Social anxiety disorder (SAD) is characterized by excessive fear of negative evaluation in social situations that contributes to functional impairment and/or clinically significant distress (American Psychiatric Association, 2022). SAD is one of the most common psychiatric disorders, with a lifetime and 12-month prevalence of 12.1% and 6.8%, respectively (Kessler & Wang, 2008) and is associated with worse academic achievement, reduced involvement and productivity in the workforce, higher financial costs, and lower health-related quality of life (Stein et al., 2005; Stein & Stein, 2008). Unfortunately, only about 40–75% of those treated for SAD reach remission, and relapse after effective treatment is common (Davidson et al., 2004; Otto et al., 2000; Rodebaugh et al., 2004; Stein & Stein, 2008). As such, there is a clear need to better understand mechanisms involved in the maintenance of SAD that may be effectively targeted during evidence-based treatment to improve outcomes.
Cognitive-behavioral models of SAD emphasize the role of biased cognitive processes in disorder maintenance (Clark & Wells, 1995; Hofmann, 2007; Rapee & Heimberg, 1997). Attentional bias (AB) to threat is one such mechanism. AB describes the preferential allocation of attentional resources towards threatening over non-threatening stimuli (Cisler & Koster, 2010). Predominant cognitive-behavioral models of SAD converge on the theorized role of ABs to disorder maintenance, but diverge on which types of attentional foci are thought to be most pathogenic. Clark and Wells’ (1995), Hofmann’s (2007), and Rapee and Heimberg’s (1997) models of SAD all propose that social situations trigger a perception of social danger (i.e., high likelihood and severity of negative evaluation), often stemming from negative beliefs about others’ high standards and one’s own shortcomings. Under this condition of threat, maladaptive ABs are activated. Clark and Wells’ (1995) and Hofmann’s (2007) theories argues that socially anxious individuals allocate disproportionate attention solely to internal sources of threat (i.e., self-focused attention), or interoceptive stimuli (e.g., blushing, sweating, feelings of anxiety) that are consistent with one’s negative self-beliefs around social inadequacy. In contrast, Rapee and Heimberg (1997) propose that socially anxious individuals are vigilant towards both internal and external sources of threat, including social-evaluative stimuli such as others’ facial expressions, body language, and behaviors that could signal negative evaluation. In both models, these attentional foci are thought to contribute to a distorted negative mental representation of the self, which heightens the perceived social danger and triggers affective, physiological, cognitive, and behavioral symptoms of anxiety. Specifically, attentional focus towards threats is thought to spike acute anxiety and contribute to negatively biased thought processes following social situations. Thus, these models imply that ABs play a causal, maintaining role in SAD, but there is debate as to whether internally and/or externally focused attention has the most harmful effects (Schultz & Heimberg, 2008).
While elevated attentional focus towards both internal and external sources of threat have been empirically observed in SAD (Amir & Foa, 2001; Gilboa-Schechtman et al., 1999; Hope et al., 1989; Mogg & Bradley, 2002; Norton & Abbott, 2016; Pishyar et al., 2004; Spurr & Stopa, 2002), they are usually studied in isolation rather than together (Norton & Abbott, 2016; Schultz & Heimberg, 2008). The few studies that have examined both types of attentional foci jointly have found stronger evidence for the presence and harmful effects of excessive attention to internal threats than external threats. Employing a dot-probe paradigm, Pineles and Mineka (2005) compared the magnitude of high and low socially anxious individuals’ ABs towards internal sources of threat (i.e., heart rate information) and external sources of threat (i.e., threatening faces). They found evidence for ABs to internal, but not external, threats among socially anxious participants, supporting Clark and Wells’s (1995) and Hofmann’s (2007) assertion that SAD is characterized by excessive self-focused attention. Taking an experimental approach, Woody and colleagues compared the effects of delivering a speech focused on one’s own versus another’s current emotions, thoughts and body sensations, and found that the self-focused speech condition caused greater anxiety in both socially phobic and non-socially phobic participants (Woody, 1996; Woody & Rodriguez, 2000). Similarly, Gaydukevych and Kocovski (2012) assigned socially anxious participants to either attend to their own thoughts, feelings, actions, and body sensations or to their partner’s words and facial expressions during a conversation task, and found that self-focused attention caused greater post-event processing, or maladaptive ruminative thinking, than other-focused attention. While these studies provide evidence for the harmful effects of internally focused attention, they do not fully probe the effects of externally focused attention. This does not allow for a more complete understanding of the extent to which attentional focus on external threats also contributes to the maintenance of SAD (Norton & Abbott, 2016).
