for these signals. The neural mechanisms that dissociate level of awareness from activation in these regions remain unknown. Using
signals of fear depends on mode of functional connectivity in amygdala pathways rather than discrete patterns of activation in these
that reentrant feedback may be necessary to afford such awareness. In contrast, responses to fear in the absence of awareness were
Human emotions are more easily influenced when we are not
consciously aware of the influence (Zajonc, 1980). Fearful ex-
pressions are biologically salient signals of potential danger (Ek-
man, 1982), with the capacity to influence our physiology in the
absence of awareness (Hatfield et al., 1992; Zald, 2003), but the
perception of these signals remain unknown.
Neuroimaging studies have considered whether distinct neu-
ral systems determine responses to consciously “seen” versus
“unseen” signals of fear. The amygdala is implicated in fear pro-
cessing regardless of awareness (Hatfield et al., 1992; Zald, 2003)
but may be recruited by parallel visual inputs. In response to
masked fearful faces, the amygdala is activated together with the
superior colliculus and pulvinar (Morris et al., 1998; Whalen et
al., 1998; Liddell et al., 2005), components of a direct subcortical
pathway that bypasses the visual (striate) cortex (Linke et al.,
1999). Consciously seen fear, conversely, elicits extensive visual
2005), consistent with an indirect striate–lateral geniculate nu-
cleus (LGN) pathway to the amygdala.
Other findings indicate that a simple subcortical–cortical di-
vision is not sufficient to dissociate levels of awareness. Early
visual association areas of the ventral processing stream are en-
gaged by both unseen (Liddell et al., 2005; Williams et al., 2006c)
and seen (Hariri et al., 2003; Das et al., 2005) fear stimuli. These
2003). Regardless of level of awareness, fear signals also elicit
medial prefrontal cortex (MPFC) activation, although noncon-
scious perception may be distinguished by more rostral ventral
activity and conscious fear by more prolonged activity in a com-
paratively dorsal portion (Williams et al., 2006c).
level of awareness and neural systems. Blind-sight patients can-
not consciously report seeing fearful faces but nonetheless dis-
criminate them in forced-choice tasks (de Gelder et al., 1999).
Conversely, parietal lesion patients show residual activity in stri-
ence of extinguished fearful faces (Vuilleumier et al., 2002). This
evidence indicates that striate–LGN input to the amygdala is not
sufficient to afford awareness for fear and that lack of such input
(Morris et al., 1999, 2001) is not sufficient to account for loss of
Connectivity between activated brain regions may be neces-
sary for determining awareness (Crick, 1984; Tononi and Edel-
man, 1998; Lamme and Roelfsema, 2000). In the visual system,
activation, supported by feedforward connections along a direct
superior colliculus–pulvinar pathway and that is too rapid for
This work was supported in part by Australian Research Council Grants DP0452237 and DP0345481 (L.M.W.,
Correspondence should be addressed to Assoc. Prof. Leanne M. Williams, Brain Dynamics Centre, Westmead
Millenium Institute, Westmead Hospital, Westmead, New South Wales, 2145, Australia. E-mail:
9264 • TheJournalofNeuroscience,September6,2006 • 26(36):9264–9271
feedback reentrants to exert an effect. In contrast, reentrant and
lateral feedback occurs at slower timescales, supported by the
indirect cortical pathway (Lamme and Roelfsema, 2000). Here,
we drew onthismodeltotestthehypothesisthatdistinctmodesof
functional connectivity within cortical and subcortical amygdala
females, seven males; mean ? SD age of 35.80 ? 9.04 years; range of
21–48 years) participated in the study [in collaboration with the Brain
Resource International Database (Gordon, 2003; Gordon et al., 2005)].
We have shown previously that neural responses to fear are relatively
the normal range of tested intelligence (mean of 105) based on the Spot-
the-Word estimate of intelligence quotient (Baddeley et al., 1993). Ex-
clusion criteria included Axis-I psychiatric
diagnosis, brain injury [via radiological assess-
(MRI) scans], history of loss of consciousness
der or genetic disorder, and substance abuse.
