Emotion recognition bias depends on stimulus morphing strategy

Abstract

Emotion recognition is vital for social interactions, and atypical (or biased) emotion recognition has been linked to mental health disorders including depression and anxiety. However, biases in emotion recognition vary across studies, and it is unclear whether this reflects genuine group differences in psychological processes underlying emotion recognition or differences in methodologies. One common method to measure biases in emotion recognition involves morphing a face between two emotional expressions in different ratios and asking participants to categorise the faces as belonging to one of the two emotion categories (‘direct-morphing’ method). However, this method creates morphed faces that are not ecologically valid. Alternatively, faces may be morphed through a neutral expression (‘morphing-through-neutral’ method), which is more ecologically valid since emotional expressions usually start from a neutral face. To compare these two approaches, we measured emotion recognition biases using two morphing techniques in 136 participants who also completed measures of anxiety (GAD-7) and depression (PHQ-9). Biases obtained using the two methods differed significantly: In the direct-morphing method, participants perceived the central 50% happy/50% angry face as slightly happy, whereas in the morphing-through-neutral method the neutral face was seen as angry. There were no associations between biases and depression or anxiety scores for either morphing method. This study is the first to directly compare emotion recognition biases obtained using two different morphing methods and is a first step towards reconciling discrepancies in the literature.