Trait Boldness and Emotion Regulation: An Event-Related Potential Investigation

Abstract

The present study sought to extend knowledge of the role of boldness, a transdiagnostic bipolar trait dimension involving low sensitivity to threat, in emotional reactivity and regulation using physiological and report-based measures. One prior study found that boldness was associated with reduced late positive potential (LPP) while passively viewing aversive images, but not during emotion regulation; a disconnect between LPP and self-reported reactivity was also observed. Here, participants (N = 63) completed an emotion regulation task in which they either passively viewed or effortfully up- or downregulated their emotional reactivity to pleasant, unpleasant, and neutral pictures while EEG activity was recorded; they later retrospectively rated the success of their regulation efforts. ANOVAs examining the interactive effects of regulation instruction and boldness on LPP amplitude revealed that lower boldness (higher trait fearfulness) was associated with paradoxical increases in LPP to threat photos during instructed downregulation, relative to passive viewing, along with lower reported regulation success on these trials. Unexpectedly, similar LPP effects were observed for affective images overall, and especially nurturance photos. Although subject to certain limitations, these results suggest that individual differences in boldness play a role not only in general reactivity to aversive stimuli, as evidenced by prior work, but in the ability to effortfully downregulate emotional response.

Emotion and its regulation play a key role in the emergence and amelioration of psychopathology. One approach to understanding and potentially mitigating clinical difficulties with emotion regulation is through use of psychophysiological measures, such as electroencephalography (EEG). The current study investigated how boldness, a transdiagnostic trait that confers risk for fear pathology at low levels and psychopathy at high levels, relates to physiological and report-based indices of emotional reactivity and regulation.

Emotion Regulation

Emotion is a multifaceted phenomenon that arises when an individual attends to a situation and deems it relevant to their current goals (Lazarus, 1991); it serves several important functions, including motivating adaptive action and facilitating social interaction (Gross, 1998). A crucial, widely recognized characteristic of emotion is its malleability. The implicit theory of emotion malleability (Tamir et al., 2007) posits that emotions are dynamic in nature and can be influenced by top-down control in the form of emotion regulation (Kneeland et al., 2016). The ability to modulate one’s emotional experience and emotional expression in alignment with one’s conscious goals and intentions (Gross, 1998; Gross, 2009) can mitigate functional impairments arising from exposure to aversive events and positively influence interpersonal functioning and general mood (Lopes et al., 2005; Gross, 1998; Gross and John, 2003; Tamir et al., 2007).

In turn, emotion dysregulation — difficulty acknowledging and accepting emotions and behaving adaptively under emotional circumstances (Stanton et al., 2016) — is a transdiagnostic process underlying a range of clinical problems, including depression, problematic substance use, chronic worry and anxiety (Garnefski & Kraaij, 2006, Mennin et al., 2007), and violence and aggression (Davidson, 2000).

Psychophysiology of Emotion Regulation: The Late Positive Potential

Emotions can be conceptualized and measured within the domains of subjective report, overt behavior, and physiology, with these modalities providing non-redundant information about affective experience (Mauss et al., 2005). Accordingly, a major line of research has investigated the psychophysiology of emotion and its regulation, with the goal of establishing a comprehensive empirical model of emotion and informing clinical assessment and intervention for forms of psychopathology characterized by emotion dysregulation. Much of this work has centered on the late positive potential (LPP), an event-related potential that is sensitive to the emotional content of visual stimuli. Pleasant and unpleasant images elicit a larger LPP than neutral images, and this effect is maintained for several seconds following stimulus presentation (Hajcak et al., 2010). The largest LPPs are observed for stimuli with strong motivational significance, such as erotica and images of bodily mutilation (Weinberg & Hajcak, 2010). However, these effects are dependent on attention (Raymond, 2009), as LPP is attenuated when attention is directed away from emotionally arousing aspects of an image (Hajcak et al., 2009). Together, these results suggest that LPP indexes “bottom-up” emotion reactivity, the inherent, raw response to the sensory properties of an attended emotional stimulus (Connor et al., 2004). Importantly, however, LPP is also sensitive to conscious, “top-down” processes, in which cortical efforts influence otherwise involuntary responses to stimuli (McRae et al., 2011). For example, findings demonstrate that LPP amplitude is modulated by instructed emotion regulation, increasing or decreasing with reappraisal of the images’ meaning (for a review, see Hajcak et al., 2010). Reappraisal-related reductions in LPP to negative images are also associated with participants’ self-report of changes in emotional experience (Hajcak & Nieuwenhuis, 2006), suggesting concordance between LPP and subjective accounts of emotion regulation. In sum, LPP appears to index both basic, bottom-up reactivity to motivationally salient stimuli and conscious, top-down engagement with emotional content.

Boldness and Emotion

Boldness is a dispositional trait with great relevance to emotion processing; it is considered the nexus of fearlessness and social dominance (Patrick et al., 2009, 2019). Specifically, boldness represents the common variance among social assurance, dominance, persuasiveness, self-confidence, optimism, resilience, valor, intrepidness, and tolerance of uncertainty (Patrick et al., 2019). Boldness operates transdiagnostically, showing strong positive relations to forms of psychopathology characterized by blunted emotional reactivity, and negative to those involving enhanced emotional reactivity. Moreover, it is considered a bipolar trait, with both high and low scores showing clear clinical relevance. For example, high boldness is considered a central component of some models of psychopathic personality (e.g., the triarchic model of psychopathy; Patrick & Drislane, 2015; Patrick et al., 2009), as original descriptions of psychopathy emphasized adaptive traits such as absence of nervousness, resilience to stress, and social charm (Cleckley, 1941).¹ Consistent with the notion of boldness as adaptive, it demonstrates consistent negative relations to internalizing disorders (Brislin et al., 2017; Eisenbarth et al., 2021; Latzman et al., 2019, 2020), especially those involving maladaptive fear (Latzman et al., 2020). The lower pole of the boldness dimension has therefore been conceptualized as reflecting high fearfulness and threat sensitivity, a risk factor for anxiety-related problems (Kramer et al., 2012, 2020; Vaidyanathan et al., 2009). However, although clearly related to lack of fear pathology, high boldness cannot be considered fully adaptive, as it is also related to narcissism, manipulativeness, and fearless sensation-seeking (Benning et al., 2005a; Brislin et al., 2015, 2017; Fanti et al., 2016; Sellbom et al., 2016; Stanley et al., 2013). In addition, some research has found positive associations with externalizing problems (Fanti et al., 2016; Hall et al., 2014; Hicks et al., 2014), especially in combination with other psychopathic traits (Baroncelli et al., 2020; Drislane et al., 2014). These associations may be attributable to the fearless sensation-seeking inherent in boldness (Anderson et al., 2021; Benning et al., 2005a; Brislin et al., 2017; Patrick et al., 2009). Taken together, prior research suggests that the bipolar boldness-trait fear dimension demonstrates transdiagnostic relevance to forms of psychopathology involving excessive or deficient fear.

