Abnormal Target Detection and Novelty Processing Neural Response in Posttraumatic Stress Disorder

Abstract

Attention impairments are common symptoms of posttraumatic stress disorder (PTSD); however, the nature of these impairments remains elusive. Attention impairment may arise as the result of either excessive response to task-irrelevant stimuli or reduced response to task-relevant information. To test the association between PTSD and response to task-relevant and task-irrelevant stimuli, we used a 3-tone novelty auditory oddball task (AOD). We hypothesized that participants with PTSD relative to trauma controls would have less response during novelty processing in the dorsolateral prefrontal cortex (dlPFC) and the anterior cingulate cortex, as well as less response in the dlPFC and the orbitofrontal cortex during target detection. Thirty-one male veterans completed a 3-tone novelty AOD task during functional magnetic resonance imaging. Compared to trauma controls, the PTSD group had reduced response during novelty processing in ventromedial prefrontal cortex, superior/middle frontal gyrus (dlPFC), supplementary motor area/caudate, and in posterior regions including bilateral posterior cingulate cortex. The current results suggest PTSD is associated with a pattern of reduced response to novel stimuli. A disturbed orienting response in these brain regions could theoretically underlie PTSD attention-related symptoms.

1. Introduction

Individuals with posttraumatic stress disorder (PTSD) report attention-related symptoms including hypervigilance, exaggerated startle, and difficulty concentrating (American Psychiatric Association, 2013, 2000). Consistent with these self-reports, neurocognitive studies find that PTSD is associated with impaired arousal and frontal systems in working memory, mental manipulation, sustained attention, and initial acquisition of information (Hayes et al., 2012; Vasterling et al., 1998). Some frameworks of PTSD suggest these impairments may result from neurologically-based differences in attentional function (Hayes et al., 2012; Morey et al., 2009), pointing to the relevance of uncovering such underlying mechanisms.

The study of non-emotional neurological attentional processes in PTSD has primarily utilized electroencephalography (EEG) and auditory oddball (AOD) target detection tasks, in which participants listen to a series of tones, responding only when infrequent target tones are presented. In two-tone variants of the AOD task, frequent “standard” tones and infrequent “target” tones are presented, whereas in three-tone novelty variants, infrequent non-repeating “novel” tones are also presented. In both cases, participants are instructed to respond only to target stimuli. People with PTSD show altered event-related potentials (ERPs) during AOD compared to controls, particularly in the latency and amplitude of the target detection Target P300 (“P3b”), a component involved in attention allocation, activation of working memory, and arousal state (Johnson et al., 2013; Karl et al., 2006; Polich and Kok, 1995).

However, PTSD attention problems may also involve altered reactivity to unexpected (e.g. novel) events resulting in difficulty maintaining focused attention. Participants with PTSD also show selective sensitivity measured by ERPs during novelty processing (Kimble et al., 2000). The Novelty P300 (“P3a”), which occurs approximately 50 milliseconds prior to the Target P3, is associated with disruption of attention monitoring (Polich and Criado, 2006). Although different cognitive functions in different brain regions are thought to be responsible for the Target P3 versus the Novelty P3, novelty detection remains relatively under-examined in PTSD compared to target detection.

While ERP studies have identified this potential difference associated with PTSD, ERPs do not offer the most precise spatial resolution to differentiate underlying contributions of specific brain regions to the psychological process. Functional magnetic resonance imaging (fMRI) and magnetoencephalography (MEG) studies allow for more precise regional variance associated with attention monitoring to be examined. fMRI research using the traditional two-tone AOD task has found greater blood oxygen level dependent (BOLD) response during target detection in dorsal (Bryant et al., 2005; Felmingham et al., 2009) and rostral anterior cingulate cortex (ACC), left amygdala (Bryant et al., 2005), supramarginal gyrus, and hippocampus (Felmingham et al., 2009) in participants with PTSD versus non-trauma controls. A related MEG study using a combat versus neutral word attentional blink task determined that PTSD was associated with altered activity during combat words suggesting deficits in rapid regulatory processes, particularly at earlier latencies (Todd et al., 2015). More recently, using MEG during a three-tone threat-related novelty AOD task, PTSD was associated with reduced activity in the dorsolateral prefrontal cortex (dlPFC) and orbito-frontal cortex (OFC) during target detection and reduced activity in the dlPFC and anterior cingulate cortex (ACC) during threat-related novelty processing compared to trauma control participants (Herz et al., 2016).

