Critically, there was a highly significant interaction of Emotion-relevance × Action-relevance that survived our stringent cluster correction in a late time-window from 650 to 790 ms (t 8 = 3.15, P = 0.018, Corr), and that reflected distinct response patterns to the FEAR and NEU faces (Fig. 3 c–f).
5 a), whereas the CIRCLE condition showed only a marginal trend ( P = 0.06, Unc, around 600 ms post-onset; Fig. 5 b).
Nevertheless, this limitation prevents us from reaching firm conclusions regarding some of our iERP results that were marginally significant (uncorrected level) and regarding behavioral results that were statistically non-significant, given that a null effect does not equate an absence of effect.
This lack of reliable interaction between emotion- and action-relevance for the early amygdala activity may suggest that some emotional processing could occur to some degree during this time-window, regardless of the task goals, even though the direct comparison of fearful and neutral faces in the non-target condition failed to reach significance when considered alone (see Fig. 3 c vs. e).