Abstract
This study aimed to discover predictors of subjective and objective difficulty in emotion perception from dynamic facial expressions. We used a multidimensional emotion perception framework, in which observers rated the perceived emotion along a number of dimensions instead of choosing from traditionally-used discrete categories of emotions. Data were collected online from 441 participants who rated facial expression stimuli in a novel paradigm designed to separately measure subjective (self-reported) and objective (deviation from the population consensus) difficulty. We targeted person-specific (sex and age of observers and actors) and stimulus-specific (valence and arousal values) predictors of those difficulty scores. Our findings suggest that increasing age of actors makes emotion perception more difficult for observers, and that perception difficulty is underestimated by men in comparison to women, and by younger and older adults in comparison to middle-aged adults. The results also yielded an increase in the objective difficulty measure for female observers and female actors. Stimulus-specific factors-valence and arousal-exhibited quadratic relationships with subjective and objective difficulties: Very positive and very negative stimuli were linked to reduced subjective and objective difficulty, whereas stimuli of very low and high arousal were linked to decreased subjective but increased objective difficulty. Exploratory analyses revealed low relevance of person-specific variables for the prediction of difficulty but highlighted the importance of valence in emotion perception, in line with functional accounts of emotions. Our findings highlight the need to complement traditional emotion recognition paradigms with novel designs, like the one presented here, to grasp the "big picture" of human emotion perception.