Boldness has been conceptualized as a normally distributed neurobehavioral trait involving low dispositional sensitivity to threat (Patrick, in press; Patrick et al., 2009; Patrick & Drislane, 2015). This notion is supported by research demonstrating negative associations between boldness and reactivity to phasic aversive stimuli. Specifically, in the context of viewing unpleasant versus neutral pictures, community participants higher in boldness show smaller potentiation of the startle response to sudden auditory probes (i.e., reduced fear-potentiated startle; Benning et al., 2005b; Esteller et al., 2016). In addition, incarcerated individuals diagnosed with psychopathy, which includes elements of boldness (Venables et al., 2014), also exhibit reduced aversive startle potentiation (Patrick, 1994; Vaidyanathan et al., 2011). Conversely, trait fear (i.e., the lower pole of the boldness dimension) is associated with greater potentiation (Cook et al., 1992; Kramer et al., 2012; Yancey et al., 2016) and accounts for the association between aversive startle potentiation and fear disorders (Yancey et al., 2015). These findings have been interpreted as evidence that individual differences in bottom-up physiological defensive reactivity to attended aversive stimuli manifest in report-based measures of normal-range personality (i.e., boldness, threat sensitivity) as well as psychopathology (i.e., psychopathy and fear pathology).

Importantly, aversive cue-elicited defensive reactivity can be tempered by top-down influences in the form of emotion regulation. However, individuals differ in their ability to exert effortful up- or downregulation of emotional responses (Myruski et al., 2019). Anxiety disorders are associated with reliance on maladaptive emotion regulation strategies and underuse of more effective techniques (Aldao et al., 2010), resulting in emotion dysregulation. In line with this finding, prior research using the current sample demonstrated that individuals high in threat sensitivity (i.e., low in boldness) reported more frequent use of expressive suppression, an ineffective regulation strategy (Perkins et al., 2019). In addition, unlike some other aspects of psychopathy (Garofalo et al., 2018), boldness appears negatively related to emotion dysregulation (Donahue et al., 2014), which helps to inhibit impulsive aggression (Long et al., 2014). Given these findings, it is plausible that boldness — the transdiagnostic factor connecting psychopathy and anxiety — may play a role in the effortful ability to up- or downregulate emotional reactivity, rather than being exclusively relevant to initial reactivity. Although the tendency to report emotion dysregulation and reliance on maladaptive strategies appear inversely related to boldness, it remains unclear whether boldness is associated with the ability to deploy more effective regulation strategies when instructed. Evidence in support of this possibility could have important implications for personalized therapy approaches. The current study sought to investigate the role of boldness in bottom-up emotional reactivity and top-down regulation, as reflected in the LPP brain response and experiential report.

Some previous research has begun to explore these questions. Yancey et al. (2016) demonstrated that a measure of threat sensitivity — known to correlate strongly, in reverse, with boldness (Patrick et al., 2019) — was positively associated with LPP amplitude to aversive (versus neutral) images. In other words, in the absence of instructions to regulate, higher-bold participants demonstrated a relatively small overall LPP to aversive images. However, that study design could not disentangle the effects of reactivity and regulation, as participants could have naturalistically downregulated their emotional response. Another study by Ellis et al. (2017) examined the facets of the triarchic model of psychopathy, including boldness, in relation to LPP during an emotion regulation task with aversive and neutral pictures. Boldness was associated with smaller LPP while passively viewing aversive photos but was unrelated to LPP during instructed regulation, replicating the finding of diminished physiological reactivity to aversive stimuli in boldness but not detecting regulation effects. Interestingly, high-bold participants in this study did not report blunted reactivity to aversive scenes; moreover, they rated their reactivity during regulation trials as higher than lower-bold participants. In other words, high-bold participants (1) did not perceive themselves to be underreactive to passively viewed aversive scenes, despite showing reduced LPP, (2) believed they were highly successful in upregulating their response to aversive photos, when the LPP results indicated otherwise, and (3) perceived their reactivity to aversive photos to be high when following instructions to downregulate it, a view again belied by null LPP effects. The authors interpreted these findings as indicating a disconnect between physiological and self-report indices of emotional reactivity and regulation among highly bold individuals, perhaps due to a lack of insight.

The Present Study

The current study sought to replicate and extend the prior literature on boldness and emotional reactivity and regulation, as indexed by the LPP, in terms of three major aims. Aim 1 was to determine whether boldness was associated with LPP during passive viewing and/or attempts to up- or downregulate reactivity to aversive and pleasant images. Regarding aversive images, based on Ellis et al. (2017), we predicted that boldness would relate specifically to smaller LPP during passive viewing, but not regulation trials, implicating a role for boldness in bottom-up reactivity to unpleasant images but not top-down regulation. This prior study did not examine pleasant images, but given the theorized specificity of the relationship between boldness and sensitivity to threat, we hypothesized that no boldness-related effects would be observed for pleasant images. Aim 2 was to examine picture content within the broader categories of aversive and pleasant images. Again, due to the conceptualization of boldness as low threat sensitivity, we anticipated that its association with reduced LPP to passively viewed images would be most evident for directly threatening content (e.g., a gun pointed at the camera; see Vaidyanathan et al., 2009). Finally, Aim 3 extended Ellis et al.’s (2017) examination of discrepancies between physiological and report-based measures of reactivity by exploring participants’ perceptions of their own emotion regulation success. Given the previous study’s contradictory findings for experiential ratings, we did not have any specific a priori hypotheses regarding high-bold participants’ perceptions of their ability to up- or downregulate reactivity according to instructions.

