Demographic factors predict magnitude of conditioned fear

International Journal of Psychophysiology 98 (2015) 59–64

Contents lists available at ScienceDirect

International Journal of Psychophysiology journal homepage: www.elsevier.com/locate/ijpsycho

Demographic factors predict magnitude of conditioned fear Blake L. Rosenbaum a, Eric Bui a,b, Marie-France Marin a,b, Daphne J. Holt a,b,c, Natasha B. Lasko a,b, Roger K. Pitman a,b, Scott P. Orr a,b, Mohammed R. Milad a,b,c,⁎ a b c

The Department of Psychiatry, Massachusetts General Hospital, Boston, MA, United States Harvard Medical School, Boston, MA, United States The HST-MIT Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA, United States

a r t i c l e

i n f o

Article history: Received 18 November 2014 Received in revised form 21 May 2015 Accepted 26 June 2015 Available online 4 July 2015 Keywords: Fear conditioning Demographics Classical conditioning Education Extinction Skin conductance

a b s t r a c t There is substantial variability across individuals in the magnitudes of their skin conductance (SC) responses during the acquisition and extinction of conditioned fear. To manage this variability, subjects may be matched for demographic variables, such as age, gender and education. However, limited data exist addressing how much variability in conditioned SC responses is actually explained by these variables. The present study assessed the influence of age, gender and education on the SC responses of 222 subjects who underwent the same differential conditioning paradigm. The demographic variables were found to predict a small but significant amount of variability in conditioned responding during fear acquisition, but not fear extinction learning or extinction recall. A larger differential change in SC during acquisition was associated with more education. Older participants and women showed smaller differential SC during acquisition. Our findings support the need to consider age, gender and education when studying fear acquisition but not necessarily when examining fear extinction learning and recall. Variability in demographic factors across studies may partially explain the difficulty in reproducing some SC findings. © 2015 Published by Elsevier B.V.

1. Introduction Classical conditioning has been used to explore fear-based learning associated with clinical anxiety (Grillon and Morgan, 1999; Pitman and Orr, 1986) as well as to elucidate mechanisms associated with the psychopathology of several psychiatric disorders, including posttraumatic stress disorder (PTSD) (Blanchard et al., 1996; Garfinkel et al., 2014), obsessive compulsive disorder (OCD) (McLaughlin et al., 2015; Nanbu et al., 2010), and schizophrenia (Graham and Milad, 2011; Milad and Quirk, 2012). A typical fear-conditioning paradigm involves the pairing of a neutral cue (conditioned stimulus, CS) with an aversive stimulus (unconditioned stimulus, US), such as an electric shock, that produces an unconditioned response (UR). Subsequent to a series of CS–US pairings, presentation of the CS comes to elicit a conditioned response (CR) when presented alone. In human fear conditioning, physiological measures such as startle, heart rate, or skin conductance (SC) are commonly used to measure the CR (Jovanovic et al., 2012; Orr and Roth, 2000; Vervliet et al., 2004). During extinction training, when the CS is no longer paired with the US, the CR will diminish, i.e. extinguish, over repeated presentations. Both increased fear conditionability and the diminished ability to extinguish fear have been suggested to be

⁎ Corresponding author at: Department of Psychiatry, Massachusetts General Hospital, 149 13th St, CNY 2614, Charlestown, MA 02129, United States. E-mail address: [email protected] (M.R. Milad).

http://dx.doi.org/10.1016/j.ijpsycho.2015.06.010 0167-8760/© 2015 Published by Elsevier B.V.

involved in the pathophysiology of anxiety disorders (Milad and Quirk, 2012; Pitman et al., 2012; Rauch et al., 2006; Shin and Liberzon, 2010). Substantial individual differences in conditionability characterize human fear conditioning studies. Some individuals rapidly acquire a strong CR that is relatively specific to the fear cue; others may acquire a CR that generalizes to non-conditioned cues, while still others fail to acquire a CR. It has been suggested that genetic, developmental and personality factors contribute to individual differences in conditionability (Hettema et al., 2003; Merrill et al., 1999). Several studies have examined the relationship between conditioned fear responses and various personality traits, especially anxiety. For example, a recent study of 46 healthy Puerto Rican subjects found that the combination of selfreport measures of personality traits and measures of physiological reactivity predicted as much as 45% of the variance observed during fear conditioning (Martinez et al., 2012). Similarly, Otto and colleagues reported that self-report measures of anxiety symptoms, mood, and personality explained about 28% of the variance in fear conditioning (Otto et al., 2007). The potential effect of demographic factors such as age, gender and education on fear conditioning is generally treated as nuisance variance in fear-conditioning studies. Consequently, studies often seek to match experimental and control groups for these variables so as to mitigate their possible influence(s) on the findings. It is unclear whether matching for these variables does, in fact, reduce variance in the conditioning indices being measured. And even if matched within a particular

