- Email: [email protected]
Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates
INTPSY-11049; No of Pages 6 International Journal of Psychophysiology xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
International Journal of Psychophysiology journal homepage: www.elsevier.com/locate/ijpsycho
Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates Brian Chin, Brady D. Nelson, Felicia Jackson, Greg Hajcak ⁎ Stony Brook University, United States
a r t i c l e
i n f o
Article history: Received 10 March 2015 Received in revised form 11 November 2015 Accepted 15 November 2015 Available online xxxx Keywords: Intolerance of uncertainty Reinforcement Startle Threat
a b s t r a c t Fear conditioning research on threat predictability has primarily examined the impact of temporal (i.e., timing) predictability on the startle reflex. However, there are other key features of threat that can vary in predictability. For example, the reinforcement rate (i.e., frequency) of threat is a crucial factor underlying fear learning. The present study examined the impact of threat reinforcement rate on the startle reflex and self-reported anxiety during a fear conditioning paradigm. Forty-five participants completed a fear learning task in which the conditioned stimulus was reinforced with an electric shock to the forearm on 50% of trials in one block and 75% of trials in a second block, in counter-balanced order. The present study also examined whether intolerance of uncertainty (IU), the tendency to perceive or experience uncertainty as stressful or unpleasant, was associated with the startle reflex during conditions of low (50%) vs. high (75%) reinforcement. Results indicated that, across all participants, startle was greater during the 75% relative to the 50% reinforcement condition. IU was positively correlated with startle potentiation (i.e., increased startle response to the CS+ relative to the CS−) during the 50%, but not the 75%, reinforcement condition. Thus, despite receiving fewer electric shocks during the 50% reinforcement condition, individuals with high IU uniquely demonstrated greater defense system activation when impending threat was more uncertain. The association between IU and startle was independent of state anxiety. The present study adds to a growing literature on threat predictability and aversive responding, and suggests IU is associated with abnormal responding in the context of uncertain threat. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Fear conditioning is a form of associative learning that is critical to the development and maintenance of anxiety disorders (Lissek et al., 2005). Laboratory studies of fear conditioning often examine differential conditioning, during which multiple conditioned stimuli (CS) are presented, and one is paired with an aversive unconditioned stimulus (UCS) and signals danger (CS +) and the others are not paired and signal safety (CS −). In humans, fear conditioning is often examined using the startle eye blink reflex, which is a cross-species index of defense system activation (Lang et al., 1990). The startle reflex is modulated by emotional valence, such that it is potentiated (i.e., increased) by aversive emotional states and attenuated (i.e., decreased) by appetitive emotional states (Giargiari et al., 2005; Lang et al., 1998). Consistent with these data, fear conditioning studies reliably show that the startle reflex is increased in the presence of the CS+ relative to the CS− (Grillon et al., 1993). There are several features of threat that can impact defense system activation. The predictability of threat has been suggested to delineate ⁎ Corresponding author at: Psychology Department, Stony Brook University, Stony Brook, NY 11794-2500, United States. E-mail address: [email protected] (G. Hajcak).
the emotional response states of fear and anxiety (Barlow, 2000; Davis, 1992; Grillon et al., 2004). Fear is associated with predictable threat and elicits a fight, flight, or freeze response, whereas anxiety is associated with unpredictable threat and elicits defensive preparedness and hypervigilance. This differentiation has been supported by neuroanatomical (Davis, 1998), neuroimaging (Walker and Davis, 2008), pharmacological (Grillon et al., 2006), and psychophysiological studies (Grillon et al., 2004). The distinction between fear and anxiety also plays an important role in several theoretical perspectives of anxiety disorders. For example, an enhanced fear response in the presence of certain objects or situations is characteristic of phobic disorders, whereas chronic anxious apprehension about the future is typical of generalized anxiety disorder (GAD; Lang et al., 2000). The majority of research investigating threat predictability has examined the impact of temporal (i.e., timing) predictability on the startle reflex (e.g., Grillon et al., 2004). In these paradigms, the exact timing of aversive stimulus delivery is either known (predictable timing condition) or unknown (unpredictable timing condition). However, there are other key features of threat that can vary in predictability. For example, Shankman et al. (2011) found that unpredictable, relative to predictable, shock intensity potentiated the startle reflex. To date, most studies have compared predictable and unpredictable threat by using conditions that are matched on the frequency with which an
http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006 0167-8760/© 2015 Elsevier B.V. All rights reserved.
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006
2
B. Chin et al. / International Journal of Psychophysiology xxx (2015) xxx–xxx
aversive stimulus is delivered, but differ in some other dimension (e.g., timing, intensity; Grillon et al., 2004, 2008). Another experimental feature that can alter the predictability of an aversive stimulus is the reinforcement rate (i.e., frequency). Theoretical models of fear conditioning have implicated the reinforcement rate of threat as a crucial factor underlying fear learning (Gallistel and Gibbon, 2000). Indeed, threat responding is based on the possibility that the CS+ is related to impending presentation of the aversive UCS (Norrholm et al., 2006). Therefore, higher reinforcement rates make the UCS more predictable and consequently the CS+/UCS contingency easier to learn, while partial (or decreased) reinforcement of a CS+ increases the number of trials required for fear acquisition and learning (Gallistel and Gibbon, 2000). The present study examined the impact of 50% and 75% UCS reinforcement rates on the startle reflex and self-reported anxiety during a fear conditioning paradigm. Specifically, 45 participants completed a CS +/CS − threat-of-shock task that contained two within-subjects conditions in a counterbalanced order: 50% shock reinforcement and 75% shock reinforcement. The startle reflex was measured during the CS + and CS − in each condition, and self-reported anxiety during these conditions was collected retrospectively at the end of the task. Critically, in the 50% reinforcement condition, the UCS was less frequent and therefore more unpredictable; on the other hand, in the 75% condition the UCS was more frequent and predictable. Consistent with prior theoretical models (Gallistel and Gibbon, 2000), we hypothesized that fear-potentiated startle (FPS; i.e., the difference between the CS+ and CS −) and self-reported anxiety potentiation (i.e., the difference between the CS+ and CS−) would be larger in the 75% relative to the 50% reinforcement condition. We also examined the relationship between individual differences in particular anxiety phenotypes and threat responding. Specifically, we examined how variations in intolerance of uncertainty (IU) related to FPS and self-reported anxiety in both the 50% and 75% condition. IU reflects the tendency to find ambiguity and uncertainty aversive, stressful, and unpleasant (Dugas et al., 2004). IU has been associated with several anxiety disorders, including GAD (Dugas et al., 2004), obsessivecompulsive disorder (OCD; Tolin et al., 2003), panic disorder (Carleton et al., 2013, 2014), posttraumatic stress disorder (PTSD; Fetzner et al., 2013), and social anxiety disorder (SAD; Boelen and Reijntjes, 2009). This has led some researchers to conceptualize IU as a potential transdiagnostic factor of psychopathology (Boswell et al., 2013). In the present study, participants completed the Intolerance of Uncertainty Scale (IUS; Freeston et al., 1994), and we examined whether IU was associated with FPS and self-reported anxiety in the context of less (50%) versus more (75%) frequent and predictable reinforcement. Since the UCS was less predictable in the 50% condition, we hypothesized that IU would be associated with a heightened startle reflex and self-reported anxiety during the 50% condition. Finally, participants also completed the State Trait Anxiety Inventory (STAI; Spielberger et al., 1983), and we examined whether the association between IU and the startle reflex and self-reported anxiety were independent of general symptoms of anxiety. We hypothesized that the relationship between IU and threat responding would remain significant after controlling for anxiety.
2. Methods 2.1. Participants Forty-five introduction to psychology students participated for course credit. The sample included 32 females (71.1%) and the racial/ ethnic distribution was 33.3% Caucasian, 33.3% Asian, 11.1% African American, and 22.2% ‘Other’. Informed consent was obtained prior to participation and the research protocol was approved by the Stony Brook University Institutional Review Board.