Method

Participants and Procedure

Participants (N = 68) were recruited from an advertisement in the student newspaper and from undergraduate psychology classes at the University of Minnesota. To be eligible for the study, individuals were required to complete a screening questionnaire to ensure that they were without visual or hearing impairments. Of this sample, 37 (54.4%) were women and the mean age was 21.66 (SD = 5.08). As for self-reported race and ethnicity, consistent with the demographics of the University of Minnesota at the time of data collection, 63 participants (92.6%) were White and non-Hispanic, two (2.9%) were Black and non-Hispanic, two (2.9%) were Asian and non-Hispanic, and one (1.5%) self-described as “another race” and Hispanic. All participants provided written informed consent before the study began and were compensated with course credit or money after completion. Participation in the study consisted of completion of questionnaires and an EEG task, described below. All procedures complied with the standards set forth in the Declaration of Helsinki and were approved by the University of Minnesota Institutional Review Board (HSC #0002S35481).

Five participants were excluded from analyses due to unusable EEG data, and multiple imputation was used to handle missing data for a given condition for 23 other participants (see “EEG Data Reduction and Quantification”). These procedures led to a final N of 63 for all analyses. A sensitivity analysis was conducted in using the pwr.f2.test function within the “pwr” package (version 1.3; Champely, 2020) in R (version 4.0.4; R Core Team, 2021). The results indicated that with our specified general linear model and sample, the minimum effect size to be detected with .80 power and α = .05 was f² = .16, which is equivalent to η² = .14, a large-sized effect (Cohen, 1988). Although a sample of this size would thus be underpowered to detect effects of the magnitude typically found in cross-modal research (small or medium), our study was intended as an initial exploration of what effects could exist, rather than attempting to draw definitive conclusions about the absence of effects. Therefore, the sample was viewed as sufficient for the purposes of this study, recognizing that any observed effects would need to be replicated in larger samples (see Discussion).

Emotion Regulation EEG Task

Task stimuli and procedure.

Following application of the EEG cap and peripheral electrodes, participants completed a task in which they viewed 60 photographs from the International Affective Picture System (IAPS; Lang et al., 2008). Of these picture stimuli, 20 depicted pleasant content, including 10 erotic images (e.g., nude individuals, intimate couples), 5 food (e.g., French fries, desserts), and 5 nurturance (e.g., babies, young animals). Another 20 contained unpleasant themes, including 10 threat images (e.g., pointed guns, snakes), 5 mutilation (e.g., burn victims, severed hand), and 5 contamination (e.g., pollution, animal carcass). The final 20 were neutral pictures consisting of household objects (e.g., hair dryer), buildings, and expressionless human faces. Each pleasant and unpleasant picture was presented 3 times throughout the task, with a different instruction before each repetition: Participants were asked to “enhance” or increase the intensity of the emotion they felt toward the picture, “suppress” or decrease the intensity of emotion, or simply “view” the picture without attempting to change their emotions. Neutral pictures were presented only once, under the “view” condition, to avoid confusing participants about changing their emotional response to non-affective images. Before beginning, the task was explained following the instructions by Jackson et al. (2000). Appropriate and inappropriate regulation strategies were discussed to encourage participants to engage in cognitive reappraisal — changing their interpretation of the stimulus — rather than, for example, generating unrelated emotions, thinking of things unrelated to the picture, averting their eyes, or only focusing on parts of the picture. Three practice trials were completed; for each, participants were asked to describe the strategies they had used and provided corrective guidance to encourage reappraisal. If participants were unable to provide satisfactory (appraisal-relevant) responses after three trials, the instructions were repeated and an additional trial took place. If any participant remained confused after four practice trials, they would have been excluded from the study (n = 0 in this sample).

In total, the task consisted of 140 trials, in which participants viewed the instruction (“enhance,” “suppress,” or “view”) for 6 seconds. While the instruction was on the screen, a high- or low-pitched tone was played, and participants were instructed to press a button when they heard the high-pitched tone to ensure attention to the instructions. Immediately following the instruction screen, the picture was presented for 6 seconds. Noise probes were delivered during 126 of 140 picture presentations (90%) to elicit the startle-blink response, with varying latency such that the probe occurred 300, 800, 3000, 4000, or 5000 milliseconds (ms) after picture onset. Instruction type and probe latency were distributed proportionately across picture valence and content categories, and trial order was randomized with the constraint that no more than two consecutive trials contained the same valence, instruction, or probe onset. Between trials, a fixation cross appeared for either 2 or 3 seconds. Participants were observed throughout the session via a closed-circuit television system to ensure they were awake and focused on the task.

Although not used in the current analyses, participants completed valence and arousal ratings for a subset of the picture stimuli using the Self-Assessment Manikin (Bradley & Lang, 1994). Along with the EEG data analyzed here, other physiological responses such as facial electromyography (EMG, including the startle-blink response mentioned above), electrocardiography, and skin conductance were recorded. Analysis of these various physiological responses, as well as more detailed information about the procedure, is reported in Bernat et al. (2011).

Stimulus delivery and physiological response measurement.

Task stimuli were presented on a 21” monitor while participants were seated in a comfortable armchair 100 centimeters away. One computer, running E-prime software (Psychology Software Tools), was used to present task stimuli and record behavioral responses. A second computer acquired physiological data using Neuroscan Acquire and a Neuroscan SynAmps amplifier. Participants were fitted with a 64-channel Neuroscan Quick-Cap, following the 10–20 system, and electrodes were referenced online to CPz. Impedances were kept below 10 KOhms. The online sampling rate was 500 Hz, and an online analog bandpass filter of .05 to 100 Hz was applied.

EEG data reduction and quantification.

