Cognitive-behavioral theories of emotional disorders assert that emotional responding comprises concordant responses (e.g., co-occurring subjective distress and physiological arousal), but this is often not observed in practice. We investigated whether concordance would be greater when social threat is higher. A sample of 46 socially anxious participants underwent experiences involving varying degrees of social interaction and evaluative threat. Affect, cognitions, behaviors, and physiological responses were assessed repeatedly. We used a network modeling approach to examine the associations among emotion response components that are typically associated with trait social anxiety. After estimating contemporaneous partial-𝜏 networks for each condition, we identified the optimal modularity statistic with absolute thresholding. Permutation testing was used to investigate hypotheses tied to absolute and relative concordance. Concordance was defined as a lack of community structure as measured by a non-significant modularity statistic, indicating a unidimensional anxiety response. Contrary to hypotheses, absolute concordance was not observed in any of the conditions involving social-evaluative threat but was observed in the non-social control network. Additionally, no significant differences in relative concordance emerged when comparing explicit evaluation vs. non-explicit evaluation or social vs. non-social networks. Our findings align with the extant literature suggesting that concordance is not a necessary or common feature of emotional episodes and have implications for our theoretical understanding of social anxiety. Future work should explore individual differences in concordance and interactions among components across different timescales, as well as the replicability of our findings in larger, more diverse samples.