2.2. Measures 2.2.1. The intolerance of uncertainty scale The IUS (Freeston et al., 1994) is a 27-item self-report questionnaire that assesses the degree to which individuals find ambiguous or uncertain situations to be stressful and unpleasant. Items are rated on a fivepoint Likert scale ranging from 1 (not at all characteristic of me) to 5 (entirely characteristic of me), with higher scores indicating greater IU. 2.2.2. The state trait anxiety inventory The STAI (Spielberger et al., 1983) is a 40-item self-report measure of anxiety and consists of two 20-item versions measuring state and trait anxiety. Items are rated on a four-point Likert scale ranging from 1 (not at all) to 4 (very much so), with higher scores indicating greater anxiety. In the present study, participants only completed the STAIState. 2.3. Stimuli The startle probe was a 50-ms, 105-dB burst of white noise with instantaneous rise and fall times, delivered binaurally through headphones. Electrical shocks served as the UCS and were delivered to the participant's left forearm. Shocks consisted of 60 Hz constant AC stimulation at an amplitude between 0 and 5 mA presented for 500 ms. In order to ensure that the shocks were significantly aversive to each participant, a workup procedure was used where increasingly stronger shocks were delivered until the participant described it as feeling “highly annoying but not painful.” The mean shock level of the final sample was 2.28 mA (SD = 0.43). 2.4. Procedure After obtaining informed consent, participants were seated in front of a 19-in computer monitor inside a sound-attenuated booth and electrodes were attached to measure the startle eye blink reflex. Participants first completed a block of startle habituation trials in which four acoustic probes were delivered in the presence of a fixation cross to elicit initial exaggerated startle responses. After the habituation trials, participants completed a shock workup procedure until a desired shock level was determined. Next, participants completed two blocks (50% vs. 75% reinforcement) of a fear learning task during which a CS + (geometric shape) was reinforced with an electric shock to the forearm on 50% or 75% of trials. The CS − (a different geometric shape of the same color) was never paired with an electric shock. The first block of trials included either a green triangle and green star or a blue circle and blue square as conditioned stimuli; the second block of trials included the other set of shapes as stimuli. The pairing between shape and color (e.g. green triangle/star or blue circle/square) were constant; however, CS assignment for each pairing (e.g., whether the green triangle or star was the CS +) was counterbalanced across participants, as was the pair used in each condition (e.g., whether the green triangle/star were the CS in the 50% or 75% conditions). The order of the reinforcement condition (i.e., 50% reinforcement condition first vs. 75% reinforcement condition first) was also counterbalanced across participants. Participants were instructed to passively view the shapes as they appeared, and were told that if they paid attention they could determine which shape was sometimes paired with the shock. Participants were not explicitly informed about the reinforcement schedule (i.e., 50% vs. 75%) in either condition. Each block consisted of 16 trials (8 CS + and 8 CS −) presented in a random order, and no trial type (CS + vs. CS −) was presented more than twice in a row. The CS + and CS − were presented for 6 s; trials were separated by an intertrial interval (ITI) that varied between 2.5 and 3.0 s, during which a fixation cross was presented. During the 50% condition, the CS+ was reinforced with an electric shock on 50% of trials (i.e., 4 of 8 trials). During the 75% condition,
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006
B. Chin et al. / International Journal of Psychophysiology xxx (2015) xxx–xxx
the CS+ was reinforced with an electric shock on 75% of trials (i.e., 6 of 8 trials). For both conditions (50% and 75%), the first CS+ trial was always reinforced with a shock. The time between the end of the first block and beginning of the second was approximately 45 s. Acoustic startle probes were presented 3.5–4.5 s following the appearance of the CS +/CS. On trials in which a startle probe was presented during the ITI, the ITI was increased by 9 s and the probe was presented between 2.5 and 4.5 s after the beginning of the ITI. The startle probe was presented on 6 of 8 CS+ trials, 6 of 8 CS− trials, and 4 of 16 ITIs. The electric shock was delivered at the offset of the CS+. Administration of the electric shock was always followed by 10 s of additional time to ensure that the shock did not interfere with the subsequent trial. Trials that occurred after administration of the electric shock were equally likely to be a CS+ or CS−. In total, each participant received 10 electric shocks (4 during the 50% and 6 during the 75% reinforcement condition) and 32 startle probes (6 during the 50% CS+, 6 during the 50% CS −, 6 during the 75% CS +, 6 during the 75% CS−, 4 during the 50% ITI, and 4 during the 75% ITI). The trials where participants heard a startle probe and/or received an electric shock were chosen at random (with the exception of the first CS + trial, which was always reinforced) to ensure a balanced presentation of stimuli across the entire sample. After completion of the task, participants were asked to rate how anxious/distressed they felt when they saw each of the four geometric shapes (i.e., 50% CS+, 50% CS−, 75% CS+, 75% CS −) on a four-point Likert scale ranging from 1 (not at all) to 4 (very). 2.5. Physiological recording and analysis Electromyography (EMG) was recorded and processed using PSYLAB (Contact Precision Instruments, London, UK). Two 4 mm Ag/AgCl electrodes were filled with electrode gel (TD-40; Mansfield R and D) and were positioned beneath the left eye over the orbicularis oculi muscle approximately 25 mm apart. A third electrode was placed on the forehead to serve as an isolated ground. EMG activity was sampled at 1000 Hz and filtered between 30 and 500 Hz. EMG responses were rectified in a 200 ms window, beginning 50 ms before the onset of the startle probe. A 6-point running average was applied to the rectified data to smooth out sharp peaks. Raw startle magnitude was baseline corrected and represented the difference between the average of the EMG in the 50 ms window prior to the startle probe and the maximum in the 150 ms post-probe window. Each participant's data was examined on a trial-by-trial basis. Trials with no perceptible eyeblink response were scored as zero and included in the overall averages (i.e., a blink response less than 10 μV); trials with excessive baseline artifacts or noise were excluded from analysis. Startle magnitude during the CS+, CS−, and ITI of the 50% and 75% reinforcement conditions was t-scored within-subjects (M = 50, SD = 10) to limit the influence of participants with large overall blink responses. 2.6. Data Analysis Ten participants were excluded from data analysis due to equipment malfunction (n = 6) or poor startle data quality (n = 4), leaving a final sample of 35 participants (n = 18 did 50% reinforcement condition). To examine threat responding during the different conditions and their association with IU, we conducted a Condition (50% reinforcement vs. 75% reinforcement) X Trial (CS+ vs. CS−) X IUS mixed-measures analysis of covariance (ANCOVA), with condition and cue as within-subjects factors and IUS as a mean-centered continuous covariate. Task order (50% reinforcement first vs. 75% reinforcement first) was also included as a dichotomous covariate. Furthermore, to examine the unique association between IU and threat responding independent of current anxiety, we conducted identical analyses but also included STAI-State as a meancentered continuous covariate. Separate analyses were conducted for self-report anxiety and the startle reflex. One participant did not
3
Fig. 1. Mean startle magnitude (top) and self-reported anxiety (bottom) for the CS+, CS−, and ITI by reinforcement condition (50% vs. 75%). Self-reported anxiety ranged from 1 (not at all) to 4 (very). Error bars represent standard error. au = arbitrary units; CS = conditioned stimulus; ITI = intertrial interval.
complete the self-reported anxiety measure, leaving a sample of 34 participants for those analyses. All analyses were conducted in IBM SPSS Statistics, Version 22.0 (Armonk, NY, USA). 3. Results 3.1. Self-report measures The IUS (M = 57.64, SD = 15.83, Cronbach's α = .93) and STAI-State (M = 47.54, SD = 2.92, α = .70) were moderately correlated, r(35) = .53, p b .01. However, neither measure was associated with shock level and did not differ between participants who completed the 50% vs. 75% reinforcement condition first (i.e., task order; ps N .38). IUS scores were comparable to those found in previous studies using undergraduates (M = 58.23, SD = 19.47; Nelson & Shankman, 2011), but lower than those reported in clinical populations (e.g., OCD checkers; M = 82.16, SD = 23.60; Tolin et al., 2003). 3.2. Startle Reflex For the startle reflex (Fig. 1, top; see Table 1 for raw startle magnitude values), results indicated a main effect of condition, F(1, 32) = 8.67, p b .01, η2p = .21, such that the startle reflex was greater during the 75% relative to the 50% reinforcement condition, and a main effect of trial, F(1, 32) = 6.30, p b .05, η2p = .16, such that the startle reflex was Table 1 Means (and standard deviations) for raw startle magnitude across the CS+, CS−, and ITI for the 50% and 75% reinforcement conditions.
CS+ CS− FPS ITI
50% reinforcement
75% reinforcement
62.73 (22.97) 56.55 (25.97) 6.17 (14.23) 55.43 (26.62)
64.79 (25.37) 59.77 (26.34) 5.01 (11.06) 55.82 (29.52)
Note. FPS was calculated by subtracting the CS− from the CS+ (i.e., CS+ minus CS−). All values are presented in microvolts. Standard deviations are presented in parentheses. CS = conditions stimulus; FPS = fear-potentiated startle; ITI = inter-trial interval.
