Biological Psychology 94 (2013) 441–449
Contents lists available at ScienceDirect
Biological Psychology
journal homepage: www.elsevier.com/locate/biopsycho
Electrocortical processing of social signals of threat in combat-related post-traumatic stress disorder
a,∗ a a,c
Annmarie MacNamara , David Post , Amy E. Kennedy ,
b a,b,c
Christine A. Rabinak , K. Luan Phan
a
Department of Psychiatry, University of Illinois at Chicago, United States
b
Department of Psychiatry, University of Michigan, United States
c
Mental Health Service Line, Jesse Brown VA Medical Center, United States
a r t i c l e i n f o a b s t r a c t
Article history: Post-traumatic stress disorder (PTSD) is characterized by avoidance, emotional numbing, increased
Received 21 March 2013
arousal and hypervigilance for threat following a trauma. Thirty-three veterans (19 with PTSD, 14 without
Accepted 30 August 2013
PTSD) who had experienced combat trauma while on deployment in Iraq and/or Afghanistan completed
Available online 8 September 2013
an emotional faces matching task while electroencephalography was recorded. Vertex positive poten-
tials (VPPs) elicited by happy, angry and fearful faces were smaller in veterans with versus without
Keywords:
PTSD. In addition, veterans with PTSD exhibited smaller late positive potentials (LPPs) to angry faces
PTSD
and greater intrusive symptoms predicted smaller LPPs to fearful faces in the PTSD group. Veterans with
Numbing
PTSD were also less accurate at identifying angry faces, and accuracy decreased in the PTSD group as
Vertex positive potential
VPP hyperarousal symptoms increased. These findings show reduced early processing of emotional faces,
Late positive potential irrespective of valence, and blunted prolonged processing of social signals of threat in conjunction with
LPP impaired perception for angry faces in PTSD.
Event-related potentials © 2013 Elsevier B.V. All rights reserved. ERP
Emotional faces
1. Introduction work have led to a substantial platform of knowledge about emo-
tional brain responses in PTSD (Nemeroff et al., 2006; Pitman et al.,
Post-traumatic stress disorder (PTSD) is a debilitating men- 2012), several inconsistent results have also emerged, thus raising
tal disorder that results following exposure to a traumatic event questions about key biological dimensions of the disorder (Lanius,
such as armed combat, motor vehicle accident or sexual assault Bluhm, Lanius, & Pain, 2006; Lanius, Brand, Vermetten, Frewen, &
(American Psychiatric Association, 2000). Patients with PTSD expe- Spiegel, 2012).
rience intense emotional reactions when reminded of their trauma Many affective neuroimaging studies in PTSD have focused on
and report exaggerated arousal (e.g., being easily startled), both the early, automatic processing of threatening stimuli as reflected
of which suggest emotional hyper-responsivity. However, individ- in amygdala activity. To this end, researchers have used masked
uals with PTSD also report anhedonia and a decreased capacity (Rauch et al., 2000) or briefly presented (e.g., 200 ms; Shin et al.,
for emotion (‘emotional numbing’), suggesting emotional hypo- 2005) pictures or words. Pictorial stimuli are often faces, with
responsivity. Characterizing the neural basis of these diverse fearful faces representing indirect or ambiguous threat and angry
emotional responses poses a major challenge to contemporary faces representing direct or non-ambiguous threat (Kim et al.,
psychiatric research (Friedman, Resick, Bryant, & Brewin, 2011). 2011). Other work has used images from the International Affective
Functional neuroimaging techniques have focused primarily on Picture System (IAPS; Lang, Bradley, & Cuthbert, 2005), or trauma-
the study of brain function related to fear perception in exam- specific stimuli (e.g., traumatic scripts, combat images; Rauch et al.,
ining emotional responsivity. While consistent findings from this 1996; Shin et al., 1997). Evidence of increased amygdala response
to threatening stimuli in PTSD has been observed in some (Rauch
et al., 1996, 2000; Shin et al., 1997, 2005) but not all of these stud-
ies (Bremner et al., 1999; Britton, Phan, Taylor, Fig, & Liberzon,
∗
Corresponding author at: Department of Psychiatry, University of Illinois at
2005; Lanius et al., 2002; Phan, Britton, Taylor, Fig, & Liberzon,
Chicago, 1747 West Roosevelt Road, Chicago, IL 60607, United States.
