Isolating Action Prediction from Action Integration in the Perception of Social Interactions

Abstract

Previous research suggests that predictive mechanisms are essential in perceiving social interactions. However, these studies did not isolate action prediction (a priori expectations about how partners in an interaction react to one another) from action integration (a posteriori processing of both partner’s actions). This study investigated action prediction during social interactions while controlling for integration confounds. Twenty participants viewed 3D animations depicting an action–reaction interaction between two actors. At the start of each action–reaction interaction, one actor performs a social action. Immediately after, instead of presenting the other actor’s reaction, a black screen covers the animation for a short time (occlusion duration) until a still frame depicting a precise moment of the reaction is shown (reaction frame). The moment shown in the reaction frame is either temporally aligned with the occlusion duration or deviates by 150 ms or 300 ms. Fifty percent of the action–reaction trials were semantically congruent, and the remaining were incongruent, e.g., one actor offers to shake hands, and the other reciprocally shakes their hand (congruent action–reaction) versus one actor offers to shake hands, and the other leans down (incongruent action–reaction). Participants made fast congruency judgments. We hypothesized that judging the congruency of action–reaction sequences is aided by temporal predictions. The findings supported this hypothesis; linear speed-accuracy scores showed that congruency judgments were facilitated by a temporally aligned occlusion duration, and reaction frames compared to 300 ms deviations, thus suggesting that observers internally simulate the temporal unfolding of an observed social interction. Furthermore, we explored the link between participants with higher autistic traits and their sensitivity to temporal deviations. Overall, the study offers new evidence of prediction mechanisms underpinning the perception of social interactions in isolation from action integration confounds.

Influence of social and semantic contexts in processing speech in noise

“…This raise the question of whether the influence of social traits on processing social stimuli is linked to the perceptual modality in itself, or would rather be linked to a domain-general mechanisms that would be generally involved in processing social interactions, independently to the modality. While further experiments will be needed to approach this question, behavioral effects that reliably characterize the performance of the individual, highlighting population-level and interindividual differences (see also Pesquita et al, 2022), appear especially worthy of attention as they can offer handy tools for detection of diseases or vulnerabilities.…”

When Corticospinal Inhibition Favors an Efficient Motor Response

“…We also know that perceptual grouping may occur even when part of the stimuli is masked [87]. Likewise, spatiotemporal simulation of a 'missing part' might be generated to be compared in real-time with the upcoming input during social interactions [11].…”

“…To achieve an efficient level of coordination, observers must simulate the unfolding of an observed request, compare the upcoming sensory input, predict the partner's intention from early movement cues [6][7][8][9][10] and finally activate the corresponding motor representation, well before the interactive request is fully expressed ( [11,12] for a review). However, how such delicate integration is reflected in modulation of corticospinal excitatory and inhibitory mechanisms is still poorly studied.…”

“…Moreover, the previous research has not fully disambiguated the relative contribution of two different mechanisms: action prediction (a priori expectations) and action integration (a posteriori processing of both partners' actions). A recent study by Pesquita and colleagues [11] was able to isolate action prediction from confounding factors related to action integration. Specifically, participants watched animated sequences with two characters (e.g., A and B involved in a handshake sequence).…”