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006
4
B. Chin et al. / International Journal of Psychophysiology xxx (2015) xxx–xxx
greater during the CS+ relative to the CS−.1 There was no Condition X Trial interaction, F(1, 32) b 0.01, ns, suggesting that FPS (i.e., CS + minus CS−) did not differ between the 50% and 75% conditions. There was also a Condition X Trial X IUS three-way interaction, F(1, 32) = 4.11, p = .05, η2p = .11. To follow-up the interaction, we first subtracted the CS− from the CS+ (i.e., FPS) separately for the 50% and 75% reinforcement conditions. This allowed for the examination of startle potentiation in both conditions. Next, we conducted separate partial correlations between the IUS and startle potentiation during the 50% and 75% conditions, controlling for task order. Results indicated that IU was positively correlated with startle potentiation during the 50%, pr(32) = .40, p b .05, but not the 75%, pr(32) = −.14, ns, reinforcement condition (Fig. 2). A comparison of the correlation coefficients using a Fisher r-to-z transformation indicated that this was a significant difference, z = 2.26, p b .05, two-tailed. In other words, greater IU was associated with increased FPS in anticipation of the less frequent but more unpredictable UCS. We also examined whether the association between the IUS and the startle reflex was independent of anxiety. To this end, we conducted an identical mixed-measure ANCOVA, but included STAI-State as an additional mean-centered covariate. Results again indicated a Condition X Trial X IUS three-way interaction, F(1, 31) = 5.06, p b .05, η2p = .14. Follow-up analyses indicated that IUS was positively correlated with startle potentiation during the 50%, pr(31) = .41, p b .05, but not the 75%, pr(31) = −.22, ns, reinforcement condition, and these correlations significantly differed, z = 2.64, p b .01, two-tailed. In contrast, there were no main effects or interactions involving STAI-State (ps N .26). Thus, the association between IU and FPS during the 50% condition was not better accounted for by broader anxiety. In the present study, participants completed a fear learning task and were not explicitly told whether they were about to complete a 50% vs. 75% reinforcement condition, which only differed by two reinforced CS+ trials (6 vs. 8, respectively). Thus, it is unlikely that IU was related to FPS until the latter half of the each condition when participants were finally able to discern this difference. To test this hypothesis, we conducted an identical Condition X Trial X IUS mixed-measures ANCOVA for the first half vs. second half of trials during each condition. For the first half of trials, there were no main effects or interactions involving IUS (ps N .41). In contrast, for the second half of trials there was a Condition X Trial X IUS interaction, F(1, 32) = 4.07, p = .05, ηp2 = .11. Follow-up partial correlation analyses indicated that IU was positively correlated with startle potentiation during the 50%, pr(32) = .44, p b .01, but not the 75%, pr(32) = −.11, ns, reinforcement condition. A Fisher r-to-z transformation again indicated this was a significant difference, z = 2.33, p b .05, two-tailed. These results support the hypothesis that IU was associated with increased FPS in anticipation of the less frequent but more unpredictable UCS during the latter half of trials.
1 There was no main effect of the covariate task order, F(1, 32) = 0.32, ns, but there was a Condition X Task Order interaction, F(1, 32) = 34.64, p b .001, η2p = .52. Results indicated that when the 75% reinforcement condition was presented first, the startle reflex was greater during the 75% (M = 53.57, SD = 2.13) relative to the 50% condition (M = 48.33, SD = 2.52), F(1, 16) = 28.80, p b .001, η2p = .64. In contrast, when the 50% reinforcement condition was presented first, the startle reflex was greater during the 50% (M = 52.24, SD = 2.23) relative to the 75% condition (M = 49.11, SD = 3.01), F(1, 17) = 9.80, p b .01, η2p = .37. Together, these results reflect an expected decrease in startle reflex across both conditions over time. We also compared the CS+ and CS− to the ITI. To this end, we conducted a Condition (50% reinforcement vs. 75% reinforcement) X Trial (CS+, CS−, ITI) repeated measures ANOVA, with task order (50% reinforcement first vs. 75% reinforcement first) included as a dichotomous covariate. Results indicated a main effect of condition, F(1, 33) = 22.41, p b .001, η2p = .40, such that the startle reflex was greater during the 75% relative to 50% reinforcement rate condition, and a main effect of trial, F(2, 66) = 5.63, p b .05, η2p = .15, such that the startle reflex was larger during the CS+ relative to the CS−, F(1, 33) = 7.38, p b .01, η2p = .18, and ITI, F(1, 33) = 8.33, p b .01, η2p = .20, but the startle reflex did not differ between the CS− and ITI, F(1, 33) = 1.99, ns. There was no Condition X Trial interaction, F(2, 66) = 2.53, ns.
Fig. 2. Scatterplot depicting the association between IUS and FPS (i.e., CS+ minus CS− difference score) during the 50% reinforcement condition. CS = conditioned stimulus; FPS = fear-potentiated startle; IUS = intolerance of uncertainty scale.
3.3. Anxiety For self-reported anxiety (Fig. 1, bottom), there was a main effect of trial, F(1, 31) = 29.00, p b .001, η2p = .48, such that participants reported greater anxiety during the CS+ relative to the CS−, and a main effect of IUS, F(1, 31) = 14.16, p b .001, η2p = .31, such that greater IU was associated with increased self-reported anxiety across all conditions, pr(32) = .56, p b .001 (see Fig. 3). There was also a Condition X Trial interaction, F(1, 31) = 4.67, p b .05, η2p = .13, such that participants reported a greater increase in self-reported anxiety to the CS+ relative to the CS− during the 75%, F(1, 31) = 30.56, p b .001, η2p = .50, relative to the 50%, F(1, 31) = 21.35, p b .001, η2p = .40, reinforcement condition. There were no interactions involving the IUS (ps N .54). Finally, we also examined the unique association between the IUS and self-reported anxiety during the fear learning task independent of broader state anxiety. To this end, we conducted an identical mixed-measure ANCOVA, but included STAI-State as an additional mean-centered covariate. Results again indicated a main effect of IUS, F(1, 30) = 14.68, p b .001, η2p = .33, such that greater IU was associated with increased self-reported anxiety across all threat conditions, pr(31) = .57, p b .001. There were no main effects or interactions for STAI-State (ps N .13).
4. Discussion The present study examined the impact of UCS reinforcement rate on the startle reflex and self-reported anxiety during a fear conditioning task. Across all participants, the startle reflex was greater during the 75% relative to the 50% reinforcement condition. This suggests that the reinforcement rate with which participants received electric shocks
Fig. 3. Scatterplot depicting the association between IUS and self-reported anxiety collapsing across the CS+ and CS− trial types and 50% and 75% reinforcement conditions. Self-reported anxiety ranged from 1 (not at all) to 4 (very). au = arbitrary units; CS = conditioned stimulus; IUS = intolerance of uncertainty scale.
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006
B. Chin et al. / International Journal of Psychophysiology xxx (2015) xxx–xxx
potentiated the defense startle reflex across all trial types. Participants also had greater self-reported anxiety to the CS+ relative to the CS− in the 75% relative to the 50% reinforcement condition. On the other hand, FPS (i.e., CS + minus CS −) did not differ between the 50% and 75% conditions. However, individual differences in IU were positively correlated with startle potentiation during the 50% (but not 75%) reinforcement condition. In other words, high IU individuals demonstrated increased defensive responding in the context of the less frequent (but more unpredictable) threat. IU was also associated with increased anxiety across both conditions. These associations were independent of state anxiety, which did not impact startle magnitude, FPS, or selfreported anxiety in either condition. This study supports the notion that the reinforcement rate of threat impacts defense system activation and this relationship is influenced by individual differences in IU. These findings add to a growing literature linking IU and psychobiological responding to uncertain threat. For example, greater IU has been associated with increased insula (Shankman et al., 2014; Simmons et al., 2008) and amygdala (Schienle et al., 2010) activation during the processing of uncertain threat. Furthermore, high IU was associated with an enhanced event-related potential P200 response during the anticipation of uncertain valenced pictures (Gole et al., 2012). The present study is one of the first to demonstrate an association between IU and the frequency of threat reinforcement. Despite receiving fewer electric shocks, high IU participants demonstrated greater startle potentiation during the 50% reinforcement condition. These results are consistent with previous work demonstrating that uncertainty can lead to suboptimal reactions that subsequently impact decision-making (Ladouceur et al., 1997; Maner et al., 2007). For example, Luhmann et al. (2011) found that during a delayed probabilistic reward task, high IU individuals were more likely to choose a smaller and less probable (but immediate) reward instead of a larger and more probable (but delayed) reward. Although we did not directly measure decision-making, the current study adds to a growing literature indicating that, in individuals with high IU, uncertainty can contribute to irrational responding. IU has been linked to a number of different psychopathological conditions, including GAD (Dugas et al., 2004), OCD (Tolin et al., 2003), panic disorder (Carleton et al., 2013), social anxiety (Boelen and Reijntjes, 2009), and depression (McEvoy