Offline processing of the EEG data was performed using BrainVision Analyzer 2.1 (BrainProducts GmbH). The raw EEG data were re-referenced to the average of the mastoids, and a 0.1 to 30 Hz bandpass filter was applied (see Tanner et al., 2015). Ocular correction utilized the Gratton, Coles, and Donchin (1983) method. The continuous EEG was epoched for the purposes of LPP analysis using −200 to 1000 ms stimulus-locked windows relative to picture onset. Trials in which the startle probe occurred 300 or 800 ms after picture onset were excluded (n=15 trials each, distributed across valence and instruction), as the neural and muscular responses to the probe could affect the LPP. Trials for which activity exceeded +/− 75 microvolts during any part of the epoch were excluded from further processing. ERP waveforms were computed as the average for each instructional condition-by-valence type (e.g., enhance pleasant images) and for each instructional condition-by-content type (e.g., enhance nurturance images). The LPP was quantified as the mean waveform amplitude at Pz between 400 and 1000 ms following picture onset, relative to a 200 ms pre-stimulus baseline. We chose to measure LPP at electrode Pz for two reasons: (1) the prior emotion regulation literature has focused on midline electrodes, including an earlier paper using this sample that selected Pz (Bernat et al., 2011), and (2) the LPP was maximal at Pz in this sample. EEG data for five participants were excluded from analyses due to equipment malfunction occurring during data collection (n = 2), unusable data for the reference electrode throughout the task (n = 2), or excessively artifact-ridden data (88% of conditions had insufficient number of trials; n = 1). Other participants with insufficient artifact-free and startle-free trials for a given instructional condition-by-valence type (n = 23) were subjected to multiple imputation based on LPP data for the full sample using the “mice” package (van Buuren & Groothuis-Oudshoorn, 2011) in R. The criteria for minimum trials within a given instructional condition were 7 trials for pleasant, unpleasant, and neutral valences (out of 20 possible trials), 5 for erotic and threat images (out of 10 possible trials), and 3 for other content types (out of 5 possible trials). Multiple imputation was employed to estimate the LPP for an average of 2.04 out of 25 conditions among these 23 participants (range = 1 to 6 conditions imputed).

Split-half reliability was calculated for each condition-by-instruction ERP as the correlation between average amplitude for even and odd trials, corrected using the Spearman-Brown prophecy formula (r_SB; see Table 1). These analyses used only observed data, not the multiply imputed data described above, to avoid artificial inflation of reliability. Reliability was acceptable to good for each instructional condition in the overall pleasant, unpleasant, and neutral valences (r_SBs ranged from .71 to .82; mean number of trials included in LPP [M_trials] ranged from 13.10 to 15.65 out of 20 presentations), apart from the unpleasant-suppress condition, which showed marginal reliability (r_SB = .68, M_trials = 15.17). For the specific content types, reliability was expectably lower given the smaller number of trials included; M_trials ranged from 7.37 to 7.79 out of 10 presentations each for erotic and threat images and 3.73 to 4.03 out of 5 presentations each for the other content types. Reliability was acceptable to good (i.e., r_SB > .70) for erotic images (all instructions) and threat images (enhance and view instructions), and marginal (i.e., r_SB > .60) for nurturance (view), contamination (enhance), and mutilation (suppress) images. Reliability for the other content-by-instruction categories was poor (r_SB < .60).

Table 1.

Descriptive statistics for main study variables.

	Mean	Std. Dev.	Range	Reliability
MPQ-Boldness	.57	.21	.00 to .95	.81a
Late Positive Potential (μV)
Unpleasant – Enhance	10.16	6.72	−1.01 to 35.09	.82b
Unpleasant – Suppress	9.01	6.10	−5.46 to 23.67	.68b
Unpleasant – View	8.54	6.09	−5.21 to 25.66	.79b
Pleasant – Enhance	9.92	6.33	−3.15 to 31.70	.77b
Pleasant – Suppress	9.16	6.32	.30 to 31.84	.73b
Pleasant – View	8.47	6.13	−5.36 to 28.74	.79b
Neutral – View	3.17	5.62	−12.96 to 21.56	.71b
Threat – Enhance	10.48	7.02	−1.97 to 35.37	.75b
Threat – Suppress	9.40	6.98	−6.79 to 24.86	.47b
Threat – View	8.76	6.90	−7.98 to 22.60	.83b
Nurturance – Enhance	9.71	7.57	−9.21 to 38.88	.48b
Nurturance – Suppress	9.30	7.79	−5.91 to 31.59	.44b
Nurturance – View	8.26	7.82	−8.84 to 26.15	.66b
Self-Reported Regulation Success
Unpleasant – Enhance	7.37	1.38	3 to 10	--
Unpleasant – Suppress	5.83	1.90	1 to 10	--
Pleasant – Enhance	7.68	1.22	4 to 10	--
Pleasant – Suppress	6.15	1.52	2 to 9	--
Threat – Enhance	7.14	1.55	4 to 10	--
Threat – Suppress	6.71	2.18	1 to 10	--
Nurturance – Enhance	7.43	1.66	3 to 10	--
Nurturance – Suppress	6.40	2.18	1 to 10	--

Note. LPP means, standard deviations, and ranges include multiply imputed data; reliability estimates do not. Self-reported regulation success was based on a single-item rating.

^aPearson’s α;

^bSpearman-Brown-corrected split-half reliability.

Regulation Success Rating Form

Self-reported emotion regulation success was measured using a questionnaire administered at the conclusion of the study visit. Participants were asked to rate their own effectiveness in suppressing their emotions when instructed to do so during the EEG task, using a 1 to 10 Likert scale, where 1 = not effective, 5 = somewhat effective, and 10 = very effective. Ratings were provided for each of the three pleasant and three unpleasant content types, as well as for overall categories of pleasant and unpleasant pictures. Participants then used the same scale to rate their effectiveness in enhancing their emotions while viewing pictures of each valence and content type.