The specific causal pathways by which attentional allocation to internal and external threats may contribute to the maintenance of SAD are also unclear. Cognitive-behavioral models imply that excessive attentional focus on threats has effects on not only anxious responding during social stressors, but also on cognitive processing that takes place afterwards that reinforces pathological social fears (Clark & Wells, 1995; Hofmann, 2007; Rapee & Heimberg, 1997). Specifically, it is hypothesized that attentional allocation towards threats during social situations makes the events more anxiety-inducing and salient, and thus triggers maladaptive mental review of the events in the form of post-event processing (PEP) (Clark & Wells, 1995; Hofmann, 2007; Rapee & Heimberg, 1997; Sluis et al., 2017). PEP is a form of repetitive negative thinking characteristic of SAD that is focused on perceived failure in past social situations (Sluis et al., 2017). Like other forms of perseverative thought such as depressive rumination or anxious worry, PEP consumes cognitive resources and is often recurrent, intrusive, and uncontrollable, thus interfering with functioning (Rachman et al., 2000). PEP is theorized to maintain SAD by increasing the salience of negative self-perceptions in social memories, triggering the retrieval of other negative social memories, and reinforcing maladaptive core beliefs about social inadequacy (Clark & Wells, 1995; Hofmann, 2007; Norton & Abbott, 2016; Rapee & Heimberg, 1997). PEP has also been shown to precipitate social withdrawal, avoidance, and safety behaviors (Campbell et al., 2016; Mitchell & Schmidt, 2014; Rachman et al., 2000).
Although existing theory links attentional focus on threat to PEP in the maintenance of SAD (Clark & Wells, 1995; Hofmann, 2007; Rapee & Heimberg, 1997), sparse literature has directly examined this relationship. Observational studies have found a positive association between elevated self-focused attention and PEP (Mellings & Alden, 2000; Rapee & Abbott, 2007), as well an indirect positive relationship between AB towards external threats (disgusted faces) and PEP, mediated by subjective emotional reactivity to a social stressor (Çek et al., 2016). Further, experimentally induced self-focused attention, whether prompted by instruction (Gaydukevych & Kocovski, 2012) or false physiological feedback (Makkar & Grisham, 2013), has been shown to cause increased PEP. Given prior research showing that ABs are associated with increased anxiety during social stressors (Çek et al., 2016; Norton & Abbott, 2016; Woody, 1996; Woody & Rodriguez, 2000), and that state anxiety during social stressors is predictive of subsequent PEP (Kiko et al., 2012; Kocovski & Rector, 2008), it may be the case that biased attention is linked to PEP specifically via effects on acute anxious responding, reflecting perceived social danger. Indeed, prior research shows that the level of anxiety experienced during a social event is strongly associated with the level of negative PEP experienced afterwards, likely because anxiety-laden events are salient and thus are more likely to be encoded and revisited in a ruminative fashion (Kiko et al., 2012; Norton & Abbott, 2016). However, research into the causal pathways by which different types of attentional foci may contribute to the maintenance of SAD is limited.
The present study aimed to examine the causal effects of internally and externally oriented attention on state anxiety following a social stressor and subsequent PEP. High socially anxious participants were randomized to either anchor their attention on internal/interoceptive information, external/social-evaluative information, or a neutral stimulus during an impromptu speech task. Immediately following the speech and 24-hours later, state anxiety and PEP were assessed. Consistent with prior research, we hypothesized that the Internal condition would report significantly higher PEP and state anxiety than the External condition. Consistent with existing cognitive-behavioral theory (Rapee & Heimberg, 1997), we also hypothesized that both the Internal and External conditions would report significantly higher PEP and state anxiety than the Control condition. Finally, we hypothesized that immediate post-speech state anxiety would mediate the relation between condition and PEP 24 h later.
Method
Participants
A total of 123 highly socially anxious participants were recruited, including 85 undergraduates and 38 community members who were equally distributed across the three experimental groups. Participants’ mean age was 24.93 (SD = 10.61) and ranged from 18 to 77 years. The sample was 78.9% female (n = 97), and the racial/ethnic composition was as follows: 67.5% White (n = 83), 26.8% Asian/Pacific Islander (n = 33), 13% Black (n = 16), 4.9% Hispanic/Latino (n = 6), and 0.8% Other (n = 1).
Measures
Mini International Neuropsychiatric Interview- Social Phobia Module (MINI; Sheehan et al., 1998). The MINI is a brief structured diagnostic interview for DSM-5 disorders. The Social Phobia module of the MINI was used in the present study to assess for the presence of clinically significant social anxiety symptoms.