The results of the 2-tone fMRI and the 3-tone MEG studies differ substantially, and this may reflect differences between using a 2-tone versus a 3-tone task. Still, together these studies provide further evidence of PTSD-associated neural abnormalities in target and novelty processing. Yet, PTSD-associated BOLD responses to distracting, task-irrelevant stimuli, such as measured with a novelty AOD paradigm, remain undetermined. Additionally, no AOD fMRI studies have examined PTSD in comparison to a control group with trauma exposure, although ERP studies have found differences in AOD response associated with trauma exposure alone (i.e., trauma-exposed individuals without PTSD versus non-trauma controls), suggesting that mere exposure to trauma may sometimes affect the orienting response in a non-pathological manner (Kimble et al., 2010).

We designed this study to determine PTSD-associated differences in attentional processing of task-relevant and task-irrelevant information by comparing a group of participants with PTSD to a trauma-exposed, non-PTSD control group (both groups exposed to combat in Iraq or Afghanistan). We hypothesized that individuals with PTSD would show altered engagement of networks implicated in attentional processing during target detection and novelty processing. Based on the novelty detection (3-tone) MEG PTSD study (Herz et al., 2016), which was the most similar in task design to our study, and the novelty detection fMRI study in healthy participants (Kiehl et al., 2001) on which our task was based, we expected participants with PTSD to show reduced activation, relative to trauma controls, during target detection in the dorsolateral prefrontal cortex dlPFC and OFC. Similarly, we also hypothesized that participants with PTSD relative to trauma controls would demonstrate less activity in dlPFC and ACC to novel (task-irrelevant) stimuli, which should trigger an increased orienting response (related to exaggerated startle) in PTSD.

2. Methods

2.1. Participants

Recruitment was restricted to Afghanistan and Iraq combat veterans to reduce variability introduced from differing trauma types (e.g. sexual assault versus combat) or disease chronicity, as chronic PTSD may involve progressive neural deterioration (Golier et al., 2006). Veterans were recruited by flyers, radio, presentations at veterans’ organizations, and word of mouth. Criteria for exclusion were: use of neuroleptics or anti-convulsants, traumatic brain injury, left handedness, neurological disorders, severe medical illness, severe suicidal or homicidal ideation, inability to consent, and self-reported prior DSM–IV Axis I disorder other than PTSD or substance use disorder. Additionally, we excluded those unable to comply with standard MRI safety protocols.

The institutional review board of the University of New Mexico approved this study and all procedures were performed in accordance with relevant guidelines and regulations. All participants gave written informed consent. All participants were equally compensated.

Out of 76 individuals who consented to a phone screen, 47 were found eligible and 41 attended their first appointment. Four participants completed assessments but not MRIs. Three participants were removed for much higher than average frequency of errors on the task (error rate > 38%) suggesting drowsiness. Females were targeted as well as males, but only males reached an analyzable sample size, thus two females in the non-PTSD group were excluded from analyses. One participant with PTSD was removed because of mean relative framewise displacement (Power et al., 2012) greater than 0.5 mm; leaving a final group of 31 male Iraq/Afghanistan combat veterans aged 22-64 years, 15 with PTSD and 16 without PTSD.

2.1. Measures

All tests were administered by a licensed clinical psychologist or trained research personnel closely supervised by a licensed clinical psychologist. PTSD diagnosis was determined using the Clinician Administered PTSD Scale (CAPS) (Weathers, 2004). The 1/2 CAPS scoring rule was used to determine DSM-IV PTSD symptoms (Weathers, 2004). PTSD severity scores were the sums of intensity and frequency scores for all symptoms (Weathers, 2004). IQ estimates were calculated using the two-subscale protocol of the Wechsler Abbreviated Scale of Intelligence (WASI) (Weschler, 1999). Alcohol consumption (past 90 days) was assessed using the Timeline Followback (Sobell and Sobell, 1996). Depression severity was measured with the Beck Depression Inventory - Second Edition (BDI-II) (Beck et al., 1996). Traumatic brain injury (TBI) was assessed with an unpublished detailed questionnaire developed for this purpose by the Mind Research Network. People with a loss of consciousness greater than 5 minutes, any experience of secondary symptoms, or a diagnosis of traumatic brain injury were excluded from the study.