60

B.L. Rosenbaum et al. / International Journal of Psychophysiology 98 (2015) 59–64

study, differences in demographic characteristics between laboratories are difficult to avoid and could potentially contribute to the lack of consistent findings across studies. In order to address this issue, we have pooled data from 222 participants that were studied in our laboratory over the past several years. This sample includes healthy participants, as well as PTSD, OCD and schizophrenia patients who underwent the same differential conditioning procedure that used SC as the measure of conditioned fear (Holt et al., 2012; Milad et al., 2009). Given the relative lack of previous research examining this domain, it is difficult to hypothesize which of these demographic factors may be associated with the acquisition and extinction of fearconditioned SC. 2. Materials and methods 2.1. Participants 141 healthy controls and 81 patients (OCD, n = 21; PTSD, n = 40; schizophrenia, n = 20) who were recruited for one of three different studies and underwent the same two-day fear conditioning protocol while in an fMRI scanner were studied. The age range for all participants was 18 to 67 years old. Neuroimaging and psychophysiological data from these participants related to the neurobiology of fear extinction in healthy subjects (Linnman et al., 2011a, 2012; Milad et al., 2007a,b) and in the different clinical populations (Holt et al., 2012; Linnman et al., 2011b; Rougemont-Bucking et al., 2011) have been previously published. 2.2. Fear conditioning paradigm A detailed description of the fear conditioning paradigm is available elsewhere (Milad et al., 2007b). Briefly, participants were presented with images of two different rooms that provided the visual context, one specific to acquisition and one specific to extinction for 3 s. Each context contained a lamp that was first presented in the off position and that then “turned on” to one of three colors: blue, red, or yellow. The colors represented different conditioned stimuli (CSs) that signaled whether or not the participant would receive an electric shock (CS+) or no shock (CS−) and were presented for 6 s. The shock was delivered through two electrodes placed on the right hand and lasted 500 ms, immediately following offset of the CS + presentations. The shock level was previously selected by the participant to be “highly annoying but not painful.” Day 1 of the protocol consisted of Habituation, Conditioning (acquisition), and Extinction Learning phases, and Day 2 consisted of Extinction Recall and Renewal phases. During the Habituation phase, the two CS + s (4 trials each) and the CS− (4 trials) were presented within both the conditioning and extinction context; no shocks were delivered during this phase. In the Conditioning phase, participants were presented with two CS + s (for example, one might be represented by a red light and the other represented by a blue light) within a specific conditioning context and paired with the electric shock (8 trials for each CS+, 62% partial reinforcement). The CS− (for example, a yellow light) was presented without shocks (16 trials). All CS's were counterbalanced and presented in a pseudo-random order. In the Extinction Learning phase, only one of the CS + s was extinguished (CS + E), in a novel extinction context and without shocks. The other CS + was not presented and remained unextinguished (CS + U). The CS + E (16 trials) and CS− (16 trials) were both presented during extinction learning. On the following day, during the Extinction Recall phase, the CS + E (8 trials), the CS + U (8 trials), and the CS− (16 trials) were presented within the extinction learning context (without the US), whereas during the Renewal phase the three CSs were presented in the conditioning context (without the US). As part of the protocol and prior to fear conditioning, participants completed several questionnaires including the Beck Anxiety (BAI) and Depression Inventories

(BDI), the Anxiety Sensitivity Index (ASI), and the State-Trait Anxiety Inventory (STAI-T). 2.3. Skin conductance recording A SC response (SCR) was calculated by subtracting the mean SC level (SCL) during the 2 s immediately preceding CS onset from the highest SCL recorded during the CS presentation. This algorithm differs slightly from the standardized method for calculating SCRs because a negative value can be produced. Therefore, throughout the text, “SCR” and “change in SC” are used interchangeably. Each SCR was square-root transformed; for negative SCRs, the square-root of the absolute value of the SC change was obtained and then the negative sign was replaced. These data were then averaged across respective CS+ and CS− trials. Differential responses were calculated by subtracting the average SCR to the CS − trials from the average SC change to the respective CS + (both E and U) trials. 2.4. Data analyses All data used in the analyses were previously collected and compiled into a master database containing demographic, SCR, and psychometric data. Because the distribution was not normal in these populations, age was dichotomized into an older (≥29 years old) and a younger group (b29 years old) using a median split. The actual median value was rounded to the closest whole number. Similarly, education was dichotomized into a lower education group (b16 years of education) and a higher education group (≥16 years of education). Sixteen years of education is equivalent to the completion of a Baccalaureate degree in the United States. Data analyses were initially conducted for all experimental phases (Conditioning, Extinction Learning, Extinction Recall, and fear Renewal). Because there were no significant relationships between the demographic variables for phases other than Conditioning, the analyses presented below focus on the Conditioning phase alone. For this phase, separate 2 × 2 × 2 analyses of variance (ANOVA) were conducted that included: patient status (patient, healthy control) as a between-group factor; stimulus type (CS+, CS−) as a within-subject factor; and demographic variable: age (younger, older), gender (men, women), or education (low, high) as a between-subject factor. Significant interaction effects were then decomposed using additional ANOVA or independent sample t-tests, as appropriate. A series of multiple regression analyses were used to examine whether age, gender and education predicted SC change to the CS− and CS+ presentations, as well as the differential SCR, after adjusting for patient status. All analyses were performed using SPSS v22. 3. Results For the entire sample (n = 222), the mean age (standard deviation) was 35.6 (15.2); 126 of the participants (56.8%) were women. The majority of participants had an education of 16 years or more, (n = 131, 59%) and 108 (48.6%) were 29 years of age or older. The overall range of education was 8 years to 24 years. 3.1. Effects for age 3.1.1. Conditioning A 3-factor ANOVA (patient status, age, stimulus) produced a marginal main effect for age (F(1,218) = 3.49, p = .062), indicating a larger change in SC for the older group. There was no significant stimulus × age interaction (F(1,218) = 1.95, p = .164). There was no significant main effect or interactions involving patient status (all F's(1,218) b 1, p's = ns) There was a significant main effect for stimulus as well (F(1,218) = 111.85, p b .001). In order to decompose the stimulus × age interaction (See Fig. 1), SCRs to the CS+ and CS− were separately compared between the two age groups. Results of these comparisons showed that SCRs to