Social Phobia Scale (SPS; Mattick & Clarke, 1998). The SPS is a self-report measure of social phobia fears. 20 items about fears of being scrutinized during social activities (e.g., “I fear I may blush when I am with others.”) are rated on a five-point Likert scale from 0 (not at all) to 4 (extremely), yielding total scores from 0 to 80. Internal consistency of the SPS in the present sample was excellent (Cronbach’s alpha = 0.91).
Subtle Avoidance Frequency Estimation (SAFE; Cuming et al., 2009). The SAFE is a self-report measure of safety behavior utilization in SAD. Participants rate the frequency with which they engage in 32 common safety behaviors in anxiety-inducing social situations (e.g., “Avoid eye contact”) on a Likert scale from 0 (Never) to 4 (Always). Total scores range from 0 to 128. Internal consistency of the SAFE in the present sample was excellent (Cronbach’s alpha = 0.92).
Anxiety Sensitivity Index-3, Social Concerns Subscale (ASI; Taylor et al., 2007). The Social Concerns Subscale of the ASI measures fear of arousal-related symptoms in social contexts. Six items (e.g., “It scares me when I blush in front of people.”) are rated on a Likert scale from 0 (very little) to 4 (very much), yielding total scores from 0 to 24. Internal consistency of the ASI in the present sample was very good (Cronbach’s alpha = 0.80).
Depression, Anxiety, and Stress Scales-21 (DASS; Lovibond & Lovibond, 1995). The DASS is a self-report measure of psychological distress. 21 items assessing past-week symptoms of depression, anxious arousal, and psychological agitation (e.g., “I felt that I had nothing to look forward to.”) are rated on a Likert scale from 0 (Did not apply to me at all) to 4 (Applied to me very much, or most of the time). Item scores are summed and multiplied by two, yielding total scores ranging from 0 to 168. The internal consistency of the DASS in the present sample was excellent (Cronbach’s alpha = 0.90).
Socially Anxious Rumination Questionnaire- Post Event Subscale (SARQ; Abbott & Rapee, 2004; Donohue et al., 2021). The SARQ (previously named, “Thoughts Questionnaire”) is a 12-item self-report measure of PEP following social stressors. Participants report the frequency with which they experienced negative thoughts following a social threat task (e.g., “I made a fool of myself.”) on a scale from 0 (Not at all) to 4 (Very frequently), yielding total scores from 0 to 48. SARQ scores were collected both three minutes and 24 h after the speech task. Internal consistency of the SARQ both immediately following the speech (Cronbach’s alpha = 0.94) and 24 h later (Cronbach’s alpha = 0.95) was excellent.
State Anxiety. Consistent with prior research assessing subjective emotional reactivity to social stressors (Çek et al., 2016), participants rated their current level of anxiety on a visual analogue scale from 0 (not at all) to 100 (very much). State anxiety ratings were collected both three minutes and 24 h after the speech task to assess for immediate and sustained effects of the attention manipulation.
Focus of Attention Questionnaire (FAQ; Woody, 1996). The FAQ is a self-report measure of attentional foci during a social task consisting of two five-item subscales: (1) self-focused attention (i.e., attending to one’s own behavior, anxiety, or impression), and (2) other-focused attention (i.e., attending to one’s social partner or environment). Participants rate items (e.g., “I was focusing on my level of anxiety”) on a five-point Likert scale ranging from 1 (not at all) to 5 (totally). The FAQ was administered following the speech task as a manipulation check. Internal consistency of the FAQ-Self subscale was good (Cronbach’s alpha = 0.76), and the FAQ-Other subscale was poor (Cronbach’s alpha = 0.52). However, it is worth noting that the FAQ- Other subscale is limited by measuring focus on one’s social partner and one’s environment together, whereas the External Threat and Control groups were instructed to differentially attend to their partner or their environment, respectively. Accordingly, the poor internal consistency of the FAQ-Other subscale may be due in part to the effect of the experimental manipulation.