2.2. Novelty Auditory Oddball Task

The AOD task used was previously described in Laurens et al. (2005). Participants were presented with 2 runs of 244 auditory stimuli each. The stimuli included repeating target tones (1500 Hz; probability of occurrence 0.10), novel tones (non-repeating digital noises; probability 0.10), and repeating standard tones (1000 Hz; probability 0.80). Three to five standard tones preceded each target or novel stimulus. Reaction time (RT) was computed for correct responses, defined as responses to targets tones made within 100-2100 milliseconds (ms) post-stimulus. Responses to novel and standard stimuli made within this time period were considered errors of commission. Failures to respond to target stimuli during this time period were considered errors of omission. Each run was 6 minutes, 18 seconds long and began with a 7500 ms fixation cross and an initial delay of 500 ms. Each tone was presented for 250 ms with a jittered interstimulus interval of 740-1740 ms. Each run concluded with a 10 second fixation cross. In order to ensure participants understood the instructions, each participant practiced for 10 trials prior to the actual fMRI task.

2.3. MRI acquisition

All MRI data were acquired on a 3.0 T Siemens Triotim scanner with a twelve-channel head coil, and EPI scans were acquired parallel to the ventral surface of the participants’ orbitofrontal cortex to reduce signal dropout and distortion in this region (Deichmann et al., 2003). Each volume acquired consisted of 33 ascending axial slices (64×64 matrix, 3.8×3.8 mm×3.5mm thickness, 30% gap, TR/TE=2s/29ms, FOV=240mm, flip angle=75°). In addition, a high-resolution T1-weighted MP-RAGE anatomical image was acquired (repetition time [TR]=2,530 ms, echo times [TE]=1.64, 3.50, 5.36, 7.22, 9.08 ms, flip angle=7°, 192 sagittal slices, 256×256 matrix, slice thickness=1mm, no gap). Participants were instructed to abstain from alcohol for 24 and nicotine for 2 hours prior to imaging sessions, which lasted 1.5 hours. Data from other tasks will be presented in the future or have already been reported (Sanjuan et al., 2013). Two mildly emotional tasks (viewing emotional faces and non-violent images from the Middle East) preceded the AOD task.

2.4. Analysis

The first 3 volumes of each functional run were discarded to allow the magnet to reach steady state. MCFLIRT (Motion Correction using FMRIB’s Linear Image Registration Tool) (Jenkinson et al., 2002) was used to motion correct images within a run; each volume was aligned to the first volume within the run. Images were deskulled using BET (Brain Extraction Tool) (Smith, 2002), spatially smoothed with a 5 mm full-width half-max Gaussian kernel, temporally filtered using a high-pass filter of 50 sec, and grand mean intensity normalized; all of these steps were performed using FMRIB’s Expert Analysis Tool (FEAT) (Smith et al., 2004).

Customized regressors were created for the three trial types of interest and a fourth regressor was derived that represented errors; each regressor was convolved with the canonical double gamma hemodynamic response function in FSL along with the temporal derivative. Time series analyses were conducted using FMRIB’s Improved Linear Model (FILM) (Woolrich et al., 2001) with local autocorrelation estimation. First-order movement parameters were included in the multiple regression. This first level analysis generated parameter estimates for each condition of interest, which, in turn, were combined to generate primary contrasts of interest: Contrasts of interest included: 1) Targets versus Standards - target stimuli relative to the standard baseline; 2) Novels versus Standards - novel stimuli relative to the standard stimulus baseline; and 3) Novels versus Targets. Contrast maps were registered to the participants’ high-resolution anatomical image and the MNI 152 brain template using FMRIB’s Linear Image Registration Tool (FLIRT) (Jenkinson et al., 2002; Jenkinson and Smith, 2001).

Group analyses were conducted using FMRIB’s Local Analysis of Mixed Effects (FLAME) (Woolrich et al., 2004) Stage 1. Independent samples t-tests were performed to examine differences in the contrasts of interest (Target > Standard; Novel > Standard) between participants with PTSD and controls. All analyses were corrected for multiple comparisons using minimum cluster sizes determined in Gaussian Random Field theory (Worsley et al., 1996), thus statistical maps were thresholded using cluster-based thresholding in FSL, with a voxelwise threshold of z>2.3 and cluster threshold of p<.05.

