Time-course of attention biases in social phobia

Highlights

• Fine-grained analysis of time-course of attention to social threat in patients with social phobia using probe-detection task.

• Results did not reveal between-group differences in vigilance (initial attention allocated to threat).

• Results suggest that patients with social phobia attended to emotional faces less over time, and specifically less to positive expressions.

• Relative attention to emotional versus neutral faces suggests that controls (but not patients) preferentially attend to positive and avoid negative.

• Findings suggest that attention “bias” in social phobia may be driven by a relative lack of bias (as compared to non-anxious controls).

Abstract

Theoretical models of social phobia implicate preferential attention to social threat in the maintenance of anxiety symptoms, though there has been limited work characterizing the nature of these biases over time. The current study utilized eye-movement data to examine the time-course of visual attention over 1500 ms trials of a probe detection task. Nineteen participants with a primary diagnosis of social phobia based on DSM-IV criteria and 20 non-clinical controls completed this task with angry, fearful, and happy face trials. Overt visual attention to the emotional and neutral faces was measured in 50 ms segments across the trial. Over time, participants with social phobia attend less to emotional faces and specifically less to happy faces compared to controls. Further, attention to emotional relative to neutral expressions did not vary notably by emotion for participants with social phobia, but control participants showed a pattern after 1000 ms in which over time they preferentially attended to happy expressions and avoided negative expressions. Findings highlight the importance of considering attention biases to positive stimuli as well as the pattern of attention between groups. These results suggest that attention “bias” in social phobia may be driven by a relative lack of the biases seen in non-anxious participants.

Introduction

Social phobia is one of the most common psychiatric disorders, with lifetime prevalence rates estimated as high as 12% (Kessler et al., 2005). In addition, social phobia is associated with substantial psychosocial stress and elevated risk for suicidality (Sareen et al., 2005, Stein and Kean, 2000). Thus, significant efforts have focused on elucidating causal and maintaining factors of social phobia. To this end, one promising area has been the accumulation of evidence that socially anxious individuals are biased to preferentially attend to threat information. Indeed, biased attention is implicated in the major theoretical models of social phobia (Clark and Wells, 1995, Heimberg et al., 2010, Rapee and Heimberg, 1997). These models converge on the prediction that biased processing of social cues plays a role in the maintenance of elevated social anxiety; however, these models diverge in their predictions about how attention unfolds over time. For example, Clark and Wells (1995) propose that individuals with social phobia avoid focusing attention on threat signals. An alternative theory claims that individuals with social phobia are vigilant to detect threat cues and furthermore, once threat signals are detected socially anxious individuals have difficulty disengaging attention from threat cues (Heimberg et al., 2010, Rapee and Heimberg, 1997).

In the anxiety literature broadly, experimental evidence has generally supported the claims that anxious individuals (both clinical and non-clinical) demonstrate preferential attention for threat cues (see Bar-Haim, Lamy, Pergamin, Bakerrmans-Kranenburg, & van IJzendoorn, 2006 for a meta-analysis). Importantly, this meta-analysis concludes that considering attention as a unitary construct is insufficient in explaining the nature of bias toward threat. Specifically, the importance of evaluating stages of processing is emphasized. That is, previous work relying on reaction time indices to measure attention biases fall short in establishing what subsystems of attention account for biased attention. This distinction is important because attention bias can be driven by different components of attention (i.e., facilitated attention or vigilance, avoidance, or difficulty disengaging attention). As has been discussed elsewhere (cf. Cisler & Koster, 2010), these different components may have different biological mechanisms and, given the vast literature on attentional biases in anxiety “the next step in attentional bias research is to elucidate how and why attention is biased toward threat” (p. 5).

Historically, the bulk of research evaluating attention biases in clinical populations has relied on reaction time indices. Reaction-time based research has been important to early evaluations of attention biases, and more recently this approach has attempted to delineate components of attention bias by varying stimulus onset time (e.g., Mogg et al., 2004a, Mogg et al., 2004b, Moriya and Tanno, 2011) or reconsidering reaction-time indices to allow for the distinction between vigilance and disengagement (e.g., Koster et al., 2004, Salemink et al., 2007). Of note, the presence of biases at different stages of attentional processing has varied theoretical (and treatment) implications. For example, attention biases characterized by hypervigilance to the detection of threat suggest patients may overestimate the likelihood of threat in their environment, whereas sustained attention on threat (i.e., maintenance) suggests the ongoing allocation of resources to possible threat in one's environment, potentially at the cost of more adaptive processing. Importantly, these strategies are not necessarily mutually exclusive of one another (see Cisler et al., 2009, Weierich et al., 2008), yet reaction-time data is limited in its ability to capture the nuances of dynamic attention. For this reason, the field has increasingly shifted toward the assessment of attention biases measuring eye movements and fixations throughout the course of experimental tasks.

This transition toward the measurement of attention in clinical populations via the tracking of eye-movements is highlighted by a recent meta-analysis of this research methodology in the affective disorders (Armstrong & Olatunji, 2012). This paper evaluated 33 experiments and results indicated that that the extant eye-tracking literature largely supports the theoretical claim that anxious participants demonstrate initial orienting biases toward threatening stimuli. Further, although results related to maintenance of attention were more tentative given less consistency in the findings, there is less support for maintained attention to (i.e., difficulty disengaging from) threatening stimuli in anxious populations. Since this meta-analysis, one additional study has contributed to the eye-tracking literature evaluating patients with social phobia and report that socially anxious participants disengaged from positive stimuli more readily than negative stimuli (Chen, Clark, MacLeod, & Guastella, 2012).

Of note, the momentum this work has gained in recent years has contributed significantly to our understanding of the time-course of attention biases among the anxiety disorders. Yet, variability in biases found in attention across time in these studies highlights the relevance of fine-grained sampling of visual attention. For example, the bulk of previous research collapsed metrics of attention across time into samples of 500 ms or longer. These segments may have obscured nuances in eye-movement data given that as many as 4 fixations can be expected per second (Rayner, 1998). Thus, evaluating attention in smaller segments over time may provide a more refined measurement of the time-course of attention. Further, as previous research has noted, the attentional processes of vigilance and maintenance are not mutually exclusive (Armstrong and Olatunji, 2012, Weierich et al., 2008). Refined time-course measurement may enable a delineation of both vigilance and maintenance within the same task (i.e., at what temporal stage one strategy is employed versus another).

In summary, further clarification of the time-course of visual attention in anxious individuals is warranted. Inconsistencies in previous studies could be due to sampling differences, variation in experimental tasks, and analytic approaches. The current study contributes to the previous research by providing an assessment of eye-movements in a clinical population while they complete the frequently cited attention bias assessment task–the emotional probe detection task. We aimed to conduct a fine-grained analysis of the time-course of eye movements; thus, attention was measured during very brief samples (i.e., every 50 ms). The goal of the current study was to evaluate the time course of attention to emotional and neutral faces in patients with social phobia. We hypothesized that, consistent with results from a recent meta-analysis and the claims of the theoretical models (Heimberg et al., 2010, Rapee and Heimberg, 1997), patients with social phobia would be initially vigilant to direct attention toward negative emotional expressions. Given less consistency in the literature regarding maintenance of attention on threat, we hypothesized in accordance with theoretical models that subsequent to initial vigilance, patients with social phobia would maintain attention on threat for a longer period of time relative to non-anxious participants.