Procedure
The university’s Institutional Review Board approved all aspects of the study. To identify high socially anxious participants, individuals aged 18 or older who scored above 39 on the Brief Fear of Negative Evaluation Scale (BFNE; Leary, 1983) were eligible to participate. A cut-off score of 39 was chosen on the basis of prior studies using the BFNE to identify high socially anxious samples (e.g., George & Stopa, 2008; Makkar & Grisham, 2013a). Participants were recruited via two channels. First, undergraduate students at a private university in the southeastern United States completed the BFNE as part of a broader screening effort in classes offering course credit for research participation. Second, community participants who opted into a local research opportunity listserv were sent study information and a link to complete the BFNE for screening via a mass distribution email. Eligible participants were contacted via email and invited to participate for either class credit or a $25 gift card. Participants were assessed at two time points. During session one, participants came into the laboratory and were briefly interviewed by a trained research assistant using the Social Phobia module of the MINI to assess for a diagnosis of SAD. Then, participants completed the study questionnaires on a laboratory computer. The research assistant then introduced the speech task, which was chosen as the social stressor on the basis of prior research showing that speeches are strong elicitors of PEP in highly socially anxious individuals (Kiko et al., 2012). Consistent with methods employed by Çek et al. (2016), participants were asked to give a three-minute speech while being video recorded and assessed by the experimenter. Participants then waited for one minute before they were given the topic of their speech (“Have music and TV become too explicit? Argue for or against.”). Participants were randomized into one of three attentional conditions: Internal, External, or Control. Depending on their condition, they received one of the following attention manipulations, adapted from Gaydukevych and Kocovski (2012; see Supplement for full attention manipulation instructions):
Internal Condition
Participants were instructed to pay close attention to their thoughts, actions, and body sensations during the speech. Since feedback about physiological arousal has been shown to increase self-focused attention (Makkar & Grisham, 2013; Wells & Papageorgiou, 2001), participants in this condition were set up with a portable heart rate monitor and told that their electrocardiogram (ECG) results would be displayed to them during the speech to help them focus on their internal sensations. In reality, all participants viewed the same sham heart rate video displaying ECG waves, corresponding beats per minute (BPM), and a very soft beeping noise. The video showed a fluctuating but gradually increasing and then modestly decreasing heartrate, ranging from 72 to 105 BPM. Importantly, the displayed heart rate was fluctuant to provide a mix of threatening and nonthreatening cues, parallel to the External condition. The video was positioned slightly to the right of participants’ line of sight and the volume was set to be barely audible so as to not be a distractor, but rather an available stimulus for sensorily anchoring attention on interoceptive information. Research assistants began showing the sham heart rate readings just before the participants began their speech.
External Condition
Participants were instructed to pay close attention to the audience member’s reactions and evaluations of them during the speech, taking note of their evaluator’s body language, facial expressions, and behaviors to generate an estimate of the evaluator’s assessment of the participant. Participants in this condition were asked to provide their best guess of the evaluator’s assessment of them at the end of the speech (from 0 to 10) to help focus their attention on the evaluator and increase the salience of the social-evaluative threat present in the environment. Importantly, the evaluator gave no explicit feedback as to their actual evaluation of the participant’s performance.
Control Condition
Participants were instructed to pay close attention to a neutral object (a faux pumpkin) placed in participants’ line of sight during the speech. This manipulation was intended to control for the visual attentional demands of the other conditions, but to not induce any particular attentional focus towards threat.
In each condition, after receiving their instructions, participants were given a notecard and asked to write down a few words to help them remember what they were meant to focus on during their speech. Participants were able to hold this card as a memory aid during their speech. The research assistant then turned on the video camera and asked participants to begin their speech. Research assistants followed the same behavioral script during each speech regardless of condition, which contained a blend of positive, negative, and ambiguous evaluative cues such as nodding, frowning, and note-taking (see Supplement for full behavioral protocol). Thus, the objective level of social threat was held constant across conditions, while the attentional foci varied across conditions. After three minutes, the research assistant asked the participant to stop and removed the heart rate monitor (if applicable). Participants then waited for three minutes before completing the SARQ, State Anxiety rating, and FAQ. This concluded session one.
24 h after session one, participants were emailed a link to complete the SARQ, State Anxiety rating, and (Internal condition only) a heart rate manipulation check remotely. The heart rate manipulation check consisted of three questions regarding to what extent participants believed their heart rate measurement was accurate (“1. How much did you believe that your heart rate was being accurately measured by the equipment? 2. How much did you believe that the heart rate information transferred to the computer was correct and accurate? 3. When you were told that your heart rate increased, to what extent did you believe this to be true?”; Makkar & Grisham, 2013) on a scale from 0 (not at all) to 9 (completely), yielding total scores from 0 to 27. Each email contained an image of the participant at the start of their speech in order to facilitate retrieval of their memory of the stressor and any corresponding anxious affect or thoughts experienced since (Ucros, 1989). Participants then received a debriefing message describing the purpose of the study and clarifying that the heart rate results they viewed were false (if applicable). Participants were compensated via class credit or a $25 gift card. All analyses were conducted in SPSS 28. Of the 123 participants enrolled, one participant dropped out prior to the speech task, and six did not provide complete follow-up data. Missingness was not significantly associated with experimental condition (p > .05). Table 1 indicates where restricted samples were used in analyses due to missing data. The datasets generated during the current study are available from the corresponding author on reasonable request.