3. Results

3.1 Sample characteristics

3.1.1. AOD performance

As expected, there were no differences between groups in accuracy for target, novel, or standard stimuli, or in RT mean or standard deviation for targets (Table 1).

Table 1.

Behavioral Performance Variables by Group

	PTSD (N=15)		No PTSD (N=16)		t	p
	Mean (SD)	Range	Mean (SD)	Range
Correct Target	47 (0.8)	46–48	47 (1.1)	44–48	0.07	0.943
Correct Novel	46 (3.1)	36–48	46 (4.0)	33–48	−0.10	0.922
Correct Standard	387 (0.6)	385–387	386 (0.9)	384–387	1.29	0.208
Reaction Time Mean (Hits)	464 (92.7)	302–644	427 (75.1)	355–626	1.25	0.223
Reaction Time SD (Hits)	116 (30.4)	68–166	105 (30.4)	72–168	1.03	0.313

^aSignificant at p<.05

The sums of possible correct responses across both runs were 48 each for Target and Novel and 392 for Standard tones.

PTSD, posttraumatic stress disorder; N, sample size; SD, standard deviation;

3.1.2. Demographics and pathology

Comparison groups did not differ in age, percent alcohol drinking days, days since last alcoholic drink, or depression severity (Table 2). As expected, the PTSD group had a higher mean CAPS PTSD severity score than the control group. Additionally, the control group had a higher average IQ than the PTSD group (t(29) = 2.597, p < .05), as has been found elsewhere (Macklin et al., 1998). The PTSD group also had higher average drinks per drinking day (t(29) = −2.109, p < .05), than the control group.

Table 2.

Demographic and Clinical Variables by Group

	PTSD (N=15)		No PTSD (N=16)		t	p
	Mean (SD)	Range	Mean (SD)	Range
Age	31 (11.6)	22–64	29 (6.0)	22–41	0.47	0.285
CAPS Severity Scorea	60 (15.7)	43–98	22 (13.7)	0–43	7.32	0.000
Drinks per drinking daya	5.2 (4.43)	0–17	2.5 (2.44)	0–10	2.11	0.044
Percent drinking days	24 (25.7)	0–99	16 (22.4)	0–94	0.91	0.372
Days since last drinkb	11 (21.3)	1–83	5.9 (5.57)	1–20	0.86	0.397
Beck Depression Inventory	18 (13.6)	1–52	10 (8.6)	0–26	1.86	0.073
IQa	104 (11.8)	79–122	116 (13.4)	94–139	−2.60	0.015
	Percent		Percent	Fisher’s Exact Test or χ2
Employed full time	53		31		χ2=2.79	0.248
White	53		88			0.054
Native American	20		25			1.000
Asian	7		0			0.484
African American	7		6			1.000
Hispanic	47		13			0.054
Army	67		56			0.716
Marine Corps	0		25			0.058
Navy	13		0			0.226
Air Force	20		19			1.000
Coast Guard	0		0			N/A

Note: Participants were instructed to check all that apply for ethnicity and race categories.

^aSignificant at p<.05

^bPTSD (N=14) no PTSD (N-14)

PTSD, posttraumatic stress disorder; N, sample size; SD, standard deviation; CAPS, Clinician Administered PTSD Scale; IQ, Intelligence Quotient.

3.2 fMRI results

Main effects for the task (i.e., Target versus Standard and Novel versus Standard contrasts) are reported in the supplementary material and were similar to previously published data (Herz et al., 2016; Kiehl et al., 2001). The PTSD group showed greater response during Standards than the non-PTSD group in right frontal operculum/inferior frontal gyrus and medial/right precentral gyrus. Group differences in the contrasts of interest are reported below.

3.2.1. Targets versus Standards

Participants with PTSD showed less difference in BOLD response compared to control participants in the Targets versus Standards contrast in a cluster encompassing the bilateral lingual gyrus, bilateral occipital pole, and bilateral lateral occipital cortex and a cluster in superior parietal lobe (SPL), and pre-/post-central gyrus. (See Table 3, Figure 1, corrected for multiple comparisons at z > 2.3, cluster p < .05). In both cases, control participants showed enhanced response during the presentation of targets compared to standards, whereas PTSD participants showed only a slight increased response in occipital regions. There were no regions where participants with PTSD had greater differences than control participants during target detection.