2006; Sakamoto et al., 2005; Shin et al., 1999). Indeed, recent meta-
Tel.: +1 312 413 4707.
analyses (Etkin & Wager, 2007; Hayes, Hayes, & Mikedis, 2012;
E-mail addresses: [email protected],
[email protected] (A. MacNamara). Lanius et al., 2006; Simmons & Matthews, 2012) have questioned
0301-0511/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.biopsycho.2013.08.009
442 A. MacNamara et al. / Biological Psychology 94 (2013) 441–449
the consistency of amygdala hyper-reactivity in PTSD. Importantly, Wager, 2007; Hayes et al., 2012; Lanius et al., 2006; Simmons &
the tendency for increased amygdala activity in PTSD to have been Matthews, 2012).
more reliably observed in response to masked or briefly-presented As an early measure of face-processing, the vertex positive
stimuli (Bryant et al., 2008; Felmingham et al., 2010; Rauch et al., potential (VPP) is a positive-going ERP component that peaks
2000; Shin et al., 2005) suggests that it may reflect an automatic at central sites within 200 ms following stimulus onset and is
arousal response (Lanius et al., 2006). larger for faces compared to other types of stimuli (e.g., Bentin
Differences in stimuli may partially explain discrepancies et al., 2007; Rossion & Jacques, 2008; Wheatley, Weinberg, Looser,
observed in prior work. For instance, increased reactivity to Moran, & Hajcak, 2011). The VPP is believed to reflect the perceptual
emotional stimuli in PTSD may be primarily observed for trauma- and structural encoding of faces (Carmel & Bentin, 2002; Jeffreys,
specific stimuli (e.g., personalized scripted imagery of trauma), 1989; Wheatley et al., 2011); moreover, the VPP responds to the
whereas decreased or normative reactivity may be more fre- emotional salience of faces, and may be especially enhanced to
quently observed for generic threat depictions (e.g., threatening threatening faces (Batty & Taylor, 2003; Blau, Maurer, Tottenham,
faces; Casada, Amdur, Larsen, & Liberzon, 1998; Pineles, Shipherd, & McCandliss, 2007; Righart & de Gelder, 2008).
Mostoufi, Abramovitz, & Yovel, 2009). Combat-related PTSD, in The late positive potential (LPP) is distinguished by its abil-
particular, has been associated with reduced response when ity to measure the sustained, elaborative processing of stimulus
processing non-traumatic threatening/aversive stimuli (Armony, salience (Cuthbert, Schupp, Bradley, Birbaumer, & Lang, 2000). The
Corbo, Clément, & Brunet, 2005; Phan et al., 2006). Additionally, LPP is a centro-parietal, positive-going ERP component that begins
the type of ‘control’ stimuli that are used may impact results: approximately 300 ms after stimulus onset and is larger for emo-
for example, among studies that use facial stimuli to evaluate tional compared to neutral stimuli (Foti, Hajcak, & Dien, 2009;
threat-processing in PTSD, neutral faces may not provide the MacNamara, Foti, & Hajcak, 2009; Schupp et al., 2000). In addition
most appropriate comparison, because anxious individuals may to tracking the affective nature of stimuli, the LPP is also sensi-
interpret even these baseline stimuli differently than non-anxious tive to individual differences in the perceived salience of stimuli.