Table 1
Group Differences in Measures of Attentional Focus, anxiety, and PEP
Internal (n = 41) | External (n = 41) | Control (n = 40) | F | p | |
|---|---|---|---|---|---|
FAQ-Self | 16.49 (3.65)a | 13.78 (4.29)b | 14.13 (4.29)b | 5.17 | 0.007 |
FAQ-Other | 7.95 (2.34)a | 9.73 (2.78)b | 8.00 (2.82)a | 5.95 | 0.003 |
Anxiety (Immediate)1 | 54.70 (26.63)a | 38.39 (28.88)b | 43.18 (29.29)a, b | 3.54 | 0.032 |
Anxiety (24 h) 2 | 30.33 (27.53) | 25.51 (24.24) | 24.31 (23.81) | 0.62 | 0.539 |
SARQ (Immediate) | 23.54 (12.91) | 19.34 (12.67) | 19.55 (12.13) | 1.44 | 0.240 |
SARQ (24 h)2 | 18.26 (12.57)a | 12.03 (11.83)b | 14.79 (10.68)a, b | 2.77 | 0.067 |
Results
Baseline Measures
We conducted univariate Analysis of Variance (ANOVA) and chi-square tests to examine baseline differences between conditions on demographic and individual difference variables (see Table 2). There were no significant differences between conditions on any variable except gender, such that there were significantly more males in the External condition. Because gender was not significantly associated with the primary outcome measures (ps > 0.40) and prior theory does not suggest that gender would significantly influence levels of the outcome variables (Clark & Wells, 1995), we did not control for gender in subsequent analyses (de Boer et al., 2015). However, when gender was included as a covariate, the results remained unchanged. Eight percent of the sample met diagnostic criteria for SAD as assessed by the MINI, and this proportion did not significantly differ across the three conditions.
Table 2
Sample demographics and descriptive statistics of baseline measures
Internal (n = 41) | External (n = 42) | Control (n = 40) | F or χ2 | p | |
|---|---|---|---|---|---|
Age | 24.80 (9.70) | 26.67 (13.55) | 23.23 (7.50) | 1.08 | 0.342 |
Gender | 12.11 | 0.002 | |||
Female | 34 (82.92) | 26 (61.90) | 37 (92.50) | ||
Male | 7 (17.07) | 16 (38.10) | 3 (7.50) | ||
Race | |||||
White | 26 (63.41) | 31 (73.81) | 26 (65.00) | 1.19 | 0.552 |
Asian | 11 (26.83) | 11 (26.19) | 11 (27.50) | 0.02 | 0.991 |
Black | 7 (17.07) | 3 (7.14) | 6 (15.00) | 2.02 | 0.365 |
Other | 1 (2.44) | 0 (0.00) | 0 (0.00) | 2.02 | 0.365 |
Ethnicity | 4.16 | 0.125 | |||
Hispanic | 4 (9.76) | 2 (4.76) | 0 (0.00) | ||
Non-Hispanic | 37 (90.24) | 40 (95.24) | 40 (100.00) | ||
Participant Type | 1.12 | 0.571 | |||
Student | 28 (68.3) | 27 (64.3) | 30 (75.00) | ||
Community | 13 (31.7) | 15 (35.7) | 10 (25.00) | ||
SPS | 24.71 (13.27) | 20.40 (13.09) | 21.68 (12.96) | 1.18 | 0.312 |
SAFE | 37.61 (18.10) | 35.10 (20.36) | 37.00 (19.98) | 0.19 | 0.829 |
ASI | 9.78 (5.12) | 9.19 (6.10) | 9.03 (5.75) | 0.20 | 0.819 |
DASS | 35.37 (21.74) | 32.81 (21.91) | 33.85 (19.96) | 0.15 | 0.859 |
Manipulation Checks
Within the Internal condition, responses to the heart rate manipulation check questions indicated that participants largely believed the accuracy of the sham heart rate feedback that they viewed, M = 21.82, SD = 5.45, with scores ranging from 3 to 27. Since no participants scored a 0, indicating complete disbelief in the heart rate feedback, all available outcome data for participants randomized to the Internal condition was used.
We tested univariate ANOVAs on the FAQ Self and Other subscales to assess for the success of the attentional conditions in inducing their intended attentional foci. As shown in Table 1, the ANOVAs revealed a significant effect of group on FAQ-Self [F(2,119) = 5.17, p = .007] and FAQ-Other [F(2,119) = 5.95, p = .003]. Follow-up tests revealed that the Internal condition scored significantly higher than the External condition (Mean Difference = 2.71, SE = 0.92, p = .004) and the Control condition (Mean Difference = 2.36, SE = 0.92, p = .012) on the FAQ-Self. Further, the External condition scored significantly higher than the Internal condition (Mean Difference = 1.78, SE = 0.59, p = .003) and the Control condition (Mean Difference = 1.73, SE = 0.59, p = .004) on the FAQ-Other. This suggests that the attention manipulation was successful in inducing significantly higher self-focused attention in the Internal condition, and other-focused attention in the External condition.