Table 3.

Clusters showing greater differences for the Control compared to the PTSD group.

Contrast	Region	Voxels	z-score	x	y	z
Targets vs. Standards
	Bilateral Occipital Pole/Lingual Gyrus	5388	4.63	−24	−102	−4
	R Pre-/Post-central/SPL	579	3.98	44	−44	56
Novels vs. Standards
	PCC/Precuneus	1236	4.32	10	−48	24
	L MFG/Precentral Gyrus/SMA	1221	4.07	−30	18	46
	R IPL	731	3.45	38	−64	48
	R Heschl’s gyrus	673	3.61	42	−18	2
	R Occipital Pole	514	3.53	18	−88	0
	Bilateral vmPFC	436	3.53	14	64	10
	R Caudate/Thalamus	424	3.66	18	−20	24
	L Lingual Gyrus	370	3.8	−4	−88	−10

All analyses were corrected for multiple comparisons using minimum cluster sizes with all clusters surpassing the cluster level p-value of .05.

Targets vs. Standards, Targets > Standards; Novels vs. Standards, Novels > Standards; R, right; L, left; SPL, Superior Parietal Lobe; PCC, Posterior Cingulate Cortex; MFG, Middle Frontal Gyrus; SMA, Supplementary Motor Area; IPL, Inferior Parietal Lobe; vmPFC, ventromedial prefrontal cortex.

Figure 1.

Analysis of BOLD activation where the trauma control group had greater activation than the PTSD group for Targets versus Standards. All analyses were corrected for multiple comparisons using minimum cluster based thresholding (voxel z>2.3, cluster p<.05) . (A) Statistical map displaying two independent clusters: superior parietal lobe (SPL)/pre-/post-central gyrus and Bilateral Occipital Pole/Lingual Gyrus. (B) Box plots displaying mean percent signal change for Standards vs. Baseline and Targets vs. Baseline for each group (PTSD and non-PTSD). Bars in the boxes represent the mean, the outer edges of the boxes represent the lower and upper quartiles of the data around the means, and the ends of the bars represent the max and min of the data, excluding outliers, which are shown as separate dots.

3.2.2. Novels versus Standards

Participants with PTSD compared to control participants also showed reduced BOLD response in the Novels versus Standards condition in PCC/precuneus, right occipital pole, left superior/middle frontal gyrus (SFG/MFG), right IPL, right Heschl’s gyrus/STG, right occipital pole, vmPFC, right caudate/thalamus, and right lingual gyrus. (See Table 3, Figure 2, corrected for multiple comparisons z > 2.3, at p < .05.) In all cases, control participants showed a relatively greater response during novels compared to standards, whereas PTSD participants showed no change. There were no regions where participants with PTSD had a greater response than control participants during novelty processing.

Figure 2.

Analysis of BOLD activation where the trauma control group had greater activation than the PTSD group for Novels versus Standards. All analyses were corrected for multiple comparisons using minimum cluster based thresholding (voxel z>2.3, cluster p<.05). (A) Statistical map displaying eight independent clusters: (1) PCC/precuneus, (2) left superior/middle frontal gyrus (MFG), supplementary motor area (SMA), precentral gyrus, (3) right IPL, (4) right Heschl’s gyrus/STG, (5) right occipital pole, (6) ventromedial prefrontal cortex (vmPFC), (7) right caudate/thalamus, and (8) right lingual gyrus. (B) Box plots displaying mean signal change for Standards vs. Baseline and Novels vs. Baseline for each group (PTSD and non-PTSD). Bars in the boxes represent the mean, the outer edges of the boxes represent the lower and upper quartiles of the data around the means, and the ends of the bars represent the max and min of the data, excluding outliers, which are shown as separate dots.

3.2.3. Exploratory analyses with additional covariates

We also examined our results with the addition of several covariates that may be important, but we didn’t include in our primary analyses because of our sample size and limited power.

Age

Entering age into the Target vs. Standard analysis resulted in a smaller, but significant cluster in left occipital pole, but the group differences in precentral gyrus and SPL were no longer significant. For the Novel vs. Standard analysis, adding age as a covariate only reduced the group difference activity observed in the occipital pole; all other regions remained significant.

