex 13, 807–814
27 Tong, F. (2003) Primary visual cortex and visual awareness. Nat. Rev.
Neurosci. 4, 219–229
28 Sterzer, P. and Kleinschmidt, A. (2005) A neural signature of colour
and luminance correspondence in bistable apparent motion. Eur. J.
Neurosci. 21, 3097–3106
29 Donner, T.H. et al. (2008) Opposite neural signatures of motion-
induced blindness in human dorsal and ventral visual cortex. J.
Neurosci. 28, 10298–10310
30 Rees, G. (2007) Neural correlates of the contents of visual awareness in
humans. Philos. Trans. R. Soc. Lond. B Biol. Sci. 362, 877–886
31 Jiang, Y. et al. (2006) A gender- and sexual orientation-dependent
spatialattentionaleffectofinvisibleimages.Proc.Natl.Acad.Sci.U.S.
A. 103, 17048–17052
Review
Trends in Cognitive SciencesVol.13 No.7
317

Page 9

32 Jiang, Y. et al. (2007) Processing of invisible stimuli: advantage of
upright faces and recognizable words in overcoming interocular
suppression. Psychol. Sci. 18, 349–355
33 Alais,D.andParker,A.(2006)Independentbinocularrivalryprocesses
for form and motion. Neuron 52, 911–920
34 Andrews, T.J. and Blakemore, C. (1999) Form and motion have
independent access to consiousness. Nat. Neurosci. 2, 405–406
35 Jiang, Y. and He, S. (2006) Cortical responses to invisible faces:
dissociating subsystems for facial-information processing. Curr. Biol.
16, 2023–2029
36 Pasley, B.N. et al. (2004) Subcortical discrimination of unperceived
objects during binocular rivalry. Neuron 42, 163–172
37 Williams, M.A. et al. (2004) Amygdala responses to fearful and happy
facial expressions under conditions of binocular suppression. J.
Neurosci. 24, 2898–2904
38 Fang, F. and He, S. (2005) Cortical responses to invisible objects in
the human dorsal and ventral pathways. Nat. Neurosci. 8, 1380–
1385
39 Sterzer, P. et al. (2008) Fine-scale activity patterns in high-level visual
areas encode the category of invisible objects. J. Vis. 8, 1–12
40 Sterzer, P. et al. (2009) Electromagnetic responses to invisible face
stimuli during binocular suppression. Neuroimage 46, 803–808
41 Andrews, T.J. et al. (2002) Activity in the fusiform gyrus predicts
conscious perception of Rubin’s vase-face illusion. Neuroimage 17,
890–901
42 Hasson, U. et al. (2001) Vase or face? A neural correlate of shape-
selective grouping processes in the human brain. J. Cogn. Neurosci. 13,
744–753
43 Fang, F. et al. (2008) Perceptual grouping and inverse fMRI activity
patterns in human visual cortex. J. Vis. 8, 2.1–9
44 Dodd, J.V. et al. (2001) Perceptually bistable three-dimensional figures
evoke high choice probabilities in cortical area MT. J. Neurosci. 21,
4809–4821
45 Williams,Z.M.etal.(2003)Parietalactivityandtheperceiveddirection
of ambiguous apparent motion. Nat. Neurosci. 6, 616–623
46 Brouwer, G.J. and van Ee, R. (2007) Visual cortex allows prediction of
perceptualstatesduringambiguous
Neurosci. 27, 1015–1023
47 Castelo-Branco, M. et al. (2002) Activity patterns in human motion-
sensitive areas depend on the interpretation of global motion. Proc.
Natl. Acad. Sci. U. S. A. 99, 13914–13919
48 Kleinschmidt, A. et al. (2002) The neural structures expressing
perceptual hysteresis in visual letter recognition. Neuron 34, 659–666
49 Muckli, L. et al. (2002) Apparent motion: event-related functional
magnetic resonance imaging of perceptual switches and States. J.
Neurosci. 22, RC219
50 Kleinschmidt, A. et al. (2002) Neural correlates of visual-motion
perception as object- or self-motion. Neuroimage 16, 873–882
51 Sterzer, P. et al. (2003) Responses of extrastriate cortex to switching
perception of ambiguous visual motion stimuli. Neuroreport 14, 2337–
2341
52 Sterzer, P. et al. (2002) Neural correlates of spontaneous direction
reversals in ambiguous apparent visual motion. Neuroimage 15, 908–
916
53 Sterzer, P. and Rees, G. (2008) A neural basis for percept stabilization
in binocular rivalry. J. Cogn. Neurosci. 20, 389–399
54 Leopold, D.A. et al. (2002) Stable perception of visually ambiguous
patterns. Nat. Neurosci. 5, 605–609
55 Pearson, J. and Brascamp, J. (2008) Sensory memory for ambiguous
vision. Trends Cogn. Sci. 12, 334–341
56 Noest, A.J. et al. (2007) Percept-choice sequences driven by interrupted
ambiguous stimuli: a low-level neural model. J. Vis. 7, 10
57 Pearson, J. et al. (2008) The functional impact of mental imagery on
conscious perception. Curr. Biol. 18, 982–986
58 Ishai, A. et al. (2000) Distributed neural systems for the generation of
visual images. Neuron 28, 979–990
59 Carter, O. and Cavanagh, P. (2007) Onset rivalry: brief presentation
isolates an early independent phase of perceptual competition. PLoS
One 2, e343
structure-from-motion.J.