individuals (Cooney, Atlas, Joormann, Eugène, & Gotlib, 2006; For instance, the LPP is larger for photographs of relatives, or one’s
Somerville, Kim, Johnstone, Alexander, & Whalen, 2004). Differ- own name or face (Grasso & Simons, 2011; Tacikowski & Nowicka,
ences in the types of control groups employed (i.e., trauma-exposed 2010) and the LPP elicited by pictures of food is larger in individuals
versus non-exposed controls) may also add to the heterogeneity of who have been food-deprived (Stockburger, Schmälzle, Flaisch,
findings in PTSD. Bublatzky, & Schupp, 2009). The LPP is also sensitive to more willful
In addition to hemodynamic measures, event-related potentials modulations of stimulus salience. For example, the LPP is smaller
(ERPs) can be used to measure affective processing in PTSD. ERPs when participants are asked to reduce the emotional salience of
have excellent temporal resolution, and can therefore be used to affective pictures (Hajcak & Nieuwenhuis, 2006; Moser, Hajcak,
examine both early and late (i.e. prolonged) threat processing in Bukay, & Simons, 2006; Parvaz, MacNamara, Goldstein, & Hajcak,
the same experimental paradigm. In addition, because ERPs do not 2012). Therefore, the LPP seems to be sensitive to both bottom-up
rely on the ‘subtraction’ analysis used in typical functional mag- (i.e., content-driven) and top-down (i.e., strategic) modulations of
netic resonance imaging (fMRI) contrast images; group differences stimulus salience, and may be an excellent tool for investigating
in neural response to threatening stimuli can be more easily dis- the elaborated processing of threatening stimuli in PTSD.
tinguished from differences in response to control stimuli. Most The current study used the VPP and the LPP to examine the
commonly, ERPs have been used to examine abnormalities in the processing of emotional stimuli in a group of returning veterans
processing of non-affective target stimuli and affective distracter with and without PTSD. Given high levels of interpersonal dys-
stimuli in PTSD (Karl, Malta, & Maercker, 2006). Few studies, how- function in combat-related PTSD (Frueh, Turner, Beidel, & Cahill,
ever, have utilized ERPs to examine the processing of task-relevant, 2001; Jordan et al., 1992; Ruscio, Weathers, King, & King, 2002),
emotional stimuli in PTSD. we used facial stimuli presented in the context of an affect match-
In one of these studies, Felmingham, Bryant, and Gordon (2003) ing task (Hariri, Tessitore, Mattay, Fera, & Weinberger, 2002; Phan
showed an angry and a neutral face to individuals with PTSD et al., 2008). We expected to observe evidence of altered threat-
and a group of controls while electroencephalography (EEG) was processing in PTSD for the LPP (Tso, Chiu, King-Casas, & Deldin,
recorded. Pictures were shown repeatedly for 110 ms in a pas- 2011), however, based on conflicting prior results (Armony et al.,
sive viewing task, and the authors used negative-going ERPs to 2005; Etkin & Wager, 2007; Grasso & Simons, 2012; Hayes et al.,
measure the processing of facial stimuli. Results revealed that 2012; Lanius et al., 2006; Simmons & Matthews, 2012), we did not
individuals in the control group evinced larger temporo-occipital have directional hypotheses for the ERPs. Based on prior work, we
negativities around 100 ms and 650 ms after picture onset, for thought it possible that we would observe either increased ampli-
angry compared to neutral faces. Individuals in the PTSD group did tudes to threatening stimuli, reflecting hypervigilance (Karl et al.,
not show this difference for angry versus neutral faces, indicating 2006; Shin et al., 2005) or decreased amplitudes, possibly reflect-
reduced discrimination between neutral and threatening stimuli. ing avoidance or numbing symptoms (Felmingham et al., 2003;
Unfortunately, Felmingham et al. (2003) did not include other face Felmingham, Bryant, Kendall, & Gordon, 2002). Behaviorally, we
types, so it could not be determined whether PTSD-related deficits expected to observe reduced accuracy for individuals with PTSD
in emotion-processing were specific to threatening faces. More- – especially for trials containing threatening expressions (Pollak,
over, pictures were presented for only 110 ms and the sustained Cicchetti, Hornung, & Reed, 2000; Shenk, Putnam, & Noll, 2012).
processing of threatening pictures, which might have indicated dif-
ferences between conditions for the PTSD group, was not examined.