and Mahoney, 2012), and evidence suggests that IU may be broadly associated with increased negative affect and anxiety symptomatology. However, there are several potential methodological limitations to self-report measures of emotional reactivity that may impact their association with IU, including demand characteristics, restricted range of responses and ceiling effects, and reliance on retrospective recall. Importantly, the present study highlights the potential utility of psychophysiological measures (e.g., startle reflex) that are more sensitive to subtle differences in threat responding and may demonstrate better specificity in relation to anxiety phenotypes (e.g., IU). The startle reflex across the CS+ and CS− was greater during the 75% relative to 50% reinforcement condition. This result may have been due to greater shock sensitization during the 75% reinforcement condition (Davis, 1989). Furthermore, as expected self-reported anxiety potentiation (i.e., difference between the CS+ and CS −) was greater during the 75% relative to 50% reinforcement condition. However, in contrast to our hypothesis FPS did not differ across 50% and 75% shock reinforcement. It is important to note that the 50% and 75% reinforcement conditions only differed by two reinforced trials, and it is possible that a greater difference between reinforcement rates (e.g., 25% vs. 75%) may impact psychophysiological indices of fear learning. It is also possible that lengthening the acquisition phase (i.e., initial trials) and better separating the acquisition and ‘test’ phase (i.e., later trials after contingency has been learned) may have influenced how the reinforcement rate impacted FPS. The present study had several limitations that warrant consideration. First, the experience of two consecutive varying reinforcement
5
rates may have altered fear acquisition and attentional processes, and future studies should consider including a comparison group with consistent reinforcement rates. Second, the present study only measured individual differences in broad state anxiety, and it is unclear whether IU was associated with startle potentiation during the 50% condition independent of trait anxiety. However, we speculate this effect would be independent of trait anxiety given that the STAI-State and Trait measures have been shown to be strongly correlated (Grös et al., 2007). Finally, participants did not provide expectancy ratings for each reinforcement condition, which limited our ability to examine potential differences in contingency awareness. In conclusion, the present study found that the reinforcement rate of threat impacted emotional and motivational responding during a fear conditioning task and this was influenced by individual differences in IU. Specifically, across all participants greater UCS frequency was associated with an increased startle reflex. However, differences in FPS varied as a function of IU, such that IU was associated with enhanced startle potentiation during the more uncertain (but less frequent) reinforcement condition. These findings add to a growing literature demonstrating the relationship between IU and psychophysiological measures of sensitivity to uncertain threat. Future research should examine the impact of different reinforcement rates on other aspects of fear learning, including extinction and reinstatement. Furthermore, future studies might leverage this paradigm in clinically anxious individuals who are high on IU, and/or determine whether treating IU reduces sensitivity to unpredictable threat. References Barlow, D.H., 2000. Unraveling the mysteries of anxiety and its disorders from the perspective of emotion theory. Am. Psychol. 55, 1247. Boelen, P.A., Reijntjes, A., 2009. Intolerance of uncertainty and social anxiety. J. Anxiety Disord. 23, 130–135. Boswell, J.F., Thompson-Hollands, J., Farchione, T.J., Barlow, D.H., 2013. Intolerance of uncertainty: a common factor in the treatment of emotional disorders. J. Clin. Psychol. 69, 630–645. Carleton, R.N., Fetzner, M.G., Hackl, J.L., McEvoy, P., 2013. Intolerance of uncertainty as a contributor to fear and avoidance symptoms of panic attacks. Cogn. Behav. Ther. 42, 328–341. Carleton, R.N., Duranceau, S., Freeston, M.H., Boelen, P.A., McCabe, R.E., Antony, M.M., 2014. “But it might be a heart attack”: intolerance of uncertainty and panic disorder symptoms. J. Anxiety Disord. 28, 463–470. Davis, M., 1989. Sensitization of the acoustic startle reflex by footshock. Behav. Neurosci. 103, 495–503. Davis, M., 1992. The role of the amygdala in fear and anxiety. Annu. Rev. Neurosci. 15, 353–375. Davis, M., 1998. Are different parts of the extended amygdala involved in fear versus anxiety? Biol. Psychiatry 44, 1239–1247. Dugas, M.J., Buhr, K., Ladouceur, R., 2004. The role of intolerance of uncertainty in etiology and maintenance. In: Heimberg, R.G., Turk, C.L., Mennin, D.S. (Eds.), Generalized Anxiety Disorder: Advances in Research and Practice. Guilford Press, New York, NY, pp. 143–163. Fetzner, M.G., Horswill, S.C., Boelen, P.A., Carleton, R.N., 2013. Intolerance of uncertainty and PTSD symptoms: exploring the construct relationship in a community sample with a heterogeneous trauma history. Cogn. Ther. Res. 37, 725–734. Freeston, M.H., Rhéaume, J., Letarte, H., Dugas, M.J., Ladouceur, R., 1994. Why do people worry? Personal. Individ. Differ. 17, 791–802. Gallistel, C.R., Gibbon, J., 2000. Time, rate, and conditioning. Psychol. Rev. 107, 289–344. Giargiari, T.D., Mahaffey, A.L., Craighead, W.E., Hutchison, K.E., 2005. Appetitive responses to sexual stimuli are attenuated in individuals with low levels of sexual desire. Arch. Sex. Behav. 34, 547–556. Gole, M., Schäfer, A., Schienle, A., 2012. Event-related potentials during exposure to aversion and its anticipation: the moderating effect of intolerance of uncertainty. Neurosci. Lett. 24, 112–117. Grillon, C., Ameli, R., Foot, M., Davis, M., 1993. Fear-potentiated startle: Relationship to the level of state/trait anxiety in healthy subjects, Bio. Psychiatry 15, 566–574. Grillon, C., Baas, J.P., Lissek, S., Smith, K., Milstein, J., 2004. Anxious responses to predictable and unpredictable aversive events. Behav. Neurosci. 118, 916. Grillon, C., Baas, J.M., Pine, D.S., Lissek, S., Lawley, M., Ellis, V., Levine, J., 2006. The benzodiazepine alprazolam dissociates contextual fear from cued fear in humans as assessed by fear-potentiated startle. Biol. Psychiatry 60, 760–766. Grillon, C., Lissek, S., Rabin, S., McDowell, D., Dvir, S., Pine, D.S., 2008. Increased anxiety during anticipation of unpredictable but not predictable aversive stimuli as a psychophysiologic marker of panic disorder. Am. J. Psychiatr. 165, 898–904. Grös, D.F., Antony, M.M., Simms, L.J., McCabe, R.E., 2007. Psychometric properties of the state-trait inventory for cognitive and somatic anxiety (STICSA): comparison to the state-trait anxiety inventory (STAI). Psychol. Assess. 19, 369–381.
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006
6
B. Chin et al. / International Journal of Psychophysiology xxx (2015) xxx–xxx
Ladouceur, R., Talbot, F., Dugas, M.J., 1997. Behavioral expressions of intolerance of uncertainty in worry experimental findings. Behav. Modif. 21, 355–371. Lang, P.J., Bradley, M.M., Cuthbert, B.C., 1990. Emotion, attention, and the startle reflex. Psychol. Rev. 97, 377–395. Lang, P.J., Bradley, M.M., Cuthbert, B.N., 1998. Emotion, motivation, and anxiety: brain mechanisms and psychophysiology. Biol. Psychiatry 44, 1248–1263. Lang, P.J., Davis, M., Ohman, A., 2000. Fear and anxiety: animal models and human cognitive psychophysiology. J. Affect. Disord. 61, 137–159. Lissek, S., Powers, A.S., McClure, E.B., Phelps, E.A., Woldehawariat, G., Grillon, C., Pine, D.S., 2005. Classical fear conditioning in the anxiety disorders: a meta-analysis. Behav. Res. Ther. 43, 1391–1424. Luhmann, C.C., Ishida, K., Hajcak, G., 2011. Intolerance of uncertainty and decisions about delayed, probabilistic rewards. Behav. Ther. 42, 378–386. Maner, J.K., Richey, J.A., Cromer, K., Mallott, M., Lejuez, C.W., Joiner, T.E., Schmidt, N.B., 2007. Dispositional anxiety and risk-avoidant decision-making. Personal. Individ. Differ. 42, 665–675. McEvoy, P.M., Mahoney, A.E., 2012. To be sure, to be sure: intolerance of uncertainty mediates symptoms of various anxiety disorders and depression. Behav. Ther. 43, 533–545. Nelson, B.D., Shankman, S.A., 2011. Does intolerance of uncertainty predict anticipatory startle responses to uncertain threat? Int. J. Psychophysiol. 81, 107–115.
Norrholm, S.D., Jovanovic, T., Vervliet, B., Myers, K.M., Davis, M., Rothbaum, B.O., Duncan, E.J., 2006. Conditioned fear extinction and reinstatement in a human fear-potentiated startle paradigm. Learn. Mem. 13, 681–685. Schienle, A., Köchel, A., Ebner, F., Reishofer, G., Schäfer, A., 2010. Neural correlates of intolerance of uncertainty. Neurosci. Lett. 479, 272–276. Shankman, S.A., Robison-Andrew, E.J., Nelson, B.D., Altman, S.E., Campbell, M.L., 2011. Effects of predictability of shock timing and intensity on aversive responses. Int. J. Psychophysiol. 80, 112–118. Shankman, S.A., Gorka, S.M., Nelson, B.D., Fitzgerald, D.A., Phan, K.L., O'Daly, O., 2014. Anterior insula responds to temporally unpredictable aversiveness: an fMRI study. Neuroreport 25, 596–600. Simmons, A., Matthews, S.C., Paulus, M.P., Stein, M.B., 2008. Intolerance of uncertainty correlates with insula activation during affective ambiguity. Neurosci. Lett. 430, 92–97. Spielberger, C.D., Gorsuch, R.L., Lushene, R., Vagg, P.R., Jacobs, G.A., 1983. Manual for the State-Trait Anxiety Inventory. Consulting Psychologists Press, Palo Alto, CA. Tolin, D.F., Abramowitz, J.S., Brigidi, B.D., Foa, E.B., 2003. Intolerance of uncertainty in obsessive-compulsive disorder. J. Anxiety Disord. 17, 233–242. Walker, D.L., Davis, M., 2008. Role of the extended amygdala in short-duration versus sustained fear: a tribute to Dr. Lennart Heimer. Brain Struct. Funct. 213, 29–42.