All participants provided written informed
consent to participate, in accordance with Medi-
Threshold setting for behavioral task. The pa-
rameters for presenting face stimuli under un-
ditions were based on signal detection findings
from an initial psychophysics study (Williams
bers of fear, neutral, and blank screen stimuli
were presented at durations below and above
20 ms, each followed immediately by a neutral
face mask. Stimulus series were presented in
counterbalanced order, and, for each duration,
subjects were asked to respond (via button
press) if they could (1) detect the presence of a
face versus a blank screen stimulus and (2) dis-
criminate the valence (fear vs neutral) of the
to detect the presence of face versus blank
screen stimuli did not differ from chance level.
At the duration of 170 ms and above, stimulus
detection and emotion discrimination were
well above chance level (Williams et al., 2004).
These psychophysics data provided the thresh-
olds for nonconscious and conscious percep-
tion of fearful faces, respectively.
(four female and four male individuals) were
selected from a standardized series of grayscale
ity and size. Facial signals of fear have been
found to elicit more robust amygdala activa-
stimuli were presented under both conscious
based on the thresholds determined in the
above psychophysics study (Williams et al.,
2004). In the conscious perception condition,
face stimuli were presented for 500 ms, well
above the threshold for conscious discrimina-
tion. In the nonconscious perception condi-
tion, stimulus duration was 16.7 ms [less than
the 20 ms threshold established previously
(Williams et al., 2004)], followed immediately
by a neutral face mask (150 ms). The presenta-
tion of the mask after a very short delay (?20
ms) meant that it would not impact the initial feedforward sweep but
would interrupt reentrant feedback from higher areas (Lamme and Ro-
Experimental software ensured that stimulus onset was synchronized
fear and 120 neutral), presented in a pseudorandom sequence of 30
mini-blocks (containing eight fear or eight neutral stimuli each). In the
nonconscious condition, a stimulus was defined as a stimulus–mask
pairing. The interstimulus interval (ISI) was 767 ms in the conscious
perception condition and 1100 ms in the nonconscious condition to
ensure that the total stimulation period (stimulus plus ISI) was equiva-
by ?500 ms to ensure that stimulus onset did not coincide with a con-
stant slice position during image acquisition.
Instructions were presented in synchronized visual and audio
(through headphones) format. Subjects were asked to actively attend to
Williamsetal.•AmygdalaConnectivityandAwarenessforFear J.Neurosci.,September6,2006 • 26(36):9264–9271 • 9265
the face stimuli, in preparation for a postscanning briefing about these
to focus on the first face in particular.
scious as well as conscious fear stimuli, including conditioned fear, have
been found to elicit skin conductance responses (SCRs) (Ohman and
Soares, 1998; Williams et al., 2004, 2006c), we recorded SCRs in this
study as an independent index that physiological responses were pro-
duced in the absence of awareness. SCRs were recorded simultaneously
using a pair of silver–silver chloride electrodes with 0.05 M sodium chlo-
ride gel placed on the distal phalanges of digits II and III of the left hand.
The electrode pairs were supplied by a constant voltage, and the current
change representing conductance was recorded using the direct current
condition, an event was defined by the target face in each target–mask
ical model that allows each SCR to be linked to the individual eliciting
stimulus and potentially overlapping SCRs in short interstimulus inter-
val paradigms to be disentangled (Lim et al., 1997). In both conscious
and nonconscious conditions, there was an average of 16 SCR events.
SCR amplitude was analyzed using within-subjects, repeated-measures
vs neutral) as within-subject factors. Paired t tests were used to explore
the a priori contrasts of interest.
Functional MRI acquisition. Imaging was performed on a 1.5 T Sie-
mens Vision Plus scanner using an echo planar protocol. A total of 90
functional T2*-weighted volumes (three per stimulus block) were ac-
quired: 15 noncontiguous slices parallel to the intercommissural (ante-
rior commissure–posterior commissure) line, with 6.6 mm thickness;
acquired to ensure blood oxygen level-dependent (BOLD) saturation.
Analysis of signal-to-noise ratio has demonstrated previously that this
functional imaging protocol is able to elicit robust signal change in the
amygdala (Williams et al., 2006a) and other subcortical regions (Liddell
et al., 2005).