Effects of Attention Condition on Anxiety and PEP
We tested univariate ANOVAs to examine the effect of experimental condition on State Anxiety and SARQ scores collected both immediately following the speech task and 24-hours later. Results are displayed in Table 1. We found a significant group effect for immediate state anxiety [F(2,118) = 3.54, p = .032]. Follow up tests revealed that the Internal condition reported significantly higher state anxiety immediately following the speech task than the External condition (Mean Difference = 16.31, SE = 6.29, p = .011), but no other group comparisons were significant. We also found a marginally significant group effect for SARQ 24 h post-speech [F(2,114) = 2.77, p = .067]. Since our primary hypothesis concerned the comparison of PEP between the Internal and External conditions, and planned contrasts are a more appropriate test of a priori predictions than ANOVAs (which have reduced power due to the inclusion of all potential contrasts and interactions; Wilkinson, 1999), we proceeded to test the pairwise comparisons of 24 h later SARQ scores. Pairwise contrasts revealed that the Internal condition had significantly higher SARQ scores 24 h post-speech than the External condition (Mean Difference = 6.23, SE = 2.65, p = .021), but no other group comparisons were significant.
Mediating Role of State Anxiety
Based on the observed pattern of group differences only between the Internal and External conditions, we included only these two conditions in our primary mediation model to preserve power and simplicity, yielding an n = 77 (however, see the Supplement for the multicategorical mediation model including all three conditions). Using the PROCESS macro for SPSS (Hayes, 2017), we tested whether immediate state anxiety mediated the link between condition (i.e., Internal vs. External) and SARQ scores 24 h later. As shown in Fig. 1, condition significantly predicted subsequent SARQ via effects on immediate state anxiety, after controlling for gender. A 95% bias-corrected bootstrap (n = 5,000) confidence interval (CI) for the indirect effect of state anxiety did not contain zero (95% CI: 0.62, 8.53; estimate = 4.40; standard error = 1.96), suggesting significant mediation effects. We estimated the size of this indirect effect by computing the ratio of the indirect effect to the total effect (Alwin & Hauser, 1975). The indirect effect of state anxiety explained 68.50% of the total effect of condition on subsequent PEP.
Fig. 1
Mediational model of the relation between condition, immediate state anxiety, and delayed Socially Anxious Rumination Questionnaire (SARQ) scores. Path c’ represents the direct effect of condition on SARQ
×
![]()
Discussion
The present study aimed to clarify the role of internally and externally focused attention during social interactions in the maintenance of SAD. We found that internally focused attention during a social stressor caused significantly higher acute anxiety and subsequent PEP than externally focused attention. We also found that the effects of attention (i.e., internally vs. externally focused) on PEP the next day were fully mediated by differences in state anxiety immediately after the speech task. The observed pattern of results is consistent with prior research demonstrating a link between self-focused attention, state anxiety, and PEP (Gaydukevych & Kocovski, 2012; Makkar & Grisham, 2013; Norton & Abbott, 2016; Sluis et al., 2017; Woody, 1996; Woody & Rodriguez, 2000). Findings are also in line with cognitive-behavioral models of SAD, which posit that self-focused attention is a key mechanism involved in disorder maintenance (Clark & Wells, 1995; Hofmann, 2007; Rapee & Heimberg, 1997). An excessive internal focus may have contributed to elevated state anxiety and subsequent PEP in a number of ways.