B.L. Rosenbaum et al. / International Journal of Psychophysiology 98 (2015) 59–64

Effect of Age

A

Healthy

Younger than 29

0.4

29 and older

*

61

was not significant, we explored whether or not there might be tendencies for men and women to respond differently to the CS + and CS −. Comparisons of SCRs to the CS+ yielded a marginal difference between genders (F(1,218) = 3.413, p = .066); comparisons of SCRs to the CS− suggest a more substantial difference between men and women (F(1,218) = 8.652, p = 0.004). Overall, men exhibited smaller SCRs than women (see Fig. 2).

0.2

0.1

0 Differential

Patient

B 0.4 0.3

ΔSC (uS)

AVG CS-

3.3. Effects for education

*

0.2

0.1

0 AVG CS+

AVG CS-

Differential

Fig. 1. Effects of age on conditioned responses during the fear acquisition phase for (A) healthy subjects (n = 141) and (B) patients (n = 81). Data are shown for the conditioned stimulus paired with the shock (CS+), conditioned stimulus not paired with the shock (CS−), and the differential responding (CS+–CS−). Bar figures show data from conditioning. The age grouping was based on a median split of the entire cohort. *p b 0.05.

the CS + did not differ between the younger and older cohorts (F(1, 221) b 1, p = ns); whereas, SCRs to the CS− were significantly smaller in the younger, compared to the older, participants (F(1,221) = 11.48, p = .015). This suggests stronger differential conditioning in the younger, compared to older, group (See Fig. 1). 3.1.2. Psychometrics A 2-factor ANOVA (patient status, age) produced significant main effects for patient status for: STAI-T (F(1,139) = 70.96, p b .001) ASI (F(1,139) = 41.95, p b .001), BDI (F(1,139) = 69.08, p b .001), and BAI (F(1,139) = 80.32, p b .001.) There was a significant main effect for age for BDI (F(1,139) = 8.97, p = .003). There were no significant patient status × age interactions (all F(1,139)'s b 2.0, p's N .204). The BDI was entered in as covariate in the 3-factor ANOVA and a main effect of patient status was found F(1,213) = 3.98, p = .047. 3.2. Effects for gender 3.2.1. Conditioning The 3-factor ANOVA (patient status, gender, stimulus) yielded a significant main effect of gender (F(1,218) = 6.690, p = .010), indicating an overall larger change in SC for women, compared to men. There was also a significant main effect of stimulus (F(1,218) = 112.80, p b .001), indicating a larger change in SC to the CS + s, compared to CS −, as would be expected for the Conditioning phase. The main effect for patient status was not significant (F(1,218) = 1.54, p = .302) and there were no significant interactions (all F's(1,218) b 1.10, p's = ns). Although the gender × stimulus interaction

3.3.1. Conditioning The 3-factor ANOVA that included education yielded a significant education × stimulus type interaction (F(1,218) = 17.348, p b .001). The main effects for patient status and education were not significant (all F's(1,218) b 1, p's = ns). Decomposing the education × stimulus interaction (Fig. 3), a comparison of SCRs to the CS + between the higher and lower educated groups revealed comparably large SC magnitudes (F(1,218) = 2.634, p = .106). However, comparison of SC magnitudes to the CS− were found to be smaller in the higher educated group (F(1,218) = 4.973, p = 0.027), suggesting better differential conditioning in the higher educated participants.