Image processing and analysis. Preprocessing and statistical analysis
was implemented within the statistical parametric mapping software
package SPM2 (http://www.fil.ion.ucl.ac.uk/spm/software). For each
subject, functional images were first corrected for susceptibility-by-
movement artifacts and then realigned to the first volume of the time
series. Realigned images were spatially normalized into the standard ste-
a Gaussian kernel (full-width half-maximum of 8 mm) to minimize
anatomical differences. The experimental sequences (conscious/non-
response function with temporal derivative, and a high-pass filter was
applied to remove low-frequency fluctuations in the BOLD signal.
the conscious and nonconscious perception conditions separately; (2) a
tional connectivity analysis was undertaken to identify the relationships
between regions of activation in response to fear (vs neutral) for the
conscious and nonconscious conditions considered separately.
In each stage of analysis, we used a priori, hypothesis-driven search
regions of interest (ROIs). To define the search ROIs, we used standard-
mated anatomical labeling protocol (Tzourio-Mazoyer et al., 2002). To
test our hypotheses, we focused on the bilateral amygdala, brainstem,
visual regions (fusiform and inferior occipital gyri, encompassing BA18
and BA37), and medial prefrontal regions, including the anterior cingu-
late cortex (ACC) (encompassing BA24, BA25, and BA32, and medial
functional connectivity (Das et al., 2005).
Within-condition activation analyses. Brain activation for each search
ROI was first examined for the contrast of fear relative to neutral face
stimuli, within both conscious and nonconscious perception conditions
5 voxels per cluster, consistent with previous studies of fear processing
second-level random effects model to determine regions of significant
activation in response to fear, within conscious and nonconscious con-
Conjunction analyses. We used conjunction analysis to determine the
search ROIs that were significantly activated during both conscious and
nonconscious perception of fear versus the neutral baseline. These two
experimental conditions were entered into the conjunction analysis.
Functional MRI connectivity analysis. Psychophysiological interaction
(PPI) analysis is used to assess how activity in brain regions of interest
may covary together in relation to a source region, in response to the
experimental condition (Friston et al., 1997). This method of assessing
condition-dependent covariation has been used as an index of “func-
examined functional covariation specific to fear (vs the neutral baseline)
within both conscious and nonconscious conditions using the specifi-
cally developed function in SPM2. Within each condition, we examined
functional connectivity between the amygdala as the source region of
interest and each search ROI shown to be significant in the within-
condition activation analyses. Connectivity analyses were undertaken
separately for the left and right amygdala.
PPI analysis relied on two first-order effect terms, representing the
time series of activity in the amygdala source region and the condition
(conscious and nonconscious fear vs neutral). Time series were based
on the first eigenvariate, extracted from a sphere of 6 mm radius
centered on the most significantly active voxel revealed in the ROI
contrasts. The interaction between the amygdala time series and con-
dition represented the second-order effect. Before creating this inter-
action term, the BOLD signal was deconvolved with the hemody-
namic response function to represent the interaction at the neuronal
rather than hemodynamic level.
This interaction term was fitted at each voxel in the other search ROIs
using the contrast [1 0 0]. The resulting individual contrast images were
we used a statistical threshold of p ? 0.01 (uncorrected), with an extent
threshold of at least 5 voxels per cluster. This procedure produces a
nonconscious perception conditions. The amplitude of SCRs is shown in microsiemens. Fear
9266 • J.Neurosci.,September6,2006 • 26(36):9264–9271 Williamsetal.•AmygdalaConnectivityandAwarenessforFear
regression coefficient for the interaction term as a measure of PPI. A
significant coefficient indicates significant functional connectivity be-
tween the amygdala source region and the ROI in response to conscious
and nonconscious fear (relative to neutral), and the direction of this
coefficient indicates the direction (positive or negative) of the covaria-
tion. However, it is noted that, because this is a correlational technique,
the causal direction of covariation cannot be determined.
In this study, we also used physiophysiological interaction analysis (a
to examine how the interaction between components of the cortical and
subcortical visual pathways covaries with activation in the other regions
of interest. This analysis focuses on the “physiological” interaction be-
tween two brain regions rather than the interaction between one brain
region and the “psychological” impact of the condition manipulation.
