Social anxiety (SA) is a common, chronic, and debilitating condition characterized by persistent anxiety of social situations in which there is a potential for scrutiny by others (APA, 2013). Even at subclinical levels, SA is linked to intra- and interpersonal difficulties (Fehm et al., 2008). SA is associated with maladaptive beliefs about the self and others (Gilboa-Schechtman et al., 2017; Zabag et al., 2018). Recent theories and research suggest that these maladaptive beliefs in SA persist due to inflexible belief updating (Everaert et al., 2018; Kashdan and Rottenberg, 2010). Indeed, SA was associated with difficulty in positively updating negative beliefs about the self. For example, individuals high in SA struggle to update their negative self-evaluation after receiving positive feedback (Button et al., 2015). Similarly, SA is associated with a deficit in using positive information to update negative interpretations of social scenarios (Everaert et al., 2018). SA is also related to deficits in updating beliefs about others— in a computerized ball-catching game, SA was associated with enhanced avoidance (fewer throws) to previously punishing avatars (Beltzer et al., 2019). In other studies, participants first learned that stimuli (facial images) were associated with either a positive or negative outcome. Later, the outcomes were revised, such that participants needed to change their beliefs about stimuli-outcome associations. It was found that SA was associated with difficulties in negative-to-positive updating: High-SA participants found it difficult to learn that social stimuli, which previously had to be avoided, should now be approached, but not vice-versa (Zabag et al., 2023). This difficulty was specific to social stimuli and absent in tasks with non-social stimuli (Zabag et al., 2022).

Tasks that are commonly used to study learning and updating biases in SA provide only "factual-information"—when participants receive partial feedback about their chosen behavior and the consequences of that choice (Palminteri and Lebreton, 2022). However, in real-life, people often learn and update their beliefs using more abstract, inferential processes, including learning from forgone outcomes (counterfactual-information) (Coricelli and Rustichini, 2010). Indeed, previous research suggests that counterfactual-information facilitates performance and that learning and updating are enhanced when complete information is provided, about both their decision and the alternative decision that was not implemented (Salvador et al., 2017). Moreover, neuroimaging studies have demonstrated differential brain activation when factual and counterfactual-information is encountered (Boorman et al., 2011; Tobia et al., 2014). In SA, offering counterfactual-information about the consequences of avoidance, a core mechanism of the disorder, may alleviate avoidance tendencies. Exploring the impact of counterfactual-information on updating biases in SA is the main goal of the present study.

To this end, we created a dynamic social learning task in which information about both factual (chosen) and counterfactual (unchosen) decisions is provided. As in former studies (Zabag et al., 2022, 2023), participants chose whether to "approach" or "avoid" a social stimulus (a person). In contrast to previous studies, in the present task, participants received information about the results of their decision to either approach or avoid a stimulus. We explored whether SA will be associated with selective positive updating biases even under these learning-enhancing conditions.