Highly socially anxious individuals are prone to thorough self-focused observation in social situations, with particular attentional vigilance towards negative thoughts, feelings, and interoceptive sensations that signal the potential for negative evaluation (Clark & Wells, 1995; Hofmann, 2007; Norton & Abbott, 2016; Rapee & Heimberg, 1997; Spurr & Stopa, 2002). Such attentional allocation towards internal threats may set off a series of effects that are temporally sequenced and disorder-maintaining. The self-focused attention induced by the Internal condition may have contributed to a greater awareness of one’s own anxiety and accordingly more negative evaluations of the self, thus heightening the sense of immediate social danger (Clark & Wells, 1995; Hofmann, 2007; Rapee & Heimberg, 1997). Indeed, the anxiety sensitivity literature has demonstrated that anxious individuals tend to make inferences about the present level of danger based on their own anxiety responses, and interpret symptoms of arousal themselves as threatening (Arntz et al., 1995; Mansell & Clark, 1999; Olatunji & Wolitzky-Taylor, 2009; Reiss et al., 1986). In the short term, such elevated perceptions of threat may have contributed to the observed spike in state anxiety, thus increasing the overall salience of the speech task in memory (Makkar & Grisham, 2013; Norton & Abbott, 2016; Woody, 1996; Woody & Rodriguez, 2000). In the longer term, such salient memories are more likely to be the subject of PEP, which may be more negatively distorted due to the selective attention paid towards threatening internal information during the social event (Abbott & Rapee, 2004; Clark & Wells, 1995; Gaydukevych & Kocovski, 2012; Hofmann, 2007; Kiko et al., 2012; Perini et al., 2006; Rapee & Heimberg, 1997). The temporal nature of the present findings, whereby internally focused attention caused elevated immediate, but not sustained, state anxiety and elevated delayed, but not immediate, PEP, supports this theorized role of self-focused attention. It is encouraging that self-focused attention was also found to cause increased PEP following a conversation task (Gaydukevych & Kocovski, 2012), supporting the generalizability of the present findings across different types of social situations.
Alternatively, it is possible that the heart rate feedback used to induce self-focused attention in the Internal condition’s attention manipulation was more anxiety-inducing than the other conditions (Makkar & Grisham, 2013; Papageorgiou & Wells, 2002; Wells & Papageorgiou, 2001; Wild et al., 2008). However, several steps were taken to mitigate this risk, including the placement of the heart rate video outside of participants’ direct line of sight, the use of a barely audible beeping sound, and the inclusion of both increases and decreases in heart rate in the video, thus displaying both threatening and nonthreatening cues. Importantly, prior research testing the effects of false heart rate feedback on negative affect and PEP found that self-focused attention fully mediated the effects of the heart rate feedback on outcomes (Makkar & Grisham, 2013), tempering concerns about confounding effects of the heart rate feedback. However, replication of the present study using a different paradigm to induce self-focused attention would strengthen confidence in our findings.
Contrary to prediction, neither the Internal nor External conditions reported significantly higher state anxiety or PEP than the Control condition at either time point. This may have been due to the failure of the Control condition to successfully invoke an affectively neutral attentional state. Since the recruited sample was highly socially anxious (i.e., BFNE scores > 39), and such individuals have been shown to display ABs towards threat (Amir & Foa, 2001; Gilboa-Schechtman et al., 1999; Hope et al., 1989; Mogg & Bradley, 2002; Norton & Abbott, 2016; Pishyar et al., 2004; Spurr & Stopa, 2002), the Control condition’s attentional anchor (i.e., a faux pumpkin) may not have been engaging enough to override these preexisting tendencies towards attentional vigilance towards threats. Indeed, the Control group’s scores on the FAQ Self and Other subscales were in between those of the Internal and External conditions, indicating that the Control condition’s attention manipulation did not function as intended to completely disengage participants’ attention from sources of threat. Accordingly, it is unclear what participants in this condition were consistently paying attention to and thus we cannot draw firm conclusions about the relative effects of attentional allocation towards threats versus non-threats.
The finding that internally focused attention caused significantly higher state anxiety and subsequent PEP than externally focused attention does have important theoretical implications. Rapee and Heimberg’s (1997) model of SAD asserts that socially anxious individuals simultaneously attend to internal cues and external indicators of negative evaluation, and that both of these attentional foci interact to inform individuals’ mental representation of themselves in social situations and thus contribute to the cascade of somatic, cognitive, and behavioral symptoms of anxiety. Prior research does support the notion that ABs towards both threat types are present in SAD (Amir & Foa, 2001; Gilboa-Schechtman et al., 1999; Hope et al., 1989; Mogg & Bradley, 2002; Norton & Abbott, 2016; Pishyar et al., 2004; Spurr & Stopa, 2002), but our results contend with the idea that both types of attentional foci have equally harmful effects. It may be the case that attentional allocation towards external threats is harmful only insofar as it triggers internally focused attention, though this prediction needs to be more thoroughly tested.