2. Method
Briefly displayed images may reveal ‘bottom-up’ hyperreactiv-
ity often reported by patients with PTSD (Etkin & Wager, 2007),
2.1. Participants
whereas longer stimulus presentation durations may be neces-
sary to investigate ‘top-down’ processes (e.g., emotion regulation Participant demographics, symptomatology and comorbid diagnoses are pre-
sented in Table 1. Participants were 33 combat-exposed male veterans who had
deficits, avoidance) that may contribute to the pathophysiology of
returned from deployment in Operation Enduring Freedom or Operation Iraqi Free-
PTSD (Felmingham et al., 2008; Frewen & Lanius, 2006). Moreover,
dom (OEF/OIF). Participants were recruited from the Veterans Affairs (VA) Ann
an examination of both early and late threat-processing in PTSD
Arbor Healthcare System, as well as from the surrounding community via paper
could help resolve prior discrepancies in the literature (Etkin & advertisements and posted flyers.
A. MacNamara et al. / Biological Psychology 94 (2013) 441–449 443
Table 1
Participant demographics. Top section: means (and standard deviations) for age, years of education and symptomatology. Bottom section: ethnicity and comorbid diagnosis
by number and percentage of participants.
CEC PTSD Group comparison
n = 14 n = 19
Age 34.71 (8.80) 29.95 (8.03) ns
Years of education 15.64 (1.50) 12.89 (1.33) p < 0.001
CAPS total 3.43 (4.42) 72.79 (14.34) p < 0.001
Re-experiencing 0.43 (1.09) 18.11 (6.79) p < 0.001
Avoidance 1.36 (1.98) 27.37 (7.16) p < 0.001
Hyperarousal 1.64 (2.56) 27.32 (4.35) p < 0.001
PCL-M 24.93 (11.69) 56.21 (10.52) p < 0.001
CES 21.64 (5.37) 24.21 (7.30) ns
Ham-D 1.36 (1.95) 12.74 (3.23) p < 0.001
Ethnicity n % n %
Caucasian 12 85.70 19 100.0 ns
Other 2 14.30 – –
Comorbid diagnoses
MDD – – 4 21.0 ns
Alcohol abuse – – 3 15.8 ns
Group comparisons were performed using independent t-tests, except for ethnicity and comorbidity comparisons, which were performed using Fisher’s Exact Test to account
for low cell counts. ns = non-significant.
The Structured Clinical Interview for the Diagnostic and Statistical Manual of screen. Face-matching trials could be fearful, angry or happy. On shape-matching
Mental Disorders IV (SCID-I/NP for DSM-IV, First, Spitzer, Gibbon, & Williams, 1995) trials, participants were instructed to choose the shape at the bottom of the screen
was used to assess for past and present psychological disorders; PTSD sympto- that matched (i.e., had the same form as) the target shape at the top of the screen.
matology and severity was also measured using the Clinician Administered PTSD In line with previous EFT studies (e.g., Labuschagne et al., 2010; Phan et al., 2008)
Scale (CAPS; Blake et al., 1995) and the PTSD-Checklist, military version (PCL- we used geometric shapes as control stimuli instead of neutral faces, because neu-
M; Blanchard, Jones-Alexander, Buckley, & Forenia, 1996). Combat exposure was tral faces may be more influenced by individual differences (e.g., anxiety levels;
assessed using the Combat Exposure Scale (CES; Keane et al., 1989). Somerville et al., 2004).