IQ

For the contrast of Target vs. Standard, adding IQ resulted in a smaller cluster in bilateral occipital pole and no significant cluster in right SPL. For the contrast of Novel vs. Standard, adding IQ to the group difference model resulted in significant clusters in PCC, right SPL, and left MFG. The other clusters were no longer significant. There was no correlation between IQ and any of the task performance measures. There was no correlation between IQ and the contrast of Targets vs.Standards, but there was a significant positive correlation between IQ and Novels vs. Standards in right amygdala/hippocampus, right postcentral gyrus, and bilateral occipital fusiform gyrus/occipital pole.

Alcohol

For the contrast of Target vs.Standard, adding drinks per drinking day eliminated the cluster in right SPL, but did not affect the cluster in occipital pole. In addition, a group difference emerged in medial superior frontal gyrus when controlling for drinks per drinking day. Adding percent drinking days to the group difference model had a very similar effect as drinks per drinking day: the right SPL cluster was no longer significant, the occipital pole cluster remained significant, and a cluster including medial superior frontal gyrus and supplementary motor cortex were now significant.

For the contrast of Novel vs. Standard, adding drinks per drinking day eliminated the significant cluster in ventromedial PFC, right SPL, and right STG, but no other clusters were affected. Finally, adding percent drinking days eliminated group difference clusters in for ventromedial PFC only.

PTSD symptom severity

There was no correlation between PTSD symptom severity and any of the task performance measures. Similarly, PTSD symptom severity score showed no correlations with the Targets vs. Standards contrast. However, there was a significant negative correlation Novels vs. Standards with PTSD severity score in posterior cingulate cortex.

4. Discussion

Our findings supported the general hypothesis in that individuals with PTSD, relative to trauma-controls, showed decreased engagement of attentional processing networks during target detection and novelty processing, and specifically that there was a reduced PTSD-associated response in the dlPFC (left SFG, MFG) during novelty processing. Our overall finding that participants with PTSD on average showed little or no increase in activation to targets or novels versus standards, while control participants had greater activation to both targets and novels versus standards is consistent with the recent MEG results described in Herz, et al. (Herz et al., 2016). We also identified additional regions (bilateral lingual gyrus, occipital pole, lateral occipital cortex, SPL, and pre-/post-central gyrus) where participants with PTSD had less activation during target detection than trauma controls. Moreover, we found several regions (PCC/precuneus, right occipital pole, SFG/MFG, right IPL, right Heschl’s gyrus/STG, right occipital pole, vmPFC, right caudate/thalamus, and right lingual gyrus) associated with attention (Corbetta and Shulman, 2002; Kim, 2014; Leech et al., 2012; Singh-Curry and Husain, 2009) where activation during novelty processing was lower in the PTSD group compared to trauma controls.

Differences emerged between trauma control and PTSD groups during both target detection (Targets vs. Standards) and novelty processing (Novels vs. Standards) similar to those found by Herz et al. (Herz et al., 2016), although there were no group differences in behavioral performance on the task. These conditions both required detection and interrogation of infrequent stimuli. Target detection required a participant response, whereas processing of novel stimuli required a non-response. During target detection, neural activity in participants with PTSD failed to increase in bilateral lingual gyrus (occipital pole) and right SPL/pre-/post-central gyrus compared to controls, consistent with the MEG findings (Herz et al., 2016). Less activation in the ventral attention system (i.e. bilateral lingual gyrus) in PTSD participants is also similar to prior ERP studies showing reduced amplitude of early auditory attentional processing in PTSD (Bae et al., 2011; Felmingham et al., 2002; Metzger et al., 2009). Lower activation may indicate reduced cognitive control and associated ability to identify task-relevant cues in the environment. Further, the lower response in the dorsal attention system suggests impaired orienting and preparation of motor responses to task relevant stimuli. Although the AOD task is one that most people do not have difficulty performing, these neural impairments may underlie difficulties experienced by PTSD patients during more challenging day-to-day tasks.