60 Hesselmann, G. et al. (2008) Spontaneous local variations in ongoing
neural activity bias perceptual decisions. Proc. Natl. Acad. Sci. U. S. A.
105, 10984–10989
61 Hesselmann, G. et al. (2008) Ongoing activity fluctuations in hMT+ bias
the perception of coherent visual motion. J. Neurosci. 28, 14481–14485
62 Leopold, D.A. et al. (2005) Binocular rivalry and the illusion of
monocular vision. In Binocular Rivalry (Alais, D. and Blake, R.,
eds), pp. 231–258, MIT Press
63 Lumer, E.D. et al. (1998) Neural correlates of perceptual rivalry in the
human brain. Science 280, 1930–1933
64 Corbetta, M. and Shulman, G.L. (2002) Control of goal-directed and
stimulus-driven attention in the brain. Nat. Rev. Neurosci. 3, 201–215
65 Dehaene, S. (2004) Neural mechanisms for access to consciousness. In
The Cognitive Neurosciences (3rd edn) (Gazzaniga, M.S., ed.), pp. 1145–
1158, MIT press
66 Gold, J.I. and Shadlen, M.N. (2007) The neural basis of decision
making. Annu. Rev. Neurosci. 30, 535–574
67 Parker, A.J. and Krug, K. (2003) Neuronal mechanisms for the
perception of ambiguous stimuli. Curr. Opin. Neurobiol. 13, 433–439
68 Sterzer, P. and Kleinschmidt, A. (2007) A neural basis for inference in
perceptual ambiguity. Proc. Natl. Acad. Sci. U. S. A. 104, 323–328
69 Windmann, S. et al. (2006) Role of the prefrontal cortex in attentional
control over bistable vision. J. Cogn. Neurosci. 18, 456–471
70 Bonneh, Y.S. et al. (2004) Abnormal binocular rivalry in unilateral
neglect: evidencefora non-spatial
Neuroreport 15, 473–477
71 Britz, J. et al. (2009) Right parietal brain activity precedes perceptual
alternation of bistable stimuli. Cereb. Cortex 19, 55–65
72 Haxby, J.V. et al. (2000) Distinguishing the functional roles of multiple
regions in distributed neural systems for visual working memory.
Neuroimage 11, 145–156
73 Raemaekers, M. et al. (2009) Widespread fMRI activity differences
between perceptual states in visual rivalry are correlated with
differences in observer biases. Brain Res. 1252, 161–171
74 Chong, S.C. et al. (2005) Endogenous attention prolongs dominance
durations in binocular rivalry. J. Vis. 5, 1004–1012
75 Hancock, S. and Andrews, T.J. (2007) The role of voluntary and
involuntary attention in selecting perceptual dominance during
binocular rivalry. Perception 36, 288–298
76 Kohler, A. et al. (2008) Deciding what to see: the role of intention and
attentionintheperceptionofapparentmotion.VisionRes.48,1096–1106
77 Meng, M. and Tong, F. (2004) Can attention selectively bias bistable
perception? Differences between binocular rivalry and ambiguous
figures. J. Vis. 4, 539–551
78 van Ee, R. et al. (2005) Voluntary control and the dynamics of
perceptual bi-stability. Vision Res. 45, 41–55
79 Bregman, A.S. (1990) Auditory Scene Analysis. MIT Press
80 Warren, R.M. and Gregory, R.L. (1958) An auditory analogue of the
visual reversible figure. Am. J. Psychol. 71, 612–613
81 Fishman, Y.I. et al. (2001) Neural correlates of auditory stream
segregation in primary auditory cortex of the awake monkey. Hear.
Res. 151, 167–187
82 Micheyl, C. et al. (2005) Perceptual organization of tone sequences in
the auditory cortex of awake macaques. Neuron 48, 139–148
83 Pressnitzer, D. et al. (2008) Perceptual organization of sound begins in
the auditory periphery. Curr. Biol. 18, 1124–1128
84 Gutschalk, A. et al. (2005) Neuromagnetic correlates of streaming in
human auditory cortex. J. Neurosci. 25, 5382–5388
85 Cusack, R.(2005) The intraparietal
organization. J. Cogn. Neurosci. 17, 641–651
86 Basirat, A. et al. (2008) Parieto-frontal gammaband activity during the
perceptual emergence of speech forms. Neuroimage 42, 404–413
87 Klink, P.C. et al. (2008) General validity of Levelt’s propositions reveals
common computational mechanisms for visual rivalry. PLoS One 3,
e3473
88 Hupe, J.M. et al. (2008) Bistability for audiovisual stimuli: perceptual
decision is modality specific. J. Vis. 8, 1.1–15
89 Sato, M. et al. (2007) Visual contribution to the multistable perception
of speech. Percept. Psychophys. 69, 1360–1372
mechanism of extinction.
sulcusand perceptual
Review
Trends in Cognitive Sciences Vol.13 No.7
318