Participants were eligible for the PTSD group (n = 19) if they met current crite- The task was divided into two blocks, with each block having 12 angry, 12
ria for combat-related PTSD and had scores of at least a 40 or greater on the CAPS, fearful, 12 happy and 12 shape-matching trials; trials were presented randomly
reflecting moderate to severe PTSD symptoms. Comorbid Axis I disorders were per- within each block, for a total of 96 trials across both blocks. The inter-trial interval
mitted in the PTSD group, with the exception of current or prior psychosis or mania varied between 1000 and 3000 ms, during which time a white fixation cross was
or hypomania (see Table 1). History of head trauma, loss of consciousness, and centrally presented on a black background. During presentation of the images, par-
traumatic brain injury (of any severity) were exclusionary criteria for all partici- ticipants were instructed to maintain focus on the screen, but were permitted to
pants. Participants were eligible to be in the combat-exposed control group (CEC; look freely at the images. Participants performed 6 practice trials prior to beginning
n = 14) if they did not meet criteria for PTSD or any other Axis I disorder (disorders the experiment, which used facial stimuli that were not used in the experiment.
>6 months prior were allowed and were asymptomatic for PTSD according to the
CAPS (scores <20). Compared to participants in the CEC group, those in the PTSD
2.4. EEG recording and data reduction
group had significantly higher scores on the PCL-M (t(31) = 8.05, p < 0.001) and CAPS
(t(31) = 19.85, p < 0.001); participants in the CEC and PTSD groups did not differ on
Continuous EEG was recorded using an elastic cap and the ActiveTwo BioSemi
level of combat exposure (t(31) = 1.16, p > 0.25; see Table 1). Participants in the PTSD
system (BioSemi, Amsterdam, Netherlands). Thirty-four electrode sites (standard
group also reported greater levels of depression than participants in the CEC group
32 channel setup, as well as FCz and Iz) were used, based on the 10/20 system; in
(t(31) = 11.69, p < 0.001; see Table 1), as measured by the Hamilton Depression Rating
addition, one electrode was placed on each of the left and right mastoids. The elec-
Scale (Ham-D; Williams, 1988).
trooculogram (EOG) generated from eyeblinks and eye movements was recorded
None of the participants were taking psychiatric medications (at least 4 weeks
from four facial electrodes: vertical eye movements and blinks were measured with
prior to the EEG session) and none had significant medical or neurological disorders.
two electrodes placed approximately 1 cm above and below the right eye; horizon-
All participants were paid for their time, and gave written informed consent, as
tal eye movements were measured using two electrodes placed approximately 1 cm
approved by the VA Ann Arbor Healthcare System Institutional Review Board.
beyond the outer edge of each eye. The EEG signal was pre-amplified at the electrode
to improve the signal-to-noise ratio. The data were digitized at 24-bit resolution
2.2. Materials with a Least Significant Bit (LSB) value of 31.25 nV and a sampling rate of 1024 Hz,
using a low-pass fifth order sinc filter with a −3 dB cutoff point at 208 Hz. The voltage
Twenty-four angry, 24 fearful and 24 happy faces were selected from the Gur from each active electrode was referenced online with respect to a common mode
emotional faces set (Gur et al., 2002); half of the faces depicted male actors and the sense active electrode producing a monopolar (non-differential) channel.
other half depicted female actors. In addition, 3 geometric shapes – a circle, a square
and a triangle – were used as control stimuli (see task description, below). The shapes
2.5. Data reduction and analysis
were presented in white on a black background; faces were presented in color on a
black background, on a 19 computer screen using Presentation software (Neurobe-
Off-line analyses were performed using Brain Vision Analyzer 2 software (Brain
havioral Systems, Inc., Albany, CA). Participants were seated approximately 60 cm
Products, Gilching, Germany). Data from correct trials were segmented for each
from the screen.