Effect of Gender

A

Healthy

Male Female

ΔSC (uS)

AVG CS+

3.2.2. Psychometrics A 2-factor ANOVA (patient status, gender) produced significant main effects for patient status for: STAI-T (F(1,139) = 101.80, p b .001) ASI (F(1,139) = 48.16, p b .001), BDI (F(1,139) = 94.41, p b .001), and BAI (F(1,139) = 99.85, p b .001). There were significant main effects for gender for: STAI-T (F(1,139) = 4.87, p = .029) and BAI (F(1,139) = 6.22, p = .014). However, there was no significant patient status × gender interactions (all F(1,139)'s b 1, p's N .3). The STAI-T and BAI scores were entered into the 3-factor ANOVA as covariates and yielded similarly significant results.

B

*

Patient

ΔSC (uS)

ΔSC (uS)

0.3

Fig. 2. Effects of gender on conditioned responses during the fear acquisition phase for (A) healthy subjects (n = 141) and (B) patients (n = 81). Data are shown for the conditioned stimulus paired with the shock (CS+), conditioned stimulus not paired with the shock (CS−), and the differential responding (CS+–CS−). Bar figures show data from conditioning. *p b 0.05.

62

B.L. Rosenbaum et al. / International Journal of Psychophysiology 98 (2015) 59–64

Models for Differential

Effect of Education

A

<16 years of education

Healthy

Healthy Controls

Patients

Education

Education p=.050

>16 years of education

*

ΔSC (uS)

**

p=.015

Diff

Age

Gender

r2=.070, p=.019

B

Patient

ΔSC (uS)

*

**

Fig. 3. Effects of education on conditioned responses during the fear acquisition phase for (A) healthy subjects (n = 141) and (B) patients (n = 81). Data are shown for the conditioned stimulus paired with the shock (CS+), conditioned stimulus not paired with the shock (CS−), and the differential responding (CS+–CS−). Bar figures show data from conditioning. *p b 0.05, **p b 0.01.

3.3.2. Psychometrics A 2-factor ANOVA (patient status, education) produced significant main effects for patient status for: STAI-T (F(1,139) = 92.17, p b .001) ASI (F(1,139) = 47.61, p b .001), BDI (F(1,139) = 97.137, p b .001), and BAI (F(1,139) = 86.38, p b .001). There was a significant education main effect for BDI (F(1,139) = 5.59, p = 0.02).There were no significant patient status × education interactions (all F(1,139)'s b 2.00, p's N .16). The BDI scores were entered into the 3-factor ANOVA for education and did not alter the significance of any findings.

3.4. Multiple regression models To test the potential collective effect of all three demographic factors on SC during fear conditioning, we conducted a series of multiple linear regressions that predicted response to CS+, response to CS− and the differential SCR, entering the three demographic variables to assess their collective contribution to the SC variance. The regression model that included age, gender and education as predictors of the differential SCR, only reached statistical significance for the Healthy control group (F(3,137) = 3.41, R2 = 0.070, p = .019), explaining 7.0% of the variance. In this model, education was independently associated with the differential SCR (B(SE) = 0.23(0.009), p = .015). A similar model conducted among patients was not significant (Fig. 4). The regression model including age, gender and education to predict SCR to the CS + was not significant for patient group: F(3, 77) = 1.12, R2 = 0.042, p = 0.347, but yielded a significant relationship for the Healthy control group: F(3,137) = 3.67, R2 = 0.074, p = 0.014. Similar to the differential model, education was independently associated with response to the CS + (B(SE) = .037(.013), p = .005). The regression model including age, gender and education to predict SC to the CS − yielded statistically significant relationships for the patient (F(3, 77) = 4.03,

Diff

Age

Gender

Model not significant

Fig. 4. Multiple regression models taking into consideration the influence of all demographic variables on differential responding for healthy controls and patients. Shaded circles indicate the factor with most significant contribution to the model predicting change in skin conductance.

R2 = 0.35, p = .010) as well as the healthy control group (F(3, 137) = 4.43, R2 = 0.088, p = .005, Fig. 5). In the patient group, age was independently associated with a larger change in SC to the CS − (B(SE) = 0.004(0.001), p = 0.006) while in the healthy control group, gender was independently associated with SC changes to the CS− (B(SE) = 0.116(0.047), p = 0.015, (Fig. 6).

4. Discussion The findings presented suggest that age, gender, and education contribute to variance in SC during the acquisition phase of fear conditioning, but not during extinction learning or extinction recall. Older participants, regardless of psychopathology, tend to show a smaller differential SCR that is due to increased responding to the CS−. When covarying for the psychometric measures, however, this difference is no longer significant for age. Gender differences during fear conditioning primarily reflected an exaggerated SC to the CS − in healthy women. Findings pertaining to level of education show larger differential responses in subjects with at least a college education. Although both patients and healthy controls with higher education showed larger differential SCRs, the reason for these larger SCRs differed between the groups. In the healthy group, the larger differential SCR in the college graduates was due to larger responses to the CS +, whereas in the patient group the larger differential SCR in those with college degrees

Models for CS+ Patients

Healthy Controls Education

Education

p=.005

Age

CS+

r2=.074, p=.014

Gender

Age

CS+

Gender

Model not significant

Fig. 5. Multiple regression models taking into consideration the influence of all demographic variables on SCR to the CS+ for healthy controls and patients. Shaded circles indicate the factor with most significant contribution to the model predicting change in skin conductance.