Thus, in the analysis steps outlined above, the interaction between the
time series for the amygdala source region and the condition was re-
placed by the interaction between two time series. For the conscious
condition, we examined the effect of the interaction between the striate
effect of the interaction between the thalamic pulvinar and brainstem su-
perior colliculus. Time series (using the first eigenvariate) were extracted
In postscan briefings, subjects confirmed they were unable to
with above chance accuracy. In contrast, consciously presented
expressions were recognized with well above chance accuracy
(fear, 83%; neutral, 78%), indicating that
any differential effects in brain activity
were unlikely to be attributable to visual
processing or discrimination difficulties.
Fearful faces elicited significantly en-
hanced SCRs (relative to neutral) in both
conditions (Fig. 1), although this en-
hancement was most pronounced in re-
sponse to conscious fear, reflected in the
trend level interaction between condition
and stimulus (F(28)? 2.74; p ? 0.07).
Contrasts confirmed that SCRs to con-
scious fear were significantly greater than
those to neutral (t(14)? 4.28; p ? 0.001),
whereas there was a trend toward a differ-
tral (t(14)? 1.89; p ? 0.08). These SCR
data provide independent evidence that
physiological responses occur in the ab-
sence of awareness.
Consciously viewed fear elicited signifi-
dala (and right amygdala at trend level),
bilateral thalamus (including a region
consistent with the LGN), and striate cor-
tex (Table 1, Fig. 2). Significant, albeit less
pronounced, activation was also revealed
in a brainstem region corresponding to
the superior colliculus, suggesting parallel
input from the subcortical visual pathway
(Table 1, Fig. 2). Additional cortical acti-
vation was observed in the extrastriate vi-
sual cortices [both inferior occipital gyrus (GOi) and fusiform
including ACC (Table 1, Fig. 2).
tivity in corresponding subcortical regions, including the bilat-
eral amygdala, pulvinar region of the thalamus, and brainstem
region consistent with the superior colliculus (Table 1, Fig. 2).
conscious fear, it was nonetheless observed in extrastriate re-
gions, GOi and Gf, and ventral rostral MPFC (Table 1, Fig. 2).
Using conjunction analysis, we determined which regions of ac-
tivation were common to both conscious and nonconscious fear
(relative to neutral). In common, conscious and nonconscious
fear elicited activity in the amygdala, thalamus (encompassing
regions consistent with LGN and pulvinar), brainstem, and ex-
trastriate GOi and Gf regions, albeit primarily at marginal levels
of significance (Table 1, Fig. 3). Small regions of common, trend
level activation were also observed in a dorsal portion of the
MPFC and ACC (Table 1, Fig. 3), although activation in these re-
gions reached significance only in response to conscious fear for
and extrastriate (including fusiform and inferior occipital gyri). Activations were also observed in the superior colliculus and
hypothalamus regions of the brainstem. For nonconscious fear, significant ( p ? 0.05, small-volume corrected) activity was
Williamsetal.•AmygdalaConnectivityandAwarenessforFearJ.Neurosci.,September6,2006 • 26(36):9264–9271 • 9267
Using PPI analysis, we examined func-
tional connectivity between the amygdala
well as the medial prefrontal and visual
association cortices. For conscious fear,
the interaction between striate visual cor-
tex and thalamic LGN covaried negatively
ing negative neuromodulation within the
indirect cortical pathway (Table 2, Fig. 4).
Activation in the bilateral amygdala also
covaried negatively with both the striate
the LGN for the right amygdala (Table 2,
Fig. 4). Similarly, in the parallel subcorti-
cal pathway, the amygdala showed nega-
of the brainstem.
During conscious fear, connectivity
between the amygdala and visual associa-
tion and medial prefrontal cortices varied
with both laterality and dorsoventral gra-
dient. Whereas the left amygdala showed
negative functional connectivity with ex-
trastriate visual regions, the right amyg-
dala covaried positively with the right ex-
trastriate fusiform gyrus (Table 2, Fig. 4).