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006
Contents lists available at ScienceDirect
International Journal of Psychophysiology journal homepage: www.elsevier.com/locate/ijpsycho
Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates Brian Chin, Brady D. Nelson, Felicia Jackson, Greg Hajcak ⁎ Stony Brook University, United States
a r t i c l e
i n f o
Article history: Received 10 March 2015 Received in revised form 11 November 2015 Accepted 15 November 2015 Available online xxxx Keywords: Intolerance of uncertainty Reinforcement Startle Threat
a b s t r a c t Fear conditioning research on threat predictability has primarily examined the impact of temporal (i.e., timing) predictability on the startle reflex. However, there are other key features of threat that can vary in predictability. For example, the reinforcement rate (i.e., frequency) of threat is a crucial factor underlying fear learning. The present study examined the impact of threat reinforcement rate on the startle reflex and self-reported anxiety during a fear conditioning paradigm. Forty-five participants completed a fear learning task in which the conditioned stimulus was reinforced with an electric shock to the forearm on 50% of trials in one block and 75% of trials in a second block, in counter-balanced order. The present study also examined whether intolerance of uncertainty (IU), the tendency to perceive or experience uncertainty as stressful or unpleasant, was associated with the startle reflex during conditions of low (50%) vs. high (75%) reinforcement. Results indicated that, across all participants, startle was greater during the 75% relative to the 50% reinforcement condition. IU was positively correlated with startle potentiation (i.e., increased startle response to the CS+ relative to the CS−) during the 50%, but not the 75%, reinforcement condition. Thus, despite receiving fewer electric shocks during the 50% reinforcement condition, individuals with high IU uniquely demonstrated greater defense system activation when impending threat was more uncertain. The association between IU and startle was independent of state anxiety. The present study adds to a growing literature on threat predictability and aversive responding, and suggests IU is associated with abnormal responding in the context of uncertain threat. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Fear conditioning is a form of associative learning that is critical to the development and maintenance of anxiety disorders (Lissek et al., 2005). Laboratory studies of fear conditioning often examine differential conditioning, during which multiple conditioned stimuli (CS) are presented, and one is paired with an aversive unconditioned stimulus (UCS) and signals danger (CS +) and the others are not paired and signal safety (CS −). In humans, fear conditioning is often examined using the startle eye blink reflex, which is a cross-species index of defense system activation (Lang et al., 1990). The startle reflex is modulated by emotional valence, such that it is potentiated (i.e., increased) by aversive emotional states and attenuated (i.e., decreased) by appetitive emotional states (Giargiari et al., 2005; Lang et al., 1998). Consistent with these data, fear conditioning studies reliably show that the startle reflex is increased in the presence of the CS+ relative to the CS− (Grillon et al., 1993). There are several features of threat that can impact defense system activation. The predictability of threat has been suggested to delineate ⁎ Corresponding author at: Psychology Department, Stony Brook University, Stony Brook, NY 11794-2500, United States. E-mail address: [email protected] (G. Hajcak).
the emotional response states of fear and anxiety (Barlow, 2000; Davis, 1992; Grillon et al., 2004). Fear is associated with predictable threat and elicits a fight, flight, or freeze response, whereas anxiety is associated with unpredictable threat and elicits defensive preparedness and hypervigilance. This differentiation has been supported by neuroanatomical (Davis, 1998), neuroimaging (Walker and Davis, 2008), pharmacological (Grillon et al., 2006), and psychophysiological studies (Grillon et al., 2004). The distinction between fear and anxiety also plays an important role in several theoretical perspectives of anxiety disorders. For example, an enhanced fear response in the presence of certain objects or situations is characteristic of phobic disorders, whereas chronic anxious apprehension about the future is typical of generalized anxiety disorder (GAD; Lang et al., 2000). The majority of research investigating threat predictability has examined the impact of temporal (i.e., timing) predictability on the startle reflex (e.g., Grillon et al., 2004). In these paradigms, the exact timing of aversive stimulus delivery is either known (predictable timing condition) or unknown (unpredictable timing condition). However, there are other key features of threat that can vary in predictability. For example, Shankman et al. (2011) found that unpredictable, relative to predictable, shock intensity potentiated the startle reflex. To date, most studies have compared predictable and unpredictable threat by using conditions that are matched on the frequency with which an
http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006 0167-8760/© 2015 Elsevier B.V. All rights reserved.
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006
2
B. Chin et al. / International Journal of Psychophysiology xxx (2015) xxx–xxx
aversive stimulus is delivered, but differ in some other dimension (e.g., timing, intensity; Grillon et al., 2004, 2008). Another experimental feature that can alter the predictability of an aversive stimulus is the reinforcement rate (i.e., frequency). Theoretical models of fear conditioning have implicated the reinforcement rate of threat as a crucial factor underlying fear learning (Gallistel and Gibbon, 2000). Indeed, threat responding is based on the possibility that the CS+ is related to impending presentation of the aversive UCS (Norrholm et al., 2006). Therefore, higher reinforcement rates make the UCS more predictable and consequently the CS+/UCS contingency easier to learn, while partial (or decreased) reinforcement of a CS+ increases the number of trials required for fear acquisition and learning (Gallistel and Gibbon, 2000). The present study examined the impact of 50% and 75% UCS reinforcement rates on the startle reflex and self-reported anxiety during a fear conditioning paradigm. Specifically, 45 participants completed a CS +/CS − threat-of-shock task that contained two within-subjects conditions in a counterbalanced order: 50% shock reinforcement and 75% shock reinforcement. The startle reflex was measured during the CS + and CS − in each condition, and self-reported anxiety during these conditions was collected retrospectively at the end of the task. Critically, in the 50% reinforcement condition, the UCS was less frequent and therefore more unpredictable; on the other hand, in the 75% condition the UCS was more frequent and predictable. Consistent with prior theoretical models (Gallistel and Gibbon, 2000), we hypothesized that fear-potentiated startle (FPS; i.e., the difference between the CS+ and CS −) and self-reported anxiety potentiation (i.e., the difference between the CS+ and CS−) would be larger in the 75% relative to the 50% reinforcement condition. We also examined the relationship between individual differences in particular anxiety phenotypes and threat responding. Specifically, we examined how variations in intolerance of uncertainty (IU) related to FPS and self-reported anxiety in both the 50% and 75% condition. IU reflects the tendency to find ambiguity and uncertainty aversive, stressful, and unpleasant (Dugas et al., 2004). IU has been associated with several anxiety disorders, including GAD (Dugas et al., 2004), obsessivecompulsive disorder (OCD; Tolin et al., 2003), panic disorder (Carleton et al., 2013, 2014), posttraumatic stress disorder (PTSD; Fetzner et al., 2013), and social anxiety disorder (SAD; Boelen and Reijntjes, 2009). This has led some researchers to conceptualize IU as a potential transdiagnostic factor of psychopathology (Boswell et al., 2013). In the present study, participants completed the Intolerance of Uncertainty Scale (IUS; Freeston et al., 1994), and we examined whether IU was associated with FPS and self-reported anxiety in the context of less (50%) versus more (75%) frequent and predictable reinforcement. Since the UCS was less predictable in the 50% condition, we hypothesized that IU would be associated with a heightened startle reflex and self-reported anxiety during the 50% condition. Finally, participants also completed the State Trait Anxiety Inventory (STAI; Spielberger et al., 1983), and we examined whether the association between IU and the startle reflex and self-reported anxiety were independent of general symptoms of anxiety. We hypothesized that the relationship between IU and threat responding would remain significant after controlling for anxiety.
2. Methods 2.1. Participants Forty-five introduction to psychology students participated for course credit. The sample included 32 females (71.1%) and the racial/ ethnic distribution was 33.3% Caucasian, 33.3% Asian, 11.1% African American, and 22.2% ‘Other’. Informed consent was obtained prior to participation and the research protocol was approved by the Stony Brook University Institutional Review Board.
