Are Effects of Emotion Expression on Trait Impressions Mediated by Babyfaceness? Evidence From Connectionist Modeling

1. We did not analyze activation of the adult unit because it is a reciprocal of the baby unit activation.

2. For the interested reader, we note the following qualifications to the emotion main effects by sex of face: In Study 1, the manipulation check ratings of anger and surprise were stronger for female faces, although highly significant for both sexes; the tendency for surprise expressions to elicit more activation of the eye-region baby unit than neutral expressions was stronger for female than male faces, with the latter marginally significant; the tendency for anger expressions to be less attractive than neutral ones was significant only for female faces; and the tendency for happy expressions to be more attractive than neutral ones was stronger for female than male faces, with the latter marginally significant. In Study 2, female faces were rated significantly lower in dominance than male faces in the angry and happy models and significantly higher in affiliation in the surprise and happy models. Although there was a significant Face Sex × Emotion interaction for perceived affiliation in the anger model, the lower perceived affiliation of anger than neutral expressions was highly significant for both sexes. More details about these results are available from the first author.

3. Additional findings of interest were that surprise expressions were rated more happy than angry, and happy expressions were rated more surprised than angry, which is consistent with the fact that surprise can be a precursor to happiness. Also, although neutral faces were rated lowest of all expressions in surprise and happiness, they were rated second only to anger faces in anger, which is consistent with other evidence concerning similar reactions to neutral and anger expressions (e.g., Vrana, 2004).

4. Although Montepare and Dobish (2003) found that surprise expressions elicited impressions of high power in contrast to the lower power perceived in fear expressions, they attributed this result to a confusion between surprise and happy expressions rather than between surprise and fear expressions in their study. Given our Study 1 results, we expected surprise to yield impressions similar to fear expressions in Study 2.

5. In the interest of brevity, we have not reported results for ratings of intelligence and health for which we had no a priori predictions because these traits had not been included in the dominance or affiliation composites in previous research investigating impressions of emotion faces. Compared with neutral expressions, surprise was perceived as less intelligent, anger was perceived as less healthy, and happy was perceived as more intelligent and healthy, but none of these effects were mediated by their resemblance to babies.

6. The gamma coefficient for expression was not strengthened with attractiveness controlled in Model 2a despite its suppressing effect on impressions of happy expressions' low dominance because the simultaneous control of babyface ratings exerted an influence in the opposite direction, even though babyface ratings failed to qualify as a mediator. The gamma coefficient for expression with only babyfaceness added to the model, γ = —.11, p < .10, was strengthened to γ = —.14, p < .05 when attractiveness was also added to the model, revealing the suppression effect.