Trait Boldness

Boldness was assessed using a well-validated scale (Brislin et al., 2015, 2017) composed of items from the Multidimensional Personality Questionnaire – Brief Form (MPQ-BF; Patrick et al., 2013). The MPQ-Boldness scale includes construct-relevant items from the MPQ-BF’s Harm Avoidance, Stress Reaction, Social Potency, Well-Being, Achievement, Control, and Unlikely Virtues scales. Prior research with undergraduate and community samples indicates that MPQ-Boldness has good reliability and converges strongly with other established measures of the same construct (Brislin et al., 2015, 2017). Its construct validity is supported by findings of negative associations with multiple measures of dispositional fear, anxiety, and negative affect, and positive relations to sensation-seeking, positive affect, sociability, and narcissistic personality (Brislin et al., 2015, 2017). In the current sample, MPQ-Boldness demonstrated good internal consistency reliability (α = .81).

Analytic Approach

Analyses were conducted in SPSS version 27 (IBM Corp., Armonk, NY). The first aim was to examine boldness’s association with LPP during passive viewing versus up- and downregulation of emotional response to unpleasant and pleasant images. An omnibus analysis of variance (ANOVA) was used to test a 2-by-3 valence-by-instruction model (within-subjects factors) with boldness included as a continuous between-subjects factor. If valence were found to interact with the other factors, separate ANOVAs would be run with instruction and boldness factors for unpleasant and pleasant images to clarify the nature of effects. We planned to follow up on significant instruction-by-boldness interactions by examining correlations between boldness and LPP for each instructional condition, as well as with between-condition difference scores (e.g., for suppress versus view). Again, it was hypothesized that boldness would be selectively associated with reduced LPP during unpleasant picture-viewing, but not under either regulation condition, and would not be associated with LPP to pleasant pictures. For the second aim, pertaining to specific pleasant and unpleasant image contents, the same analytic approach was used to test a 6-by-3 content-by-instruction model, again with boldness as a continuous between-subjects factor. Any significant interactions with content would again be probed by conducting separate ANOVAs for each content type, and instruction-by-boldness interactions by computing correlations between boldness and LPP modulation. It was hypothesized that LPP responses while viewing directly threatening stimuli would show the greatest association with boldness. For the last aim, to investigate potential discrepancies between physiological and self-report findings, correlations were used to assess whether boldness was associated with self-reported efficacy of regulation for pleasant or unpleasant images, as well as specific content types. Effect sizes were interpreted according to Cohen (1988): partial η²s of .01, .06, and .14 and Pearson’s rs of .10, .30, and .50 were considered small, medium, and large, respectively.

Results

Table 1 contains descriptive statistics for main study variables; complete descriptive statistics can be found in Supplemental Table 1. Waveform plots for the LPP for each valence (pleasant, unpleasant, neutral) and instructional condition (enhance, suppress, view) are presented in Figure 1; topographical maps are shown in Figure 2. Waveform plots and topographical maps for each picture content type are provided in Supplemental Figures 1–6.²

Figure 1.

Waveform plots by valence and instructional condition.

Figure 2.

Topographical maps by valence and instructional condition.

Preliminary analyses focusing on overall valence and instruction conditions were performed to confirm basic effects. First, a one-way ANOVA was used to test for expected differences in LPP to pleasant, neutral, and unpleasant images under passive viewing instructions. The model was significant, F(2,124) = 50.24, p < .001, partial η² = .45. Follow-up paired-samples t-tests demonstrated that pleasant and unpleasant images elicited a larger LPP than neutral images, ts(62) = 8.21 and 8.66, partial η²s = .52 and .55, respectively, ps < .001. These differences remained significant when tested against Tukey’s critical value (α = .05) of t = 1.84. Pleasant and unpleasant LPPs did not differ from one another, t(62) = −.12, p = .91, partial η² < .001. Next, a 3-by-2 repeated-measures ANOVA tested for differential LPP responses as a function of instructional condition (enhance, suppress, view) for pleasant and unpleasant images. This analysis revealed a significant main effect only for instruction, F(2,124) = 8.19, p < .001, partial η² = .12. Follow-up paired-samples t-tests of averaged pleasant and unpleasant images demonstrated that enhance instructions resulted in a larger LPP than suppress instructions, t(62) = 2.45, p = .02, partial η² = .09, and view instructions, t(62) = 3.81, p < .001, partial η² = .19, and these statistics exceeded Tukey’s critical value (α = .05) of t = 1.84. LPPs under suppress and view instructions did not differ significantly, t(62) = 1.63, p = .11, partial η² = .04. Valence and instruction did not interact, F(2, 124) = .12, p = .89, partial η² < .01, indicating a similar effect of instruction on LPP to both pleasant and unpleasant images.

Aim 1: Boldness and LPP to Unpleasant and Pleasant Images

To address Aim 1, a three-way (valence by instruction type by boldness) ANOVA was conducted for LPP to unpleasant and pleasant images. This omnibus model revealed a main effect of instruction type, F(2,122) = 3.30, p = .04, partial η² = .05, but not valence or boldness, Fs(1,61) < .47, ps > .50, partial η²s < .01. Instruction and boldness interacted, F(2,122) = 3.21, p = .04, partial η² = .05; no other two- or three-way interactions were observed. Given the lack of main or interactive effects for valence, follow-up analyses to clarify directionality were conducted on averaged LPP amplitudes across unpleasant and pleasant images (i.e., affective images overall). Boldness was correlated with the enhance–suppress (r = .27, p = .03) and suppress–view difference scores (r = −.30, p = .02). In other words, for individuals lower in boldness, LPP reactivity to affective images was larger during suppression than enhancement and passive viewing. In contrast, those higher in boldness showed a smaller LPP when instructed to suppress, as compared to enhance, and generally similar LPP amplitude when suppressing and passively viewing (see Table 2 and Figures 3 and 4). A follow-up regression analysis in which the two difference scores (enhance–suppress, suppress–view) were entered as predictors of boldness scores revealed that these effects accounted for shared variance in boldness (ßs = .18 and −.22, ps = .19 and .11, respectively).

Table 2.

Correlations of boldness with LPP amplitudes.