Unlike the target detection contrast, the novelty processing contrast probed not only the reaction to distracting infrequent stimuli, but also the process of identifying stimuli as irrelevant for task performance. Individuals with PTSD report substantial difficulty concentrating that impairs the ability to succeed in the workplace and in educational settings. Another distressing PTSD symptom is exaggerated startle, where individuals are hypersensitive to stimuli in the environment (noises, touch, sights) that would be easily ignored by people without PTSD. Exaggerated startle may involve disrupted novelty processing where novel stimuli are incorrectly first identified as relevant when they are not. Another potentially related PTSD symptoms is hypervigilance, where individuals have difficulty ceasing to scan the environment for threat, even when no threat is likely. Comparisons between the PTSD and trauma control groups revealed differences in novelty processing within several regions implicated in early attentional orienting and cognitive control that were also similar to the MEG results (Herz et al., 2016). First, neural activity in participants with PTSD on average lacked increased responses during novels in posterior regions including rIPL, right Heschl’s gyrus/STG, and PCC. IPL is implicated in the orienting response, and the lack of response in this region may indicate that PTSD involves disturbed control processes involved in expectancy violations (i.e., participants are expecting standard cues not novel cues) (Corbetta et al., 2008; Corbetta and Shulman, 2002; Kim, 2014). The PCC is a complex structure highly connected to cortical and subcortical networks (Vogt et al., 2006), integrating diverse types of information to monitor the environment for change (Leech et al., 2012). It interacts with the subgenual ACC to assess content and personal relevance of sensory information (Vogt et al., 2006). Thus PCC activity should be high during periods requiring broad unfocused information gathering (e.g. when anticipating cues for action) and PCC activity should fall during more specific attention demands (e.g. when detecting targets). The reduced engagement of these posterior regions in the PTSD group may be associated with exaggerated salience of novelty stimuli, consistent with to the enhanced processing of emotionally salient combat words found by Todd, et al. using MEG during the attentional blink task (Todd et al., 2015).

In addition to posterior regions, we found differences in cognitive control and motoric response regions. Because novel stimuli normally elicit an orienting response followed by suppression of motor action, the initial engagement of cognitive/motor control regions was expected. Possibly related to disturbed processing in posterior regions, participants with PTSD did not have the increased responses in vmPFC, superior/middle frontal gyrus or SMA/caudate circuits observed in control participants. These regions have been shown to be responsive to changes in the environment regardless of task relevance and may be involved in detecting those environmental changes, as well as in response planning (Downar et al., 2001; Monchi et al., 2006). Other research has found neural deficits in many of the same regions during mental flexibility tasks were associated with PTSD (Dunkley et al., 2015) and such mental flexibility deficits may underlie our findings, as the AOD task also requires some degree of mental flexibility to switch between target detection and novelty processing.

Contrary to our hypothesis, we did not find any differences between the two groups in ACC activation during novelty processing nor OFC during target detection. Lack of ACC activation may be due to differences in our novel stimuli task and the one used by Herz, et al. (Herz et al., 2016). Their study used a threatening distractor, while ours used neutral novel tones. Thus, neural abnormalities during neutral novelty processing in PTSD may differ somewhat from neural abnormalities associated with threat processing in PTSD. It is also possible that the differences in temporal resolution between fMRI and MEG contributed to the regional activation differences between our results and those of the similar MEG study (Herz et al., 2016). MEG allows the results to be divided into an early response versus a later response more typical for Target P3 versus Novelty P3, whereas the poor temporal resolution of fMRI is unable to differentiate these components with the current design.

While the current study provides insight into potential neural mechanisms underlying attentional deficits in PTSD, we also note that we did not replicate prior PTSD fMRI target detection studies (Bryant et al., 2005; Felmingham et al., 2009) that found PTSD, versus non-trauma controls, to be associated with greater activity in several regions. This may be due to differences between our using a three-tone AOD task instead of the two-tone task used previously. Indeed, our results are more similar to those found in the MEG 3-tone study. Although ERP responses to target tones correlate across two- and three-stimulus paradigms (Katayama and Polich, 1996), PTSD meta-analysis (Karl et al., 2006) showed amplitudes for target detection vary depending on distractor stimuli characteristics. Repeated disruption of attention monitoring by novelty distractor stimuli may affect neural responses to target tones, and PTSD may enhance this effect. Alternatively, the discrepancy may be due to differences in our control groups (i.e., trauma control versus non-trauma control). Finally, divergent fMRI findings for targets between our study and the two others may also not be surprising given the diverse results among ERP studies reporting disparate results for target detection in PTSD (Johnson et al., 2013; Karl et al., 2006). There are many potential design differences across these studies (e.g., control group characteristics, trauma characteristics, types of tones, task design). For example, many studies have examined participants with various types of trauma or veterans from older and recent conflicts, while we were able to recruit veterans uniformly exposed to more recent combat. This suggests the need for carefully controlled comparison groups, in order to determine potential mechanisms underlying PTSD.