We observed negative connections be-
tween the right amygdala and both dorsal
MPFC and ventral ACC but positive con-
nections between bilateral amygdala and
dorsal ACC (Table 2, Fig. 4).
In contrast to conscious fear, PPI anal-
ysis revealed a predominance of positive
tical visual pathway, the interaction of ac-
tivation in brainstem superior colliculus
and thalamic pulvinar showed positive functional connectivity
with the right amygdala (Table 3, Fig. 4). The bilateral amygdala
also covaried positively with the thalamic pulvinar, and the right
amygdala showed additional positive covariation with the mid-
functional connectivity between the striate cortex and amygdala,
confirming the preferential reliance on positive connections
within the subcortical pathway.
Nonconscious fear also elicited distinctive profiles of func-
tal association cortices. Activation in the bilateral amygdala co-
MPFC (as opposed covariation with the dorsal portion for con-
tional connectivity between the right amygdala and ventral MPFC
tivity between the right amygdala and bilateral extrastriate inferior
These findings demonstrate for the first time that the mode of
functional connectivity in cortical and subcortical amygdala
pathways, and not the pattern of activation in these pathways,
provides a striking dissociation of the level of awareness for bio-
logically significant signals of fear.
In common, conscious and nonconscious perception of fear
elicited activation in the amygdala, brainstem, thalamus, and ex-
trastriate visual cortices, along with a small region of the MPFC
ness for fear were observed in the striate and prefrontal cortices.
The striate visual cortex was engaged only by consciously per-
ceived fear signals, consistent with a cortical pathway to the
brainstem superior colliculus, pulvinar, and extrastriate regions.
Conscious perception of fear was also distinguished by pro-
nounced dorsal MPFC/ACC activity, extending ventrally for the
ACC, whereas MPFC activation localized to the ventral portion
lesion evidence that facial signals of fear elicit amygdala activity
even in the absence of awareness (Morris et al., 1998; Whalen et
al., 1998; Liddell et al., 2005; Williams et al., 2006c). They also
accord with the view that visual input to the amygdala relies on
the preferential engagement of parallel ventral and dorsal path-
ways, according to level of awareness (Morris et al., 1999, 2001;
Williams et al., 2006c). However, the overlap in regions of acti-
fear (vs neutral). In conjunction, these conditions showed significant (p ? 0.05, small-volume corrected) activations in the
Small regions of common, trend-level activation were also observed in a dorsal portion of the MPFC and ACC. There were no
9268 • J.Neurosci.,September6,2006 • 26(36):9264–9271 Williamsetal.•AmygdalaConnectivityandAwarenessforFear
vation indicates it is unlikely that level of awareness for fear is
determined solely by recruitment of localized amygdala
Psychophysiological and physiophysiological interaction
analyses indicated that it is the mode of
ways that is crucial to determining level of
awareness. Conscious attention to fear
elicited negative functional connectivity
contrast, we observed only positive con-
nectivity within the subcortical pathway
to the amygdala in the absence of aware-
ness. Although BOLD activity changes
over the slower timescale of neuromodu-
lator action, rather than the millisecond
timescale of neural coupling, evidence
concerning the relationship between neu-
ral firing and subsequent hemodynamic
changes is accumulating (Attwell and Ia-
decola, 2002; Logothetis, 2003). In this
context, opposing functional relation-
activity may reflect a different balance of
excitatory and inhibitory neuromodula-
tory action on synaptic activity. Although
it is also not possible to infer causal direc-
tion from the connectivity analysis, this
proposal also accords with the distinct
modes of neural cooperation described in
models of visual consciousness (Lamme
and Roelfsema, 2000). That is, negative
liance of conscious fear perception on re-
entrant feedback, whereas positive con-
nectivitymay reflectthe excitatory
feedforward connections that support processing of fear signals
in the absence of awareness.
