Retrodictive inferences from spontaneous and simulated facial expressions

This thesis aimed to investigate whether people can accurately retrodict from a brief sample of others’ behaviour to the antecedent event, from spontaneous and simulated expressions. Additionally, the thesis aimed to investigate if retrodiction involves the inference of inner states, unlike a classification task. The experimental procedure enables us to capture a variety of brief, dynamic and spontaneous expressions which closely resembles the expressions seen in real-life social situations. At the same time, it enables us to measure participants’ performance against an objective criterion. This paradigm consists of two phases, targets and perceivers. Targets captured in video-clips were either looking at emotional expressions (Experimental Chapter 1 and 2) or pictures from the International Affective Picture System (IAPS) (Experimental Chapter 3 and 4). As a cover activity to provoke spontaneous reactions, targets were told to remember the pictures for a memory test. In Experimental Chapter 3 and 4, they were also asked to simulate expressions which were either congruent or incongruent with what they were viewing. In order to further identify if perceivers were truly inferring targets’ inner states, their (“retrodicting perceivers”) behavioural responses and eye movements were compared against perceivers who were asked to classify target responses (“classifying perceivers”). The findings revealed that retrodicting perceivers were slightly above chance level in judging the emotional expressions viewed by targets (Experimental Chapter 1 and 2) from spontaneous reactions. As for simulated expressions, perceivers were also significantly above chance in judging whether targets were displaying an incongruent expression only when targets were looking at a negative picture (Experimental Chapter 3). However, when perceivers were asked to retrodict to the type of picture viewed, they performed significantly above chance only when targets were behaving spontaneously and when they were asked to simulate congruent expressions. In contrast, they were systematically inaccurate when targets were simulating incongruent expressions. Although this may suggest that perceivers were fooled by targets’ reactions, perceivers’ responses discriminated between occasions when targets were simulating congruent and incongruent expressions. Furthermore, perceivers’ responses also discriminated between occasions when targets were behaving spontaneously and when they were asked to simulate congruent expressions. In this respect, perceivers were not entirely fooled by targets’ simulations. Finally, there was insufficient evidence to support the hypothesis that retrodicting perceivers engaged in a different strategy than classifying perceivers and hence it is not possible to conclude that retrodicting perceivers were truly inferring targets’ inner states