Our results, in conjunction with those from other AOD studies, suggest that further examination of tasks directly relevant to the wide range of PTSD symptoms are warranted. There is a long list of things we do not know about PTSD (Jakovljević et al., 2012) including underlying pathophysiological disturbances. Responses to threat or trauma cues are more commonly examined than are responses to neutral stimuli in PTSD fMRI studies. However, electrophysiology research utilizing neutral stimuli suggests that neural abnormalities to threatening stimuli in PTSD generalize to neutral stimuli responses (Johnson et al., 2013; Karl et al., 2006). Additionally, PTSD is associated with an imbalance between salience-based and goal-directed attentional systems in studies utilizing non-threatening stimuli (Esterman et al., 2013). It is worth remembering that, although Criterion B PTSD symptoms (i.e., responses often elicited by threat or trauma cues) are disabling and disturbing to patients, DSM5 Criterion E symptoms (including problems with attention) in PTSD are potentially equally disabling to patients, as they interfere substantially with educational and vocational success. Our results also suggest that attention training interventions (Badura-Brack et al., 2015) might be developed for non-threat attention control as a treatment in PTSD.

4.1. Limitations and strengths

Although we targeted women and men, the female subsample did not reach an analyzable size. Thus, our results may not generalize to women. Similarly, we did not have the resources to recruit a third non-trauma control group, so we cannot examine differences in novelty processing due to trauma exposure alone. Additionally, in order to make inferences about PTSD specifically, we excluded veterans reporting other DSM-IV Axis 1 disorders other than a substance use disorder. This may limit generalizability to more severe groups of PTSD patients, such as those with schizophrenia, major depression, or bipolar disorder. We also conducted several tasks during the fMRI session, including an emotional faces task. It is possible that factors such as this one as well as others including being inserted into the MRI machine bore, may have increased anxiety disproportionately for the PTSD group. There were also some statistical differences in alcohol consumption between our groups, although the clinical relevance of those differences appeared small. We did not exclude people with alcohol use disorders, because this would have caused our sample to poorly represent veterans with PTSD. We did additional analyses to control for the effects of alcohol as well, and still found most of our results remained. However, our small sample size and the known challenges to measuring accurate dose/toxicity with retrospective alcohol consumption self-reports in general make it difficult to fully disentangle alcohol effects on these results. Finally, our small sample size may have limited our ability to detect smaller effects.

This study also had strengths to offset these limitations. We compared participants with PTSD to a trauma control group exposed to the same type of trauma during the same conflicts. This allowed us to draw conclusions about the role of PTSD specifically not conflated with the role of just trauma, an acknowledged limitation in the two prior AOD fMRI studies on PTSD (Bryant et al., 2005; Felmingham et al., 2009). We carefully controlled for amount of time since combat exposure by recruiting veterans from the same conflict (e.g., Afghanistan/Iraq only, not Korea or Vietnam). Limiting our sample to combat veterans also reduced potential trauma-type variability. We recruited from the community instead of the Veterans Administration Health Services (VA), thus our sample may be more representative of combat veterans overall than a VA-only sample, since many veterans do not utilize VA services. Finally, using a three-tone task allowed us to build on prior research and examine neural differences in response to attention disruption during novelty processing in addition to target detection.

4.2. Conclusions

This study reports further evidence of altered brain activity associated with PTSD during a non-threatening attention task. We found lower activation associated with PTSD during novelty processing and target detection in different default mode and attention network regions. We did not find any regions where PTSD was associated with greater activation during either condition. The AOD is an attention-demanding, but non-arousing, task that most people do not have difficulty performing. It is noteworthy that even during a task that is not overly challenging or emotionally arousing, brain activity differences remain associated with PTSD in attention and default mode networks. This finding may be especially representative of daily functioning problems in the PTSD population.

Highlights.

• Reduced activation in neural attention networks was associated with PTSD compared to combat controls during novelty processing as well as during target detection.

• Reduced activation in the dlPFC and several other attention processing regions was associated with PTSD during novelty processing.

• Reduced activation in the occipital pole and right SPL/pre/post-central gyrus was associated with PTSD during target detection.