Reentrant feedback to posterior visual areas is thought to be
of interest during conscious fear (A) and nonconscious fear (B) relative to neutral. Red arrows represent positive functional
left and right amygdala. Negative connectivity was also observed between the bilateral amygdala and other cortical regions
Williamsetal.•AmygdalaConnectivityandAwarenessforFearJ.Neurosci.,September6,2006 • 26(36):9264–9271 • 9269
necessary in affording conscious awareness for visual stimuli
(Lamme and Roelfsema, 2000). Feedback involving visual areas
romodulators such as GABA. Indeed, the mutual inhibition of
GABAergic interneurons is implicated as a fundamental mecha-
nism in the neural synchrony that binds information for aware-
ness (Bush and Sejnowski, 1996; Lee et al., 2003). Numerical
modeling demonstrates that neural synchrony requires an opti-
mal ratio of 4:1 inhibitory to excitatory connections (Bush and
Sejnowski, 1996), consistent with the predominance of negative
connections in cortical pathways to the amygdala.
Complementary animal evidence demonstrates the presence
of inhibitory GABAergic circuitry in the amygdala (Szinyei et al.,
2000). Across the sustained timescale of conscious attention to
fear, inhibition may serve to downregulate parallel sources of
visual input to the amygdala, allowing a shift toward stimulus
elaboration in the amygdala and higher-order association areas.
This proposal accords with models of visual consciousness, in
which selective stimulus processing is facilitated by the broad
inhibition of surrounding sensory networks (Dehaene et al.,
2003). Terms such as “searchlight” and “surround inhibition”
have been used to describe this phenomenon (Crick, 1984). Bio-
physical and large-scale network models also highlight the im-
portance of inhibitory modulation for se lective attention to sa-
lient information (Rennie et al., 2002; Dehaene et al., 2003).
Masking of fear faces may disrupt reentrant feedback from
higher areas, such that nonconscious fear perception proceeds
with rapid feedforward connections between the amygdala and
components of the subcortical pathway. Such connections may
be reflected in the positive relationships within the brainstem–
of rapid-acting neuromodulators, such as glutamate, may facili-
tate feedforward connections along direct amygdala pathways,
allowing automatic responses to low-level fear stimulation to
proceed despite an absence of awareness. Whereas feedforward
connections may also be sufficient to activate visual and medial
prefrontal association areas, the blocking of reentrant feedback
these areas from reaching conscious awareness (Lamme and Ro-
Despite the preponderance of negative connectivity for con-
scious fear and positive connectivity for nonconscious fear, co-
variation between the amygdala and association areas was more
complex under both conditions. The lateralized profile of posi-
tive and negative connectivity between visual association areas
and the amygdala during conscious fear processing accords with
amygdala activity during the maintenance of conscious visual
attention may involve a dynamical shift in the balance of excita-
the striate cortex in this condition and thus a lack of visual elab-
oration in the higher-order areas to which it projects.
For medial prefrontal areas, awareness was distinguished by
showed a positive connection with the dorsal ACC for conscious
fear but with the rostral portion for nonconscious fear. This
three-way dissociation provides compelling support within the
one study for previous evidence using consciously seen negative
facial expressions (Kim et al., 2003, 2004) and unseen fearful eye
whites (Whalen et al., 2004), that the direction of amygdala rela-
tionship varies according to dorsal versus rostral ventral medial
prefrontal regions and level of awareness. The findings also con-
tradict the view that the ACC has a specific role in the conscious
modulation of emotional states (Damasio, 1996). Rather, a feed-
forward sweep of positive connections may be the means by
which the ACC is recruited as part of the networks for automatic
awareness, consistent with the negative connections with the
amygdala under both awareness conditions (Damasio, 1996).
association areas when conscious awareness was prevented, add-
ing weight to the view that the degree of neural connectivity
conscious awareness (Vuilleumier et al., 2002).
Our findings support a model of fear processing that goes
beyond a focus on the neuroanatomical specialization of amyg-
dala pathways and highlights the importance of the mode of
9270 • J.Neurosci.,September6,2006 • 26(36):9264–9271Williamsetal.•AmygdalaConnectivityandAwarenessforFear
functional connectivity in determining the level of awareness Download full-text
nature of fear (Hatfield et al., 1992), in which a cascade of auto-
matic physiological responses to fear to spread rapidly from one
individual to the next, without time for inhibitory feedback. On a
macro-scale, a similar cycle of covert contagion may underlie the
Attwell D, Iadecola C (2002) The neural basis of functional brain imaging
signals. Trends Neurosci 25:621–625.