2.2. Measures 2.2.1. The intolerance of uncertainty scale The IUS (Freeston et al., 1994) is a 27-item self-report questionnaire that assesses the degree to which individuals find ambiguous or uncertain situations to be stressful and unpleasant. Items are rated on a fivepoint Likert scale ranging from 1 (not at all characteristic of me) to 5 (entirely characteristic of me), with higher scores indicating greater IU. 2.2.2. The state trait anxiety inventory The STAI (Spielberger et al., 1983) is a 40-item self-report measure of anxiety and consists of two 20-item versions measuring state and trait anxiety. Items are rated on a four-point Likert scale ranging from 1 (not at all) to 4 (very much so), with higher scores indicating greater anxiety. In the present study, participants only completed the STAIState. 2.3. Stimuli The startle probe was a 50-ms, 105-dB burst of white noise with instantaneous rise and fall times, delivered binaurally through headphones. Electrical shocks served as the UCS and were delivered to the participant's left forearm. Shocks consisted of 60 Hz constant AC stimulation at an amplitude between 0 and 5 mA presented for 500 ms. In order to ensure that the shocks were significantly aversive to each participant, a workup procedure was used where increasingly stronger shocks were delivered until the participant described it as feeling “highly annoying but not painful.” The mean shock level of the final sample was 2.28 mA (SD = 0.43). 2.4. Procedure After obtaining informed consent, participants were seated in front of a 19-in computer monitor inside a sound-attenuated booth and electrodes were attached to measure the startle eye blink reflex. Participants first completed a block of startle habituation trials in which four acoustic probes were delivered in the presence of a fixation cross to elicit initial exaggerated startle responses. After the habituation trials, participants completed a shock workup procedure until a desired shock level was determined. Next, participants completed two blocks (50% vs. 75% reinforcement) of a fear learning task during which a CS + (geometric shape) was reinforced with an electric shock to the forearm on 50% or 75% of trials. The CS − (a different geometric shape of the same color) was never paired with an electric shock. The first block of trials included either a green triangle and green star or a blue circle and blue square as conditioned stimuli; the second block of trials included the other set of shapes as stimuli. The pairing between shape and color (e.g. green triangle/star or blue circle/square) were constant; however, CS assignment for each pairing (e.g., whether the green triangle or star was the CS +) was counterbalanced across participants, as was the pair used in each condition (e.g., whether the green triangle/star were the CS in the 50% or 75% conditions). The order of the reinforcement condition (i.e., 50% reinforcement condition first vs. 75% reinforcement condition first) was also counterbalanced across participants. Participants were instructed to passively view the shapes as they appeared, and were told that if they paid attention they could determine which shape was sometimes paired with the shock. Participants were not explicitly informed about the reinforcement schedule (i.e., 50% vs. 75%) in either condition. Each block consisted of 16 trials (8 CS + and 8 CS −) presented in a random order, and no trial type (CS + vs. CS −) was presented more than twice in a row. The CS + and CS − were presented for 6 s; trials were separated by an intertrial interval (ITI) that varied between 2.5 and 3.0 s, during which a fixation cross was presented. During the 50% condition, the CS+ was reinforced with an electric shock on 50% of trials (i.e., 4 of 8 trials). During the 75% condition,
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006
B. Chin et al. / International Journal of Psychophysiology xxx (2015) xxx–xxx
the CS+ was reinforced with an electric shock on 75% of trials (i.e., 6 of 8 trials). For both conditions (50% and 75%), the first CS+ trial was always reinforced with a shock. The time between the end of the first block and beginning of the second was approximately 45 s. Acoustic startle probes were presented 3.5–4.5 s following the appearance of the CS +/CS. On trials in which a startle probe was presented during the ITI, the ITI was increased by 9 s and the probe was presented between 2.5 and 4.5 s after the beginning of the ITI. The startle probe was presented on 6 of 8 CS+ trials, 6 of 8 CS− trials, and 4 of 16 ITIs. The electric shock was delivered at the offset of the CS+. Administration of the electric shock was always followed by 10 s of additional time to ensure that the shock did not interfere with the subsequent trial. Trials that occurred after administration of the electric shock were equally likely to be a CS+ or CS−. In total, each participant received 10 electric shocks (4 during the 50% and 6 during the 75% reinforcement condition) and 32 startle probes (6 during the 50% CS+, 6 during the 50% CS −, 6 during the 75% CS +, 6 during the 75% CS−, 4 during the 50% ITI, and 4 during the 75% ITI). The trials where participants heard a startle probe and/or received an electric shock were chosen at random (with the exception of the first CS + trial, which was always reinforced) to ensure a balanced presentation of stimuli across the entire sample. After completion of the task, participants were asked to rate how anxious/distressed they felt when they saw each of the four geometric shapes (i.e., 50% CS+, 50% CS−, 75% CS+, 75% CS −) on a four-point Likert scale ranging from 1 (not at all) to 4 (very). 2.5. Physiological recording and analysis Electromyography (EMG) was recorded and processed using PSYLAB (Contact Precision Instruments, London, UK). Two 4 mm Ag/AgCl electrodes were filled with electrode gel (TD-40; Mansfield R and D) and were positioned beneath the left eye over the orbicularis oculi muscle approximately 25 mm apart. A third electrode was placed on the forehead to serve as an isolated ground. EMG activity was sampled at 1000 Hz and filtered between 30 and 500 Hz. EMG responses were rectified in a 200 ms window, beginning 50 ms before the onset of the startle probe. A 6-point running average was applied to the rectified data to smooth out sharp peaks. Raw startle magnitude was baseline corrected and represented the difference between the average of the EMG in the 50 ms window prior to the startle probe and the maximum in the 150 ms post-probe window. Each participant's data was examined on a trial-by-trial basis. Trials with no perceptible eyeblink response were scored as zero and included in the overall averages (i.e., a blink response less than 10 μV); trials with excessive baseline artifacts or noise were excluded from analysis. Startle magnitude during the CS+, CS−, and ITI of the 50% and 75% reinforcement conditions was t-scored within-subjects (M = 50, SD = 10) to limit the influence of participants with large overall blink responses. 2.6. Data Analysis Ten participants were excluded from data analysis due to equipment malfunction (n = 6) or poor startle data quality (n = 4), leaving a final sample of 35 participants (n = 18 did 50% reinforcement condition). To examine threat responding during the different conditions and their association with IU, we conducted a Condition (50% reinforcement vs. 75% reinforcement) X Trial (CS+ vs. CS−) X IUS mixed-measures analysis of covariance (ANCOVA), with condition and cue as within-subjects factors and IUS as a mean-centered continuous covariate. Task order (50% reinforcement first vs. 75% reinforcement first) was also included as a dichotomous covariate. Furthermore, to examine the unique association between IU and threat responding independent of current anxiety, we conducted identical analyses but also included STAI-State as a meancentered continuous covariate. Separate analyses were conducted for self-report anxiety and the startle reflex. One participant did not
3
Fig. 1. Mean startle magnitude (top) and self-reported anxiety (bottom) for the CS+, CS−, and ITI by reinforcement condition (50% vs. 75%). Self-reported anxiety ranged from 1 (not at all) to 4 (very). Error bars represent standard error. au = arbitrary units; CS = conditioned stimulus; ITI = intertrial interval.
complete the self-reported anxiety measure, leaving a sample of 34 participants for those analyses. All analyses were conducted in IBM SPSS Statistics, Version 22.0 (Armonk, NY, USA). 3. Results 3.1. Self-report measures The IUS (M = 57.64, SD = 15.83, Cronbach's α = .93) and STAI-State (M = 47.54, SD = 2.92, α = .70) were moderately correlated, r(35) = .53, p b .01. However, neither measure was associated with shock level and did not differ between participants who completed the 50% vs. 75% reinforcement condition first (i.e., task order; ps N .38). IUS scores were comparable to those found in previous studies using undergraduates (M = 58.23, SD = 19.47; Nelson & Shankman, 2011), but lower than those reported in clinical populations (e.g., OCD checkers; M = 82.16, SD = 23.60; Tolin et al., 2003). 3.2. Startle Reflex For the startle reflex (Fig. 1, top; see Table 1 for raw startle magnitude values), results indicated a main effect of condition, F(1, 32) = 8.67, p b .01, η2p = .21, such that the startle reflex was greater during the 75% relative to the 50% reinforcement condition, and a main effect of trial, F(1, 32) = 6.30, p b .05, η2p = .16, such that the startle reflex was Table 1 Means (and standard deviations) for raw startle magnitude across the CS+, CS−, and ITI for the 50% and 75% reinforcement conditions.
CS+ CS− FPS ITI
50% reinforcement
75% reinforcement
62.73 (22.97) 56.55 (25.97) 6.17 (14.23) 55.43 (26.62)
64.79 (25.37) 59.77 (26.34) 5.01 (11.06) 55.82 (29.52)
Note. FPS was calculated by subtracting the CS− from the CS+ (i.e., CS+ minus CS−). All values are presented in microvolts. Standard deviations are presented in parentheses. CS = conditions stimulus; FPS = fear-potentiated startle; ITI = inter-trial interval.