B.L. Rosenbaum et al. / International Journal of Psychophysiology 98 (2015) 59–64

Models for CSHealthy Controls

Patients

Education

Education

Age

CS-

r2=.088, p=.005

Gender p=.015

Age p=.006

CS-

Gender

r2=.135, p=.010

Fig. 6. Multiple regression models taking into consideration the influence of all demographic variables on SCR to the CS− for healthy controls and patients. Shaded circles indicate the factor with most significant contribution to the model predicting change in skin conductance.

was due to smaller responses to the CS −. Thus, healthy individuals responded more strongly to the fear cue, whereas patients responded less strongly to the safety cue. Results from our multi-linear regression models examined the collective contribution of the demographic factors. Results from these models further implicated education as the most significant contributor to the variance in differential SCR during fear conditioning, while age and gender were factors that contributed to variance in responding to the CS −. These models were able to explain 7%–14% of the SC variance during fear conditioning. Neither extinction learning nor subsequent recall of the extinction memory was influenced by any of the demographic factors that we examined. A similar study that examined personality and anxiety traits also found no relationships with fear extinction and extinction recall (Martinez et al., 2012). Thus, our findings and those of Martinez and colleagues suggest that the effect of demographic factors may be limited to the acquisition of a conditioned fear response. In contrast, resting metabolism in the dorsal anterior cingulate cortex (dACC) was found to predict SCR during extinction recall in healthy subjects (Linnman et al., 2012). It may be that a biologically-based dimensional measure will prove to be a better predictor of fear extinction than demographic measures. The differential SCR provides a measure of one's ability to distinguish fear cues from safety cues; a smaller differential score suggests that the participant has greater difficulty discriminating the CS+ and CS−, or may have failed to respond to either one. The present results indicate that participants with more education showed larger differential SCRs. Importantly, in the patient group, the high and low education subgroups did not differ significantly in the magnitude of their SC to the CS+, suggesting that the larger differential SCR of the high education group was not simply a consequence of increased reactivity to the fear cue, but rather a better recognition of the safety cue. In the healthy group, however, the difference seems to reflect a larger SC change to the CS + in the higher education group suggesting a greater amount of conditioned responding. Thus, given these results, a higher level of education facilitates discriminatory learning either by increased responding to the fear cue or decreased responding to the safety cue. Results of regression analyses further support this finding, as education was the only individual difference variable that reached significance in both the differential SCR model and the model for responses to the CS+. Although education and IQ are likely associated (Ritchie et al., 2012), it is not clear whether IQ contributed to the difference between the high versus low education groups in differential SCRs. Education may enhance one's ability to recognize patterns such that individuals with more education become more adept at pattern recognition and

63

therefore show greater differential conditioning, due to greater recognition of the CS–US contingency. It is also possible that, given the stressful nature of higher education, those participants who have greater educational experience are more adept at coping with and managing stress. Stress in school and test-anxiety are often found to be strong predictors of school performance (Fontana and Dovidio, 1984). Thus, individuals who develop better coping mechanisms for regulating autonomic responses to anxiogenic stimuli may be more successful in school and therefore, are more likely to continue to pursue higher education. Although the differential SCR did not appear to be significantly different in the two age groups we examined, the change in SC was larger to the CS− in the older participants. Reduced conditioning in aging has been previously reported with eye-blink conditioning in healthy subjects (Bellebaum and Daum, 2004; Cheng et al., 2010). Although older individuals have been reported to show overall diminished responses (Labar et al., 2004), in the present study the older cohort was somewhat more reactive than the younger group, especially in their responses to the CS −. One possible interpretation of this finding is that the ability to clearly distinguish between the fear and safety cues is reduced with age, suggesting increased fear generalization to cues presented within a threatening situation. Labar and colleagues also noted that awareness of the CS–US contingency decreased with age. Klucken and colleagues found that contingency awareness may affect psychophysiological responses like SCR, however these differences did not extend to brain activity (Klucken et al., 2009). Deficits in working memory associated with general aging are another possible source of decreased differential responding. Gender did not affect differential conditioning but did influence responses to the CS −; women showed a larger change in SC to the CS − compared to men, contrary to our predictions. Some studies have found that men and women show comparable differential SCRs during fear conditioning (Lebron-Milad et al., 2012a; Zorawski et al., 2005). We have previously observed significantly larger differential SCRs during fear conditioning in men, compared to women (Milad et al., 2006, 2010). In patients diagnosed with PTSD, women were found to exhibit larger differential SCRs during fear conditioning, compared to men (Inslicht et al., 2013). These differences may be due to variability in gonadal or stress hormones in the participants of different studies, and suggest that perhaps given a sufficiently large enough of sample, sex-differences in fear conditioning may be negligible. Nevertheless, subjects' menstrual cycle was not tracked in the present study, so potential sex differences may be occluded. The complexity of the role that sex differences may play during fear acquisition supports the need for further research in this area (Jazin and Cahill, 2010; Lebron-Milad et al., 2012b). Results of the present study may help to explain, in part, the lack of consistency in psychophysiological findings across studies. For example, we've previously shown that PTSD patients did not differ in SC changes to the CS + compared to trauma-exposed normal controls (Linnman et al., 2011b). Several other studies have also found no group differences in fear acquisition between patient and healthy groups (Blechert et al., 2007; Peri et al., 2000). Conversely, other studies have shown heightened psychophysiological responding in PTSD populations (Keane et al., 1998). While some of these discrepancies may be due to the different types of trauma experienced by the different cohorts in these studies, our findings suggest that demographic differences among the studied populations may also contribute to the seemingly inconsistent findings. Peri and colleagues did see group differences when controlling for age, which further suggests the possible role of demographic factors. Nevertheless, several factors limit the conclusions that can be drawn from these data. The study population is limited to healthy controls and subjects with PTSD, OCD, and schizophrenia, and therefore the findings cannot be applied to people with other anxiety disorders. In summary, our data support the continuation of controlling for demographic factors as precisely as possible when examining fear conditioning, and especially when comparing findings across studies.