trial beginning 200 ms prior to picture onset and continuing for 3200 ms (3000 ms
beyond picture onset); baseline correction for each trial was performed using the
2.3. Task 200 ms prior to picture onset. Offline, data were re-referenced to the average of the
two mastoids and band-pass filtered with high-pass and low-pass filters of 0.01 and
Participants completed a version of the Emotional Face-Matching Task (Hariri 30 Hz, respectively. Eye blink and ocular corrections used the method developed by
et al., 2002), which has proven useful in characterizing threat-processing in anxious Gratton, Coles, and Donchin (1983). Artifact analysis was used to identify a voltage
and non-anxious participants (Labuschagne et al., 2010) and which would facilitate step of more than 50.0 V between sample points, a voltage difference of 300.0 V
comparison with prior ERP work in PTSD (Felmingham et al., 2003). On each trial, 3 within a trial, and a maximum voltage difference of less than 0.50 V within 100 ms
◦ ◦
images (each subtending a visual angle of approximately 11 × 13 ) were presented intervals. Trials were also inspected visually for any remaining artifacts, and data
for 3000 ms, in a ‘triangular’ arrangement – i.e., one image was centered in the top- from individual channels containing artifacts were rejected on a trial-to-trial basis.
half of the screen and the other two images were presented in the bottom-half of For figures, a digital low-pass (12 Hz) filter was applied offline before plotting the
the screen (one to the left and one to the right). There were “face-matching” and waveforms; statistical analyses were conducted using the original filter settings.
“shape-matching” trials. The VPP was scored at fronto-central pooling, Fz, FC1, FC2, FCz and Cz (Foti, Olvet,
On each face-matching trial, the faces of three different actors were presented: Klein, & Hajcak, 2010). For each condition and participant, peak VPP amplitude was
two were always emotional and one always bore a neutral expression. Participants defined as the local positive peak evident in the 150–300 ms period after stimulus
were instructed to select one of the faces at the bottom of the screen that bore the onset evident in the grand-average waveforms (Fig. 2). Of note, the search window
same emotional expression as the ‘target’ face centered in the top portion of the used for detection of VPP peak amplitudes was wider than in some previous studies,
444 A. MacNamara et al. / Biological Psychology 94 (2013) 441–449
Fig. 1. Bar graphs depicting reaction time (RT, ms; left) and accuracy (% correct; right) for participants in the CEC group (light bars) and the PTSD group (dark bars). Bars
represent standard error of the mean. Significant differences using Tukey’s (or Tukey–Kramer’s) tests are indicated with horizontal brackets (all ps < 0.05).
Table 2
Means (and standard deviations) for VPP and LPP amplitudes elicited by emotional faces and shapes.
CEC PTSD Group comparison
Condition
VPP (V) Angry faces 9.39 (6.72) 1.91 (3.82) p < 0.05
Fearful faces 9.09 (5.51) 3.70 (3.44) p < 0.05
Happy faces 8.61 (5.60) 4.62 (4.44) p < 0.05
Shapes 7.64 (5.32) 5.76 (3.88) ns
LPP (V) Angry faces 4.89 (3.38) 1.55 (3.90) p < 0.05
Fearful faces 2.98 (3.87) 0.55 (4.76) ns
Happy faces 1.05 (3.81) 3.80 (6.12) ns
Shapes −2.37 (6.74) −1.65 (3.53) ns
Group comparisons were performed using Tukey–Kramer’s tests, except for the LPP elicited by angry and fearful faces, for which a priori comparisons were performed using
independent t-tests. ns = non-significant.
in order to account for the later latency of VPPs elicited by shapes (Caldara et al., with faces elicited worse performance than trials with shapes
2003). After peak detection, mean amplitudes in a 20 ms window centered on peaks
and angry faces elicited the worst performance (i.e., shapes accu-
were extracted for statistical analysis. The LPP was scored by averaging amplitudes
racy > happy accuracy = fearful accuracy > angry accuracy; Tukey’s
at centro-parietal pooling, Cz, CP1, CP2 and Pz (MacNamara & Hajcak, 2010) from
1 p < 0.05). A near-significant between-groups effect (F(1,31) = 4.15,
600 to 3000 ms after picture onset (Dennis & Hajcak, 2009; Foti et al., 2009).
2