It is important to note that paying attention to external sources of threat may actually be adaptive in that it facilitates a new form of learning. When socially anxious individuals enter feared social situations, they tend to shift their attention inward to actively monitor performance (Hofmann, 2007; Salkovskis, 1991; Wells et al., 1995) and erroneously believe their negative internal experiences reflect the way they are perceived by others (McEwan & Devins, 1983), increasing the perceived likelihood of rejection or humiliation. An internal attentional focus may distract socially anxious individuals from corrective cues in the environment, whereas an external focus may direct an individual’s attention to information that can disconfirm their feared beliefs. In the present study, participants in the External condition attended to an audience member’s body language, facial expressions, and behaviors to generate an estimate of the evaluator’s assessment. It may be the case that the external allocation of attention facilitated a violation of expectancy (e.g., “the audience member did not reject me, instead they occasionally smiled and nodded during the speech”). Such expectancy violations are believed to be a mechanism of change in exposure-based treatment for anxiety disorders, including social anxiety disorder (Craske et al., 2014). Indeed, increased attention to the feared stimulus during exposure (e.g., the evaluator during a speech) is believed to enhance learning (Craske et al., 2022). Though the empirical literature in this area is limited, Wells and Papageorgiou (1998) have shown that exposure therapy that is combined with instructions to focus on the external environment is more effective than standard exposure treatment. Thus, an important area for future research will be to examine whether instructions to attend to external threats can effectively facilitate adaptive learning through increased violation of expectancy.
In light of these findings, the present study may have important clinical implications. In the past few decades, a wave of attention bias modification (ABM) interventions has emerged in the treatment of a broad range of disorders, including SAD (Mogg & Bradley, 2016). These interventions aim to curb disorder-specific ABs by systematically training attention away from threatening stimuli and towards neutral or positive stimuli (Mogg & Bradley, 2016). In SAD, ABMs have demonstrated some efficacy at curbing acute symptom severity, but findings are highly heterogeneous, effects are small, and treatment gains are not sustained over time (Heeren et al., 2015; Mogg & Bradley, 2016). Most ABMs for SAD use facial stimuli (e.g., disgusted, angry, neutral, or happy faces; Heeren et al., 2015), thereby aiming to attenuate ABs towards external, but not internal, threats. The present findings suggest that this treatment target may be misdirected, as internally focused ABs may be most harmful. Additionally, interoceptive exposure has been shown to significantly reduce the distress associated with experiencing somatic symptoms of anxiety across a range of disorders (Boettcher et al., 2016; Boswell et al., 2013), thus it may be particularly useful in interrupting the escalating cycle of anxiety triggered by self-focused attention amongst individuals with SAD (Collimore & Asmundson, 2014; Dixon et al., 2015). Future research is needed to examine the efficacy of such interventions in socially anxious individuals with elevated self-focused attention, as well as to explore their additive benefit to existing first-line cognitive-behavioral treatments for SAD (Heimberg, 2002).
Results of the present study must be interpreted in the context of several limitations. First, we employed an analogue sample of highly socially anxious individuals amongst whom only a minority met criteria for SAD. Although research suggests that analogue studies are a valid method for exploring mechanisms involved in SAD (Stopa & Clark, 2001), replication of the present study in a clinical sample is warranted. Further, the Control condition’s attention manipulation may have been ineffective in capturing participants’ attention and fully eliminating attentional allocation towards internal and external threats. Future research employing a more engaging, but affectively neutral, attentional anchor would allow for better evaluation of the effects of attentional allocation towards different threat types compared to a neutral attentional control. Additionally, the present study did not vary the level of social-evaluative threat present across conditions (i.e., researchers reacted identically towards all participants) so that the effects of different attentional foci could be isolated. As a next step, to examine if the beneficial effects of externally focused attention hold even when the external environment is highly threatening, future research may replicate the present study with the addition of high vs. low levels of social threat conditions (i.e., overtly disapproving vs. approving researcher reactions) to probe the effects of internally vs. externally oriented attention as a function of actual social threat level. Relatedly, we did not record participants’ estimates of how the researcher evaluated them (from 0 to 10) following the speech task in the External condition, which would have provided a useful indicator of the level of perceived threat present in the external environment. Moreover, the present study examined only the effects of sustained attentional allocation towards different threat types, but ABs are not unitary constructs. ABs may consist of facilitated attention/hypervigilance, difficulty disengaging, and attentional avoidance of threats (Cisler & Koster, 2010). Future research is needed to parse out the differential effects of biases at these different stages of attentional processing. Finally, the study relied exclusively on self-report measures of outcomes at only two time points. Incorporating physiological measures of anxiety, such as skin conductance response, and sampling outcomes at additional timepoints over a longer assessment period would have allowed for more objective measurement of study variables and exploration into the duration of effects.
Acknowledgements
We extend our sincere appreciation to all the participants who contributed their time to this study. We also thank the research assistants (Ashley Boyne, Stella Wang, Kavi Jakes, Shannon Kim, Sophia Mueller, and Elizabeth Woods) who played an integral role in data collection and the overall execution of this study.
Declarations
Conflict of Interest
The authors have no conflict of interest to declare.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