		Instructional Condition or Difference Score
		Enhance	Suppress	View	Enhance – View	Suppress – View	Enhance – Suppress
Picture Type	Affective	.04	−.10	.05	.00	−.30*	.27*
	Nurturance	−.05	−.20	.16	−.20	−.35**	.16
	Threat	.10	−.10	.24	−.13	−.35**	.24

Note. Values represent Pearson’s correlations (r) between boldness and LPP to pictures of each type, by instructional condition and differences between instructional conditions.

^**p < .01;

^*p < .05.

Figure 3.

Scatterplot depicting correlation between boldness and the difference in LPP to affective images between enhance and suppress conditions.

Figure 4.

Scatterplot depicting correlation between boldness and the difference in LPP to affective images between suppress and view conditions.

Aim 2: Boldness and LPP to Specific Content Types

A three-way (content-by-instruction type-by-boldness) ANOVA was conducted for LPP to each of the six types of image content (erotic, food, nurturance, threat, contamination, mutilation). This omnibus model revealed a three-way interaction among content, instruction, and boldness, F(10,610) = 1.88, p = .045, partial η² = .03. To clarify the nature of these effects, separate ANOVAs were conducted for each content. The only content types for which any significant effects were observed were threat and nurturance images. Results for threat images included a medium-sized main effect for instruction, F(2,122) = 4.43, p = .01, partial η² = .07, but not for boldness, F(1,61) = .58, p = .45, partial η² = .01. As hypothesized, instruction and boldness interacted, F(2,122) = 4.15, p = .02, partial η² = .06. Correlations indicated this was driven by the difference between suppress and view conditions (r with boldness = −.35, p = .005), such that those lower in boldness again tended to show a larger LPP to threat images on suppression trials as compared to passive viewing trials, whereas higher-bold individuals showed a smaller LPP on suppression trials as compared to viewing trials (see Figure 5).

Figure 5.

Scatterplot depicting correlation between boldness and the difference in LPP to threat images between suppress and view conditions.

For nurturance, the main effect of instruction was medium-sized, F(2,122) = 5.00, p = .008, partial η² = .08. Although there was no main effect of boldness (F(1,61) = .09, p = .77, partial η² = .00), an instructional condition-by-boldness interaction effect was observed, F(2,122) = 4.25, p = .02, with a medium effect size, partial η² = .07. Follow-up correlations revealed that this effect was driven by the correlation between boldness and the difference between LPP under suppress versus view conditions (r = −.35, p = .005). Specifically, lower-bold individuals demonstrated a larger LPP when suppressing than when viewing, whereas higher-bold participants showed the opposite. This effect is illustrated in Figure 6.

Figure 6.

Scatterplot depicting correlation between boldness and the difference in LPP to nurturance images between suppress and view conditions.

Aim 3: Boldness and Self-Reported Emotion Regulation Efficacy

Pearson’s correlations revealed that self-reported emotion regulation success was significantly correlated with boldness for suppression of reactivity to threat images specifically (r = .26, p = .04; see Figures 7 and 8). These results indicated that higher-bold individuals perceived themselves as more successful in downregulating their reactivity to threat images, relative to their lower-bold counterparts.

Figure 7.

Correlations between boldness and self-reported emotion regulation success.

Note. Negative correlation coefficients are denoted by (–).

Figure 8.

Scatterplot depicting correlation between boldness and self-reported success in downregulating one’s response to threat images.

Discussion

The purpose of the current study was to replicate and extend prior work investigating the role of dispositional boldness in emotional reactivity and regulation, using both physiological and report-based measures. Broadly, our results suggested that lower boldness (i.e., higher trait fearfulness) was associated with ineffective instructed downregulation of LPP response when viewing affective images, especially those depicting scenes of threat or nurturance. Specifically, participants lower in boldness showed a paradoxically enhanced LPP when asked to decrease their emotional response, relative to passive viewing, whereas those higher in boldness were better able to downregulate the LPP. This outcome departed from findings of previous work (Ellis et al., 2017; Yancey et al., 2016), which led us to hypothesize that boldness would be selectively related to LPP during passive viewing of unpleasant images, especially those with threat content. We did not predict any effects for regulation trials, nor for pleasant images. In contrast to this prior research, which showed that boldness was related to emotional reactivity to threat, we found that boldness played a role in instructed emotion regulation. Of note, whereas Ellis et al. (2017) found evidence of a disconnect between physiological and report-based indices of emotional reactivity in high-bold individuals, here higher boldness was associated with relatively more effective regulation of emotional reactivity to threatening images according to both LPP modulation and self-reported regulation success.

The construct of boldness is closely linked to emotional reactivity, as it theorized to involve dispositional immunity to stress (Patrick et al., 2019). Consistent with this view, higher boldness consistently relates to manipulativeness and social dominance, and lower boldness to anxiety disorders, especially those involving pathological fear (e.g., Brislin et al., 2015, 2017; Latzman et al., 2020). Further, individuals higher in boldness exhibit reduced sensitivity to threat, as evidenced by blunted potentiation of the protective startle response during aversive picture-viewing, whereas their lower-bold (more fearful) counterparts demonstrate the opposite pattern (Benning et al., 2005b; Esteller et al., 2016; Vaidyanathan et al., 2009; Yancey et al., 2016). Based on these and other findings, the boldness-fearfulness dimension has been conceptualized as reflecting individual differences in sensitivity to threat cues (Vaidyanathan et al., 2009). Prior research examining the LPP has found that higher-bold individuals exhibit smaller LPP reactivity to passively viewed aversive images than do their less-bold counterparts (Ellis et al., 2017; Yancey et al., 2016). However, our study failed to replicate this effect: Boldness was not associated with LPP amplitude while viewing affective images in our main analyses, and supplementary analyses revealed no significant correlations between boldness and (1) LPP while passively viewing unpleasant images (r = .09, p = .51), or (2) LPP modulation while passively viewing unpleasant images, relative to neutral (r = .05, p = .69). One possible explanation for this finding is that our small sample (N = 63) was underpowered to detect an effect. Indeed, sensitivity analyses (described in the Method) suggested that our sample allowed detection of a minimum η² of .14 (a large effect) with .80 power, whereas the LPP effect in Yancey et al. (2016) was small, η² = .01, and in Ellis et al. (2017), medium, η² = .08. Therefore, although we did not observe the same effect, we cannot infer a null association, given the lack of power.