64

B.L. Rosenbaum et al. / International Journal of Psychophysiology 98 (2015) 59–64

Acknowledgments This work was supported by the National Institute of Mental Health grant 1R01MH097964 and institutional funds from the Department of Psychiatry at MGH to MRM. The authors would like to thank the Translation Neuroscience team for their help preparing this manuscript. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijpsycho.2015.06.010. References Bellebaum, C., Daum, I., 2004. Effects of age and awareness on eyeblink conditional discrimination learning. Behav. Neurosci. 118 (6), 1157–1165. http://dx.doi.org/10. 1037/0735-7044.118.6.1157. Blanchard, E.B., Hickling, E.J., Buckley, T.C., Taylor, A.E., Vollmer, A., Loos, W.R., 1996. Psychophysiology of posttraumatic stress disorder related to motor vehicle accidents: replication and extension. J. Consult. Clin. Psychol. 64 (4), 742–751. Blechert, J., Michael, T., Vriends, N., Margraf, J., Wilhelm, F.H., 2007. Fear conditioning in posttraumatic stress disorder: evidence for delayed extinction of autonomic, experiential, and behavioural responses. Behav. Res. Ther. 45 (9), 2019–2033. Cheng, D.T., Faulkner, M.L., Disterhoft, J.F., Desmond, J.E., 2010. The effects of aging in delay and trace human eyeblink conditioning. Psychol. Aging 25 (3), 684–690. http://dx.doi.org/10.1037/a0017978. Fontana, A., Dovidio, J.F., 1984. The relationship between stressful life events and schoolrelated performances of type A and type B adolescents. J. Hum. Stress. 10 (1), 50–55. http://dx.doi.org/10.1080/0097840X.1984.9934959. Garfinkel, S.N., Abelson, J.L., King, A.P., Sripada, R.K., Wang, X., Gaines, L.M., Liberzon, I., 2014. Impaired contextual modulation of memories in PTSD: an fMRI and psychophysiological study of extinction retention and fear renewal. J. Neurosci. 34 (40), 13435–13443. http://dx.doi.org/10.1523/JNEUROSCI.4287-13.2014. Graham, B.M., Milad, M.R., 2011. The study of fear extinction: implications for anxiety disorders. Am. J. Psychiatry 168 (12), 1255–1265. http://dx.doi.org/10.1176/appi. ajp.2011.11040557. Grillon, C., Morgan III, C.A., 1999. Fear-potentiated startle conditioning to explicit and contextual cues in Gulf War veterans with posttraumatic stress disorder. J. Abnorm. Psychol. 108 (1), 134–142. Hettema, J.M., Annas, P., Neale, M.C., Kendler, K.S., Fredrikson, M., 2003. A twin study of the genetics of fear conditioning. Arch. Gen. Psychiatry 60 (7), 702–708. http://dx. doi.org/10.1001/archpsyc.60.7.702. Holt, D.J., Coombs, G., Zeidan, M.A., Goff, D.C., Milad, M.R., 2012. Failure of neural responses to safety cues in schizophrenia. Arch. Gen. Psychiatry 69 (9), 893–903. http://dx.doi.org/10.1001/archgenpsychiatry.2011.2310 (1356406 [pii]). Inslicht, S.S., Metzler, T.J., Garcia, N.M., Pineles, S.L., Milad, M.R., Orr, S.P., Neylan, T.C., 2013. Sex differences in fear conditioning in posttraumatic stress disorder. J. Psychiatr. Res. 47 (1), 64–71. http://dx.doi.org/10.1016/j.jpsychires.2012.08.027. Jazin, E., Cahill, L., 2010. Sex differences in molecular neuroscience: from fruit flies to humans. Nat. Rev. Neurosci. 11 (1), 9–17. http://dx.doi.org/10.1038/nrn2754. Jovanovic, T., Kazama, A., Bachevalier, J., Davis, M., 2012. Impaired safety signal learning may be a biomarker of PTSD. Neuropharmacology 62 (2), 695–704. http://dx.doi. org/10.1016/j.neuropharm.2011.02.023. Keane, T.M., Kolb, L.C., Kaloupek, D.G., Orr, S.P., Blanchard, E.B., Thomas, R.G., Lavori, P.W., 1998. Utility of psychophysiological measurement in the diagnosis of posttraumatic stress disorder: results from a Department of Veterans Affairs Cooperative Study. J. Consult. Clin. Psychol. 66 (6), 914–923. Klucken, T., Kagerer, S., Schweckendiek, J., Tabbert, K., Vaitl, D., Stark, R., 2009. Neural, electrodermal and behavioral response patterns in contingency aware and unaware subjects during a picture–picture conditioning paradigm. Neuroscience 158 (2), 721–731. http://dx.doi.org/10.1016/j.neuroscience.2008.09.049. Labar, K.S., Cook, C.A., Torpey, D.C., Welsh-Bohmer, K.A., 2004. Impact of healthy aging on awareness and fear conditioning. Behav. Neurosci. 118 (5), 905–915. http://dx.doi. org/10.1037/0735-7044.118.5.905. Lebron-Milad, K., Abbs, B., Milad, M.R., Linnman, C., Rougemount-Bucking, A., Zeidan, M.A., Goldstein, J.M., 2012a. Sex differences in the neurobiology of fear conditioning and extinction: a preliminary fMRI study of shared sex differences with stress-arousal circuitry. Biol. Mood Anxiety Disord. 2 (1), 7. http://dx.doi.org/10.1186/2045-5380-2-7. Lebron-Milad, K., Graham, B.M., Milad, M.R., 2012b. Low estradiol levels: a vulnerability factor for the development of posttraumatic stress disorder. Biol. Psychiatry 72 (1), 6–7. http://dx.doi.org/10.1016/j.biopsych.2012.04.029. Linnman, C., Rougemont-Bucking, A., Beucke, J.C., Zeffiro, T.A., Milad, M.R., 2011a. Unconditioned responses and functional fear networks in human classical conditioning. Behav. Brain Res. 221 (1), 237–245. http://dx.doi.org/10.1016/j.bbr.2011.02.045.