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006
4
B. Chin et al. / International Journal of Psychophysiology xxx (2015) xxx–xxx
greater during the CS+ relative to the CS−.1 There was no Condition X Trial interaction, F(1, 32) b 0.01, ns, suggesting that FPS (i.e., CS + minus CS−) did not differ between the 50% and 75% conditions. There was also a Condition X Trial X IUS three-way interaction, F(1, 32) = 4.11, p = .05, η2p = .11. To follow-up the interaction, we first subtracted the CS− from the CS+ (i.e., FPS) separately for the 50% and 75% reinforcement conditions. This allowed for the examination of startle potentiation in both conditions. Next, we conducted separate partial correlations between the IUS and startle potentiation during the 50% and 75% conditions, controlling for task order. Results indicated that IU was positively correlated with startle potentiation during the 50%, pr(32) = .40, p b .05, but not the 75%, pr(32) = −.14, ns, reinforcement condition (Fig. 2). A comparison of the correlation coefficients using a Fisher r-to-z transformation indicated that this was a significant difference, z = 2.26, p b .05, two-tailed. In other words, greater IU was associated with increased FPS in anticipation of the less frequent but more unpredictable UCS. We also examined whether the association between the IUS and the startle reflex was independent of anxiety. To this end, we conducted an identical mixed-measure ANCOVA, but included STAI-State as an additional mean-centered covariate. Results again indicated a Condition X Trial X IUS three-way interaction, F(1, 31) = 5.06, p b .05, η2p = .14. Follow-up analyses indicated that IUS was positively correlated with startle potentiation during the 50%, pr(31) = .41, p b .05, but not the 75%, pr(31) = −.22, ns, reinforcement condition, and these correlations significantly differed, z = 2.64, p b .01, two-tailed. In contrast, there were no main effects or interactions involving STAI-State (ps N .26). Thus, the association between IU and FPS during the 50% condition was not better accounted for by broader anxiety. In the present study, participants completed a fear learning task and were not explicitly told whether they were about to complete a 50% vs. 75% reinforcement condition, which only differed by two reinforced CS+ trials (6 vs. 8, respectively). Thus, it is unlikely that IU was related to FPS until the latter half of the each condition when participants were finally able to discern this difference. To test this hypothesis, we conducted an identical Condition X Trial X IUS mixed-measures ANCOVA for the first half vs. second half of trials during each condition. For the first half of trials, there were no main effects or interactions involving IUS (ps N .41). In contrast, for the second half of trials there was a Condition X Trial X IUS interaction, F(1, 32) = 4.07, p = .05, ηp2 = .11. Follow-up partial correlation analyses indicated that IU was positively correlated with startle potentiation during the 50%, pr(32) = .44, p b .01, but not the 75%, pr(32) = −.11, ns, reinforcement condition. A Fisher r-to-z transformation again indicated this was a significant difference, z = 2.33, p b .05, two-tailed. These results support the hypothesis that IU was associated with increased FPS in anticipation of the less frequent but more unpredictable UCS during the latter half of trials.
1 There was no main effect of the covariate task order, F(1, 32) = 0.32, ns, but there was a Condition X Task Order interaction, F(1, 32) = 34.64, p b .001, η2p = .52. Results indicated that when the 75% reinforcement condition was presented first, the startle reflex was greater during the 75% (M = 53.57, SD = 2.13) relative to the 50% condition (M = 48.33, SD = 2.52), F(1, 16) = 28.80, p b .001, η2p = .64. In contrast, when the 50% reinforcement condition was presented first, the startle reflex was greater during the 50% (M = 52.24, SD = 2.23) relative to the 75% condition (M = 49.11, SD = 3.01), F(1, 17) = 9.80, p b .01, η2p = .37. Together, these results reflect an expected decrease in startle reflex across both conditions over time. We also compared the CS+ and CS− to the ITI. To this end, we conducted a Condition (50% reinforcement vs. 75% reinforcement) X Trial (CS+, CS−, ITI) repeated measures ANOVA, with task order (50% reinforcement first vs. 75% reinforcement first) included as a dichotomous covariate. Results indicated a main effect of condition, F(1, 33) = 22.41, p b .001, η2p = .40, such that the startle reflex was greater during the 75% relative to 50% reinforcement rate condition, and a main effect of trial, F(2, 66) = 5.63, p b .05, η2p = .15, such that the startle reflex was larger during the CS+ relative to the CS−, F(1, 33) = 7.38, p b .01, η2p = .18, and ITI, F(1, 33) = 8.33, p b .01, η2p = .20, but the startle reflex did not differ between the CS− and ITI, F(1, 33) = 1.99, ns. There was no Condition X Trial interaction, F(2, 66) = 2.53, ns.
Fig. 2. Scatterplot depicting the association between IUS and FPS (i.e., CS+ minus CS− difference score) during the 50% reinforcement condition. CS = conditioned stimulus; FPS = fear-potentiated startle; IUS = intolerance of uncertainty scale.
3.3. Anxiety For self-reported anxiety (Fig. 1, bottom), there was a main effect of trial, F(1, 31) = 29.00, p b .001, η2p = .48, such that participants reported greater anxiety during the CS+ relative to the CS−, and a main effect of IUS, F(1, 31) = 14.16, p b .001, η2p = .31, such that greater IU was associated with increased self-reported anxiety across all conditions, pr(32) = .56, p b .001 (see Fig. 3). There was also a Condition X Trial interaction, F(1, 31) = 4.67, p b .05, η2p = .13, such that participants reported a greater increase in self-reported anxiety to the CS+ relative to the CS− during the 75%, F(1, 31) = 30.56, p b .001, η2p = .50, relative to the 50%, F(1, 31) = 21.35, p b .001, η2p = .40, reinforcement condition. There were no interactions involving the IUS (ps N .54). Finally, we also examined the unique association between the IUS and self-reported anxiety during the fear learning task independent of broader state anxiety. To this end, we conducted an identical mixed-measure ANCOVA, but included STAI-State as an additional mean-centered covariate. Results again indicated a main effect of IUS, F(1, 30) = 14.68, p b .001, η2p = .33, such that greater IU was associated with increased self-reported anxiety across all threat conditions, pr(31) = .57, p b .001. There were no main effects or interactions for STAI-State (ps N .13).
4. Discussion The present study examined the impact of UCS reinforcement rate on the startle reflex and self-reported anxiety during a fear conditioning task. Across all participants, the startle reflex was greater during the 75% relative to the 50% reinforcement condition. This suggests that the reinforcement rate with which participants received electric shocks
Fig. 3. Scatterplot depicting the association between IUS and self-reported anxiety collapsing across the CS+ and CS− trial types and 50% and 75% reinforcement conditions. Self-reported anxiety ranged from 1 (not at all) to 4 (very). au = arbitrary units; CS = conditioned stimulus; IUS = intolerance of uncertainty scale.
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006
B. Chin et al. / International Journal of Psychophysiology xxx (2015) xxx–xxx
potentiated the defense startle reflex across all trial types. Participants also had greater self-reported anxiety to the CS+ relative to the CS− in the 75% relative to the 50% reinforcement condition. On the other hand, FPS (i.e., CS + minus CS −) did not differ between the 50% and 75% conditions. However, individual differences in IU were positively correlated with startle potentiation during the 50% (but not 75%) reinforcement condition. In other words, high IU individuals demonstrated increased defensive responding in the context of the less frequent (but more unpredictable) threat. IU was also associated with increased anxiety across both conditions. These associations were independent of state anxiety, which did not impact startle magnitude, FPS, or selfreported anxiety in either condition. This study supports the notion that the reinforcement rate of threat impacts defense system activation and this relationship is influenced by individual differences in IU. These findings add to a growing literature linking IU and psychobiological responding to uncertain threat. For example, greater IU has been associated with increased insula (Shankman et al., 2014; Simmons et al., 2008) and amygdala (Schienle et al., 2010) activation during the processing of uncertain threat. Furthermore, high IU was associated with an enhanced event-related potential P200 response during the anticipation of uncertain valenced pictures (Gole et al., 2012). The present study is one of the first to demonstrate an association between IU and the frequency of threat reinforcement. Despite receiving fewer electric shocks, high IU participants demonstrated greater startle potentiation during the 50% reinforcement condition. These results are consistent with previous work demonstrating that uncertainty can lead to suboptimal reactions that subsequently impact decision-making (Ladouceur et al., 1997; Maner et al., 2007). For example, Luhmann et al. (2011) found that during a delayed probabilistic reward task, high IU individuals were more likely to choose a smaller and less probable (but immediate) reward instead of a larger and more probable (but delayed) reward. Although we did not directly measure decision-making, the current study adds to a growing literature indicating that, in individuals with high IU, uncertainty can contribute to irrational responding. IU has been linked to a number of different psychopathological conditions, including GAD (Dugas et al., 2004), OCD (Tolin et al., 2003), panic disorder (Carleton et al., 2013), social anxiety (Boelen and Reijntjes, 2009), and depression (McEvoy and Mahoney, 2012), and evidence suggests that IU may be broadly associated with increased negative affect and anxiety