Overconfident, but angry at least. AI-Based investigation of facial emotional expressions and self-assessment bias in human adults

Abstract

Metacognition and facial emotional expressions both play a major role in human social interactions [1, 2] as inner narrative and primary communicational display, and both are limited by self-monitoring, control and their interaction with personal and social reference frames. The study aims to investigate how metacognitive abilities relate to facial emotional expressions, as the inner narrative of a subject might project subconsciously and primes facial emotional expressions in a non-social setting. Subjects were presented online to a set of digitalised short-term memory tasks and attended a screening of artistic and artificial stimuli, where their facial emotional expressions were recorded and analyzed by artificial intelligence. Results show self-assessment bias in association with emotional expressivity – neutrality, saturation, transparency – and the display of anger and hostility as an individually specific trait expressed at modality-dependent degrees. Our results indicate that self-assessment bias interplays in subconscious communication – the expression, control and recognition of facial emotions, especially – with empathetic skills and manipulation.