Importantly, our study design allowed us to examine specific content categories under the umbrella of unpleasant images. Based on the conceptualization of boldness as low threat sensitivity, the largest association was expected for passive viewing of pictures depicting direct threat (i.e., a gun pointed at the viewer). The hypothesized interaction between boldness and instruction (enhance, suppress, view) was indeed observed for threat images. However, unexpectedly, this was driven by a paradoxically enhanced LPP when less-bold (more fearful) participants were asked to downregulate emotional reactivity, relative to passive viewing. In other words, although LPP would in theory be expected to be smaller following instructed downregulation, compared to passive viewing, in this case, more fearful participants showed a larger LPP when downregulating than viewing threat images. Again, a lack of statistical power could have contributed to the lack of effect for emotional reactivity (i.e., LPP during passive viewing). Nonetheless, LPP was enhanced during downregulation for less-bold participants to such a degree that it could be detected despite low power. In other words, the present results suggest a role for the boldness-trait fearfulness dimension in emotion regulation, more so than simple emotional reactivity to threat. The only prior psychophysiological study of boldness and emotion regulation (Ellis et al., 2017) did not specifically examine images of direct threat. This content type would be expected to show the strongest association with boldness, given prior work suggesting that the affective-interpersonal features of psychopathy confer a higher threshold for defensive responding to such images (Levenston et al., 2000). As a result, compared with their lower-bold counterparts, higher-bold individuals may disengage relatively easily from threatening content when instructed to downregulate their reactivity. This interpretation is consistent with prior findings that individuals higher in boldness exhibit less threat-related interference in performance under cognitively demanding instructions (Yancey et al., 2019), and with models of anxiety positing that trait fearfulness facilitates attention to and difficulty disengaging from threat (Cisler & Koster, 2010). By collapsing across different types of unpleasant images, not all of which would be expected to relate to boldness, the expected effects may have been obscured in Ellis et al. (2017).

Physiology is only one component of emotion, and incorporation of different measurement modalities is a crucial part of building a comprehensive model of personality’s role in emotional reactivity and regulation. Our analyses revealed a medium-sized correlation between boldness and self-reported success in downregulating reactivity to threat images. Put differently, less-bold (more fearful) individuals reported less effective downregulation when faced with threat pictures. This convergent evidence supports the notion that individual differences in boldness are related to the ability to effortfully disengage from threatening content and dampen emotional reactivity, on both neural and subjective levels, and increases confidence in our physiological findings. Ellis et al. (2017) reported evidence for a disconnect between higher-bold participants’ self-reported reactivity to passively viewed aversive images (reduced relative to lower-bold participants) and their LPP response to such stimuli (enhanced). In the current study, participants were asked to retrospectively rate how successful they had been in up- or downregulating their emotions for images of each type, rather than rating their overall reactivity to unpleasant images across regulatory and passive viewing trials. Our approach probably captures emotion regulation processes to a greater degree than the ratings in the Ellis et al. (2017) study. However, interestingly, in our sample, ratings of emotion regulation success were uncorrelated with LPP modulation (e.g., for downregulation of threat images, r = .02, p = .89), and in a regression model, their associations with boldness were unique (success rating ß = .26, p = .03; LPP modulation ß = −.35, p < .01). This finding contrasts with earlier work showing that modulation of self-reported reactivity to unpleasant images between reappraisal and viewing conditions was positively correlated with LPP modulation (Hajcak & Nieuwenhuis, 2006). It is important to note that this earlier study involved the construction of difference scores from trial-by-trial ratings of emotional intensity — a more direct corollary of LPP modulation — whereas our study asked participants at the end of the task to what degree they felt they had been successful in effortfully up- or downregulating their emotions. To our knowledge, no other study has examined a measure of retrospective self-reported regulation success, providing no basis for direct comparison. However, it stands to reason that these indices of emotion regulation — LPP modulation and reported success — would account for unique variance in boldness, as one likely reflects cognitive flexibility (i.e., relative control over emotional responsivity) and the other, appraisals of one’s performance (i.e., self-assuredness). Therefore, the current findings suggest that the boldness-fearfulness dimension is associated with multiple distinct facets of threat response downregulation.

Although we hypothesized that individual differences in boldness would be associated with reactivity to negative stimuli specifically, boldness-by-instruction interaction effects were observed for affective images as a whole (i.e., collapsing across pleasant and unpleasant pictures), and nurturance images in particular. For affective images overall, less-bold (more fearful) individuals tended to show a larger LPP while suppressing, relative not only to passive viewing trials, but even to trials in which they attempted to enhance their response. For nurturance images, lower-bold participants demonstrated a paradoxically larger LPP while attempting to suppress relative to passive viewing, similar to the effect found for threat images. These findings must be interpreted with caution, as they were not anticipated and did not extend to report-based measures. However, to the extent they prove replicable, they may reflect the role of social confidence in the boldness-trait fearfulness dimension: Individuals higher in boldness are more interpersonally dominant (Brislin et al., 2017; Patrick et al., 2019), report lower social anxiety (Latzman et al., 2019, 2020), and gravitate toward leadership roles (Neo et al., 2018). These behavioral propensities could reflect individual differences in the ability to disengage from affiliative stimuli in the service of instrumental goals, as reflected in LPP. The presence of similar effects for one content type within each valence category (i.e., threat within unpleasant, nurturance within pleasant) could have contributed to the overall finding that valence (pleasant vs. unpleasant) did not moderate instruction-by-boldness effects. Taken together, these results suggest the association between the boldness-fearfulness dimension and emotion regulation may not be specific to unpleasant images.