symptomatology. However, there are several potential methodological limitations to self-report measures of emotional reactivity that may impact their association with IU, including demand characteristics, restricted range of responses and ceiling effects, and reliance on retrospective recall. Importantly, the present study highlights the potential utility of psychophysiological measures (e.g., startle reflex) that are more sensitive to subtle differences in threat responding and may demonstrate better specificity in relation to anxiety phenotypes (e.g., IU). The startle reflex across the CS+ and CS− was greater during the 75% relative to 50% reinforcement condition. This result may have been due to greater shock sensitization during the 75% reinforcement condition (Davis, 1989). Furthermore, as expected self-reported anxiety potentiation (i.e., difference between the CS+ and CS −) was greater during the 75% relative to 50% reinforcement condition. However, in contrast to our hypothesis FPS did not differ across 50% and 75% shock reinforcement. It is important to note that the 50% and 75% reinforcement conditions only differed by two reinforced trials, and it is possible that a greater difference between reinforcement rates (e.g., 25% vs. 75%) may impact psychophysiological indices of fear learning. It is also possible that lengthening the acquisition phase (i.e., initial trials) and better separating the acquisition and ‘test’ phase (i.e., later trials after contingency has been learned) may have influenced how the reinforcement rate impacted FPS. The present study had several limitations that warrant consideration. First, the experience of two consecutive varying reinforcement
5
rates may have altered fear acquisition and attentional processes, and future studies should consider including a comparison group with consistent reinforcement rates. Second, the present study only measured individual differences in broad state anxiety, and it is unclear whether IU was associated with startle potentiation during the 50% condition independent of trait anxiety. However, we speculate this effect would be independent of trait anxiety given that the STAI-State and Trait measures have been shown to be strongly correlated (Grös et al., 2007). Finally, participants did not provide expectancy ratings for each reinforcement condition, which limited our ability to examine potential differences in contingency awareness. In conclusion, the present study found that the reinforcement rate of threat impacted emotional and motivational responding during a fear conditioning task and this was influenced by individual differences in IU. Specifically, across all participants greater UCS frequency was associated with an increased startle reflex. However, differences in FPS varied as a function of IU, such that IU was associated with enhanced startle potentiation during the more uncertain (but less frequent) reinforcement condition. These findings add to a growing literature demonstrating the relationship between IU and psychophysiological measures of sensitivity to uncertain threat. Future research should examine the impact of different reinforcement rates on other aspects of fear learning, including extinction and reinstatement. Furthermore, future studies might leverage this paradigm in clinically anxious individuals who are high on IU, and/or determine whether treating IU reduces sensitivity to unpredictable threat. References Barlow, D.H., 2000. Unraveling the mysteries of anxiety and its disorders from the perspective of emotion theory. Am. Psychol. 55, 1247. Boelen, P.A., Reijntjes, A., 2009. Intolerance of uncertainty and social anxiety. J. Anxiety Disord. 23, 130–135. Boswell, J.F., Thompson-Hollands, J., Farchione, T.J., Barlow, D.H., 2013. Intolerance of uncertainty: a common factor in the treatment of emotional disorders. J. Clin. Psychol. 69, 630–645. Carleton, R.N., Fetzner, M.G., Hackl, J.L., McEvoy, P., 2013. Intolerance of uncertainty as a contributor to fear and avoidance symptoms of panic attacks. Cogn. Behav. Ther. 42, 328–341. Carleton, R.N., Duranceau, S., Freeston, M.H., Boelen, P.A., McCabe, R.E., Antony, M.M., 2014. “But it might be a heart attack”: intolerance of uncertainty and panic disorder symptoms. J. Anxiety Disord. 28, 463–470. Davis, M., 1989. Sensitization of the acoustic startle reflex by footshock. Behav. Neurosci. 103, 495–503. Davis, M., 1992. The role of the amygdala in fear and anxiety. Annu. Rev. Neurosci. 15, 353–375. Davis, M., 1998. Are different parts of the extended amygdala involved in fear versus anxiety? Biol. Psychiatry 44, 1239–1247. Dugas, M.J., Buhr, K., Ladouceur, R., 2004. The role of intolerance of uncertainty in etiology and maintenance. In: Heimberg, R.G., Turk, C.L., Mennin, D.S. (Eds.), Generalized Anxiety Disorder: Advances in Research and Practice. Guilford Press, New York, NY, pp. 143–163. Fetzner, M.G., Horswill, S.C., Boelen, P.A., Carleton, R.N., 2013. Intolerance of uncertainty and PTSD symptoms: exploring the construct relationship in a community sample with a heterogeneous trauma history. Cogn. Ther. Res. 37, 725–734. Freeston, M.H., Rhéaume, J., Letarte, H., Dugas, M.J., Ladouceur, R., 1994. Why do people worry? Personal. Individ. Differ. 17, 791–802. Gallistel, C.R., Gibbon, J., 2000. Time, rate, and conditioning. Psychol. Rev. 107, 289–344. Giargiari, T.D., Mahaffey, A.L., Craighead, W.E., Hutchison, K.E., 2005. Appetitive responses to sexual stimuli are attenuated in individuals with low levels of sexual desire. Arch. Sex. Behav. 34, 547–556. Gole, M., Schäfer, A., Schienle, A., 2012. Event-related potentials during exposure to aversion and its anticipation: the moderating effect of intolerance of uncertainty. Neurosci. Lett. 24, 112–117. Grillon, C., Ameli, R., Foot, M., Davis, M., 1993. Fear-potentiated startle: Relationship to the level of state/trait anxiety in healthy subjects, Bio. Psychiatry 15, 566–574. Grillon, C., Baas, J.P., Lissek, S., Smith, K., Milstein, J., 2004. Anxious responses to predictable and unpredictable aversive events. Behav. Neurosci. 118, 916. Grillon, C., Baas, J.M., Pine, D.S., Lissek, S., Lawley, M., Ellis, V., Levine, J., 2006. The benzodiazepine alprazolam dissociates contextual fear from cued fear in humans as assessed by fear-potentiated startle. Biol. Psychiatry 60, 760–766. Grillon, C., Lissek, S., Rabin, S., McDowell, D., Dvir, S., Pine, D.S., 2008. Increased anxiety during anticipation of unpredictable but not predictable aversive stimuli as a psychophysiologic marker of panic disorder. Am. J. Psychiatr. 165, 898–904. Grös, D.F., Antony, M.M., Simms, L.J., McCabe, R.E., 2007. Psychometric properties of the state-trait inventory for cognitive and somatic anxiety (STICSA): comparison to the state-trait anxiety inventory (STAI). Psychol. Assess. 19, 369–381.
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006
6
B. Chin et al. / International Journal of Psychophysiology xxx (2015) xxx–xxx
Ladouceur, R., Talbot, F., Dugas, M.J., 1997. Behavioral expressions of intolerance of uncertainty in worry experimental findings. Behav. Modif. 21, 355–371. Lang, P.J., Bradley, M.M., Cuthbert, B.C., 1990. Emotion, attention, and the startle reflex. Psychol. Rev. 97, 377–395. Lang, P.J., Bradley, M.M., Cuthbert, B.N., 1998. Emotion, motivation, and anxiety: brain mechanisms and psychophysiology. Biol. Psychiatry 44, 1248–1263. Lang, P.J., Davis, M., Ohman, A., 2000. Fear and anxiety: animal models and human cognitive psychophysiology. J. Affect. Disord. 61, 137–159. Lissek, S., Powers, A.S., McClure, E.B., Phelps, E.A., Woldehawariat, G., Grillon, C., Pine, D.S., 2005. Classical fear conditioning in the anxiety disorders: a meta-analysis. Behav. Res. Ther. 43, 1391–1424. Luhmann, C.C., Ishida, K., Hajcak, G., 2011. Intolerance of uncertainty and decisions about delayed, probabilistic rewards. Behav. Ther. 42, 378–386. Maner, J.K., Richey, J.A., Cromer, K., Mallott, M., Lejuez, C.W., Joiner, T.E., Schmidt, N.B., 2007. Dispositional anxiety and risk-avoidant decision-making. Personal. Individ. Differ. 42, 665–675. McEvoy, P.M., Mahoney, A.E., 2012. To be sure, to be sure: intolerance of uncertainty mediates symptoms of various anxiety disorders and depression. Behav. Ther. 43, 533–545. Nelson, B.D., Shankman, S.A., 2011. Does intolerance of uncertainty predict anticipatory startle responses to uncertain threat? Int. J. Psychophysiol. 81, 107–115.
Norrholm, S.D., Jovanovic, T., Vervliet, B., Myers, K.M., Davis, M., Rothbaum, B.O., Duncan, E.J., 2006. Conditioned fear extinction and reinstatement in a human fear-potentiated startle paradigm. Learn. Mem. 13, 681–685. Schienle, A., Köchel, A., Ebner, F., Reishofer, G., Schäfer, A., 2010. Neural correlates of intolerance of uncertainty. Neurosci. Lett. 479, 272–276. Shankman, S.A., Robison-Andrew, E.J., Nelson, B.D., Altman, S.E., Campbell, M.L., 2011. Effects of predictability of shock timing and intensity on aversive responses. Int. J. Psychophysiol. 80, 112–118. Shankman, S.A., Gorka, S.M., Nelson, B.D., Fitzgerald, D.A., Phan, K.L., O'Daly, O., 2014. Anterior insula responds to temporally unpredictable aversiveness: an fMRI study. Neuroreport 25, 596–600. Simmons, A., Matthews, S.C., Paulus, M.P., Stein, M.B., 2008. Intolerance of uncertainty correlates with insula activation during affective ambiguity. Neurosci. Lett. 430, 92–97. Spielberger, C.D., Gorsuch, R.L., Lushene, R., Vagg, P.R., Jacobs, G.A., 1983. Manual for the State-Trait Anxiety Inventory. Consulting Psychologists Press, Palo Alto, CA. Tolin, D.F., Abramowitz, J.S., Brigidi, B.D., Foa, E.B., 2003. Intolerance of uncertainty in obsessive-compulsive disorder. J. Anxiety Disord. 17, 233–242. Walker, D.L., Davis, M., 2008. Role of the extended amygdala in short-duration versus sustained fear: a tribute to Dr. Lennart Heimer. Brain Struct. Funct. 213, 29–42.
Please cite this article as: Chin, B., et al., Intolerance of uncertainty and startle potentiation in relation to different threat reinforcement rates, Int. J. Psychophysiol. (2015), http://dx.doi.org/10.1016/j.ijpsycho.2015.11.006