Linnman, C., Zeffiro, T.A., Pitman, R.K., Milad, M.R., 2011b. An fMRI study of unconditioned responses in post-traumatic stress disorder. Biol. Mood Anxiety Disord. 1 (1), 8. http://dx.doi.org/10.1186/2045-5380-1-8. Linnman, C., Zeidan, M.A., Furtak, S.C., Pitman, R.K., Quirk, G.J., Milad, M.R., 2012. Resting amygdala and medial prefrontal metabolism predicts functional activation of the fear extinction circuit. Am. J. Psychiatry 169 (4), 415–423. http://dx.doi.org/10.1176/appi. ajp.2011.10121780. Martinez, K.G., Castro-Couch, M., Franco-Chaves, J.A., Ojeda-Arce, B., Segura, G., Milad, M.R., Quirk, G.J., 2012. Correlations between psychological tests and physiological responses during fear conditioning and renewal. Biol. Mood Anxiety Disord. 2 (1), 16. http://dx.doi.org/10.1186/2045-5380-2-16. McLaughlin, N.C., Strong, D., Abrantes, A., Garnaat, S., Cerny, A., O'Connell, C., Greenberg, B.D., 2015. Extinction retention and fear renewal in a lifetime obsessive–compulsive disorder sample. Behav. Brain Res. 280, 72–77. http://dx. doi.org/10.1016/j.bbr.2014.11.011. Merrill, K.A., Steinmetz, J.E., Viken, R.J., Rose, R.J., 1999. Genetic influences on human conditionability: a twin study of the conditioned eyeblink response. Behav. Genet. 29 (2), 95–102. Milad, M.R., Quirk, G.J., 2012. Fear extinction as a model for translational neuroscience: ten years of progress. Annu. Rev. Psychol. 63, 129–151. http://dx.doi.org/10.1146/ annurev.psych.121208.131631. Milad, M.R., Goldstein, J.M., Orr, S.P., Wedig, M.M., Klibanski, A., Pitman, R.K., Rauch, S.L., 2006. Fear conditioning and extinction: influence of sex and menstrual cycle in healthy humans. Behav. Neurosci. 120 (6), 1196–1203. http://dx.doi.org/10.1037/ 0735-7044.120.5.1196 (2006-22387-002 [pii]). Milad, M.R., Quirk, G.J., Pitman, R.K., Orr, S.P., Fischl, B., Rauch, S.L., 2007a. A role for the human dorsal anterior cingulate cortex in fear expression. Biol. Psychiatry 62 (10), 1191–1194. http://dx.doi.org/10.1016/j.biopsych.2007.04.032 (S00063223(07)00401-5 [pii]). Milad, M.R., Wright, C.I., Orr, S.P., Pitman, R.K., Quirk, G.J., Rauch, S.L., 2007b. Recall of fear extinction in humans activates the ventromedial prefrontal cortex and hippocampus in concert. Biol. Psychiatry 62 (5), 446–454. http://dx.doi.org/10.1016/j.biopsych. 2006.10.011 (S0006-3223(06)01298-4 [pii]). Milad, M.R., Pitman, R.K., Ellis, C.B., Gold, A.L., Shin, L.M., Lasko, N.B., Rauch, S.L., 2009. Neurobiological basis of failure to recall extinction memory in posttraumatic stress disorder. Biol. Psychiatry 66 (12), 1075–1082. http://dx.doi.org/10.1016/j.biopsych. 