The effect of stimulus set size on the sex-emotion interaction in face categorisation

Previous research has shown that invariant facial features, e.g., sex, and variant facial features, e.g., emotional expressions, interact during face categorisation. The nature of this interaction is unclear, however, and has been reported as either asymmetrical, invariant features affect emotion perception whereas emotion does not affect perception of invariant features, or symmetrical, invariant features affect perception of variant
features and vice versa. The current research investigated the source of these inconsistent results and identified the number of different posers used as the critical variable. Using faces drawn from two different databases, Experiments 1 and 2 replicated the finding of a symmetrical interaction between face sex and emotional expression when 32 different posers were used. Using subsets of four posers in the same task setup, however, did not
provide evidence for a symmetrical interaction also consistent with prior research. The same pattern of results emerged using the Garner paradigm (Experiments 3 and 4), which, with small stimulus sets, previously had yielded only evidence for an
asymmetrical interaction. This pattern of results seems to suggest that faces are processed to a lower level of detail if they are encountered repeatedly preventing interference from features that are not task relevant.