Baddeley A, Emslie H, Nimmo-Smith I (1993) The Spot-the-Word test: a
robust estimate of verbal intelligence based on lexical decision. Br J Clin
BushP,SejnowskiT (1996) Inhibitionsynchronizessparselyconnectedcor-
tical neurons within and between columns in realistic network models.
J Comput Neurosci 3:91–110.
CataniM,JonesDK,DonatoR,FfytcheDH (2003) Occipito-temporalcon-
nections in the human brain. Brain 126:2093–2107.
Crick F (1984) Function of the thalamic reticular complex: the searchlight
hypothesis. Proc Natl Acad Sci USA 81:4586–4590.
Damasio AR (1996) The somatic marker hypothesis and the possible func-
tions of the prefrontal cortex. Philos Trans R Soc Lond B Biol Sci
Das P, Kemp AH, Liddell BJ, Brown KJ, Olivieri G, Peduto A, Gordon E,
WilliamsLM (2005) Pathwaysforfearperception:modulationofamyg-
dala activity by thalamo-cortical systems. NeuroImage 26:141–148.
de Gelder B, Vroomen J, Pourtois G, Weiskrantz L (1999) Non-conscious
recognition of affect in the absence of striate cortex. NeuroReport
DehaeneS,SergentC,ChangeuxJP (2003) Aneuronalnetworkmodellink-
ing subjective reports and objective physiological data during conscious
perception. Proc Natl Acad Sci USA 100:8520–8525.
EkmanP (1982) Emotioninthehumanface.Cambridge,UK:CambridgeUP.
Friston KJ, Buechel C, Fink GR, Morris J, Rolls E, Dolan RJ (1997) Psycho-
physiological and modulatory interactions in neuroimaging. NeuroIm-
Gordon E (2003) Integrative neuroscience. Neuropsychopharmacology
Gordon E, Cooper N, Rennie C, Hermens D, Williams LM (2005) Integra-
Gur RC, Sara R, Hagendoorn M, Marom O, Hughett P, Macy L, Turner T,
Bajcsy R, Posner A, Gur RE (2002) A method for obtaining
3-dimensional facial expressions and its standardization for use in neu-
rocognitive studies. J Neurosci Methods 115:137–143.
Hariri AR, Tessitore A, Mattay VS, Fera F, Weinberger DR (2002) The
amygdala response to emotional stimuli: a comparison of faces and
scenes. NeuroImage 17:317–323.
Hariri AR, Mattay VS, Tessitore A, Fera F, Weinberger DR (2003) Neocor-
tical modulation of the amygdala response to fearful stimuli. Biol Psychi-
HatfieldE,CacioppoJT,RapsonRL (1992) Primitiveemotionalcontagion.
Emot Social Behav 14:151–177.
Hoven CW, Duarte CS, Lucas CP, Wu P, Mandell DJ, Goodwin RD, Cohen M,
P, Susser E (2005) Psychopathology among New York city public school
PJ (2003) Inverse amygdala and medial prefrontal cortex responses to
surprised faces. NeuroReport 14:2317–2322.
PJ (2004) Contextualmodulationofamygdalaresponsivitytosurprised
faces. J Cogn Neurosci 16:1730–1745.
Lamme VA, Roelfsema PR (2000) The distinct modes of vision offered by
feedforward and recurrent processing. Trends Neurosci 23:571–579.
Lee KH, Williams LM, Breakspear M, Gordon E (2003) Synchronous
model of schizophrenia. Brain Res Brain Res Rev 41:57–78.
Liddell BJ, Brown KJ, Kemp AH, Barton MJ, Das P, Peduto A, Gordon E,
WilliamsLM (2005) Adirectbrainstem-amygdala-cortical“alarm”sys-
tem for subliminal signals of fear. NeuroImage 24:235–243.
Lim CL, Rennie C, Barry RJ, Bahramali H, Lazzaro I, Manor B, Gordon E
(1997) Decomposing skin conductance into tonic and phasic compo-
nents. Int J Psychophysiol 25:97–109.