The present study must be considered in light of certain limitations. First, it must be noted that some of the basic effects expected in an emotion regulation task were not observed. In particular, although the instruction to enhance resulted in a significantly larger LPP than suppress and passive viewing conditions, there was no significant difference between LPPs during suppress and view trials. That is, within this sample as a whole, effortful downregulation did not successfully decrease the LPP relative to passive viewing, unlike in some past research (for a review, see Hajcak et al., 2010). However, among prior studies using three instruction sets (i.e., enhance, suppress, and view), as ours did, the expected pattern of enhance > view > suppress has not been consistently observed (Baur et al., 2015; Deng et al., 2019; Gardener et al., 2013; Krompinger et al., 2008; Kudinova et al., 2016; Langeslag & van Strien, 2018; Moser et al., 2006, 2009, 2010; Myruski et al., 2019; Wu et al., 2013). Therefore, our results are not completely surprising. At least one leading expert in the psychophysiology of emotion regulation informally compared three-instruction to two-instruction paradigms (i.e., only enhance and view, or only suppress and view), finding that effects were more likely to be robust and replicable for the simpler tasks, perhaps because the more complex and cognitively demanding task design interfered with participants’ ability to comply with each direction (G. Hajcak, personal communication, October 19, 2021). This informal observation suggests that the task design we used likely contributed to the absence of the expected enhance > view > suppress effect in our study.

Ambiguous task instructions may have also played a role. Participant logs revealed some initial confusion during practice trials about how to use reappraisal to up- and down-regulate emotions, which was then addressed by research assistants providing corrective guidance and engaging in discussion with participants to teach them appropriate reappraisal-based strategies, following Jackson et al. (2000). As this was done flexibly and not in a standardized manner, it is difficult to ascertain how well the participant understood the instructions by the time the task began. Mid- and post-experimental assessments completed with a subset of participants (n = 36) were coded for the specific emotion regulation strategy or strategies the participants reported using. In describing how they upregulated their emotions, every participant provided at least one response that was coded as a specific reappraisal strategy (e.g., immersion in the situation), and only two participants (5.56%) did not describe a specific reappraisal strategy for downregulation (e.g., objectification). These data provide confidence that, at the least, participants were able to describe reappraisal strategies they had tried to use. However, in the midst of a complex task (see above), it is plausible that participants had difficulty deploying such strategies, which may have been new to them. Moreover, during the task, downregulation trials instructed the participant to “SUPPRESS,” rather than a more general term such as “DECREASE” or “DOWNREGULATE.” This instruction may have unwittingly encouraged participants to engage in expressive suppression. Unlike cognitive reappraisal, expressive suppression is known to be an ineffective, paradoxical strategy that can actually increase physiological reactivity through attempts to alter only the outward expression of emotion, rather than acting on the internal activation state (Gross, 2002; Gross & Levenson, 1993, 1997). There is reason to believe the tendency to default to expressive suppression under ambiguous instructions would be more common among participants low in boldness, as anxious and fearful individuals report greater habitual use of this emotion regulation strategy (Kinney et al., 2019; Moore et al., 2008; Pan et al., 2019). Indeed, a prior study using survey data from the current sample reported greater habitual use of expressive suppression during daily life among low-bold participants (Perkins et al., 2019). Engagement in expressive suppression by some in the sample could have resulted in the unexpectedly large average LPP on “suppress” trials (cf. Pan et al., 2019), which was particularly evident for less-bold participants. In contrast, higher-bold individuals may have simply defaulted to other regulation strategies than suppression when faced with the instruction to “suppress.” Further research should more comprehensively examine the role of boldness in deployment of specific emotion regulation strategies, both in naturalistic settings and under unambiguous instruction in the lab.

Future efforts to exhaustively delineate the role of boldness in emotional reactivity and regulation should also ensure that task paradigms include a sufficient number of trials for each content type to yield reliable event-related potentials. As the present task was not designed to compare subtypes of pleasant and unpleasant images, certain content categories were presented only 5 times per instructional condition, resulting in poor reliability in some cases. The number of usable trials for LPP averages was further attenuated by the presentation of startle probes during the 400–1,000 ms epoch on 30 trials, which then had to be excluded. Moreover, this design choice prevented us from analyzing LPP throughout the full 6 seconds of picture presentation, as probes were presented on 90% of trials. However, there is reason to believe any regulation effects on LPP would be observed within our time window and would persist into later windows without much change, or even decline (Baur et al., 2015; Ellis et al., 2017; Hajcak & Nieuwenhuis, 2006; MacNamara et al., 2009; Myruski et al., 2019; Paul et al., 2013; Wu et al., 2013). Therefore, the current analyses within the 400–1,000 ms epoch shed important light on the neurophysiology of emotion regulation, despite the inability to examine a more protracted time window.

Given these issues, as well as the complex design and ambiguous task instructions discussed above, the ERP paradigm used here should be modified for subsequent research to include 1) a simpler design (comparing two instructional conditions at a time), 2) clearer and more standardized instruction regarding regulation strategies, 3) more trials per content category, and 4) fewer or no startle probes.

Another limitation of the current study is that the data were drawn from a relatively small sample of college students in Minnesota, which resulted in restricted ranges of age and ethnic/racial background. Culture and age have been found to shape emotion regulation effects on the LPP (Deng et al., 2019; Murata et al., 2013). Therefore, future investigations with similar research questions should span a broader demographic range to ensure the results are generalizable. Given the sample size, we were underpowered to draw conclusions from the absence of expected effects; our findings regarding threat images appear to be robust, as they were detected in a sample of this size, but replication is needed.

These limitations notwithstanding, the current study provides additional insight into the role of boldness in emotional reactivity and regulation. Contrary to expectations, we found that individual differences in boldness were associated with instructed downregulation of the emotional reaction to affective images — especially threat and nurturance pictures — but not with responsivity on passive viewing trials, as indexed by the LPP. Consistent with the physiological results, boldness also predicted ratings of efficacy in downregulating the response to threat pictures. The boldness-trait fearfulness dimension may provide a useful interface between physiological and subjective aspects of emotional experience, shedding further light on the role of individual differences in emotionality in transdiagnostic forms of psychopathology.

Highlights.

The boldness-trait fear dimension plays a role in effortful emotion regulation

Less-bold people showed an enhanced brain response when trying to downregulate

Boldness results were consistent across physiology and self-report for threat images

Physiological effects for boldness extended to other kinds of affective images