2009.06.026 (S0006-3223(09)00896-8 [pii]). Milad, M.R., Zeidan, M.A., Contero, A., Pitman, R.K., Klibanski, A., Rauch, S.L., Goldstein, J.M., 2010. The influence of gonadal hormones on conditioned fear extinction in healthy humans. Neuroscience 168 (3), 652–658. http://dx.doi.org/10.1016/j. neuroscience.2010.04.030. Nanbu, M., Kurayama, T., Nakazawa, K., Matsuzawa, D., Komiya, Z., Haraguchi, T., Shimizu, E., 2010. Impaired P50 suppression in fear extinction in obsessive–compulsive disorder. Prog. Neuropsychopharmacol. Biol. Psychiatry 34 (2), 317–322. http://dx.doi.org/ 10.1016/j.pnpbp.2009.12.005. Orr, S.P., Roth, W.T., 2000. Psychophysiological assessment: clinical applications for PTSD. J. Affect. Disord. 61 (3), 225–240. Otto, M.W., Leyro, T.M., Christian, K., Deveney, C.M., Reese, H., Pollack, M.H., Orr, S.P., 2007. Prediction of “fear” acquisition in healthy control participants in a de novo fearconditioning paradigm. Behav. Modif. 31 (1), 32–51. http://dx.doi.org/10.1177/ 0145445506295054. Peri, T., Ben-Shakhar, G., Orr, S.P., Shalev, A.Y., 2000. Psychophysiologic assessment of aversive conditioning in posttraumatic stress disorder. Biol. Psychiatry 47 (6), 512–519. Pitman, R.K., Orr, S.P., 1986. Test of the conditioning model of neurosis: differential aversive conditioning of angry and neutral facial expressions in anxiety disorder patients. J. Abnorm. Psychol. 95 (3), 208–213. Pitman, R.K., Rasmusson, A.M., Koenen, K.C., Shin, L.M., Orr, S.P., Gilbertson, M.W., Liberzon, I., 2012. Biological studies of post-traumatic stress disorder. Nat. Rev. Neurosci. 13 (11), 769–787. http://dx.doi.org/10.1038/nrn3339. Rauch, S.L., Shin, L.M., Phelps, E.A., 2006. Neurocircuitry models of posttraumatic stress disorder and extinction: human neuroimaging research—past, present, and future. Biol. Psychiatry 60 (4), 376–382. http://dx.doi.org/10.1016/j.biopsych.2006.06.004. Ritchie, S.J., Bates, T.C., Der, G., Starr, J.M., Deary, I.J., 2012. Education is associated with higher later life IQ scores, but not with faster cognitive processing speed. Psychol. Aging http://dx.doi.org/10.1037/a0030820. Rougemont-Bucking, A., Linnman, C., Zeffiro, T.A., Zeidan, M.A., Lebron-Milad, K., Rodriguez-Romaguera, J., Milad, M.R., 2011. Altered processing of contextual information during fear extinction in PTSD: an fMRI study. CNS Neurosci. Ther. 17 (4), 227–236. http://dx.doi.org/10.1111/j.1755-5949.2010.00152.x. Shin, L.M., Liberzon, I., 2010. The neurocircuitry of fear, stress, and anxiety disorders. Neuropsychopharmacology 35 (1), 169–191. http://dx.doi.org/10.1038/npp.2009.83 (npp200983 [pii]). Vervliet, B., Vansteenwegen, D., Eelen, P., 2004. Generalization of extinguished skin conductance responding in human fear conditioning. Learn. Mem. 11 (5), 555–558. http://dx.doi.org/10.1101/lm.77404. Zorawski, M., Cook, C.A., Kuhn, C.M., LaBar, K.S., 2005. Sex, stress, and fear: individual differences in conditioned learning. Cogn. Affect. Behav. Neurosci. 5 (2), 191–201.