Linke R, De Lima AD, Schwegler H, Pape HC (1999) Direct synaptic con-
nections of axons from superior colliculus with identified thalamo-
tical visual pathway to the amygdala. J Comp Neurol 403:158–170.
LogothetisNK (2003) TheunderpinningsoftheBOLDfunctionalmagnetic
resonance imaging signal. J Neurosci 23:3963–3971.
Morris JS, Ohman A, Dolan RJ (1998) Conscious and unconscious emo-
tional learning in the human amygdala. Nature 393:467–470.
MorrisJS,OhmanA,DolanRJ (1999) Asubcorticalpathwaytotherightamyg-
Morris JS, de Gelder B, Weiskrantz L, Dolan RJ (2001) Differential extra-
geniculostriate and amygdala responses to presentation of emotional
faces in a cortically blind field. Brain 124:1241–1252.
Ohman A, Soares JJ (1998) Emotional conditioning to masked stimuli: ex-
pectancies for aversive outcomes following nonrecognized fear-relevant
stimuli. J Exp Psychol Gen 127:69–82.
Rennie CJ, Robinson PA, Wright JJ (2002) Unified neurophysical model of
EEG spectra and evoked potentials. Biol Cybern 86:457–471.
SzinyeiC,HeinbockelT,MontagneJ,PapeHC (2000) Putativecorticaland
interneurons of the lateral amygdala. J Neurosci 20:8909–8915.
Tononi G, Edelman GM (1998) Consciousness and complexity. Science
croix N, Mazoyer B, Joliot M (2002) Automated anatomical labeling of
activations in SPM using a macroscopic anatomical parcellation of the
MNI MRI single-subject brain. NeuroImage 15:273–289.
Vuilleumier P, Armony JL, Clarke K, Husain M, Driver J, Dolan RJ (2002)
Neural response to emotional faces with and without awareness: event-
related fMRI in a parietal patient with visual extinction and spatial ne-
glect. Neuropsychologia 40:2156–2166.
Whalen PJ, Rauch SL, Etcoff NL, McInerney SC, Lee MB, Jenike MA (1998)
activity without explicit knowledge. J Neurosci 18:411–418.
Whalen PJ, Kagan J, Cook RG, Davis FC, Kim H, Polis S, McLaren DG,
Somerville LH, McLean AA, Maxwell JS, Johnstone T (2004) Human
amygdala responsivity to masked fearful eye whites. Science 306:2061.
White NS, Alkire MT (2003) Impaired thalamocortical connectivity in hu-
mans during general-anesthetic-induced unconsciousness. NeuroImage
WilliamsLM (2006) Anintegrativeneurosciencemodelofsignificancepro-
cessing. J Integr Neurosci 5:1–47.
BahramaliH,OlivieriG,DavidAS,PedutoA,GordonE (2001) Arousal
dissociates amygdala and hippocampal fear responses: evidence from si-
multaneous fMRI and skin conductance recording. NeuroImage
Williams LM, Liddell BJ, Rathjen J, Brown KJ, Gray J, Phillips M, Young A,
Gordon E (2004) Mapping the time course of nonconscious and con-
scious perception of fear: an integration of central and peripheral mea-
sures. Hum Brain Mapp 21:64–74.
don E, Bryant RA (2006a) Trauma modulates amygdala and medial pre-
Williams LM, Brown KJ, Palmer P, Liddell BJ, Kemp AH, Olivieri G, Peduto
AS, Gordon E (2006b) The mellow years?: neural basis of improving
emotional stability over age. J Neurosci 26:6422–6430.
Williams LM, Liddell BJ, Kemp AH, Bryant RA, Meares RA, Peduto AS,
Gordon E (2006c) Amygdala-prefrontal dissociation of subliminal and
supraliminal fear. Hum Brain Mapp 27:652–661.
Zajonc R (1980) Feeling and thinking: preferences need no inferences. Am
ZaldDH (2003) Thehumanamygdalaandtheemotionalevaluationofsen-
sory stimuli. Brain Res Brain Res Rev 41:88–123.
Williamsetal.•AmygdalaConnectivityandAwarenessforFearJ.Neurosci.,September6,2006 • 26(36):9264–9271 • 9271