Influences of Social Learning in Individual Perception and Decision Making in People with Autism: A Computational Approach

Abstract

The present paper proposes a computational approach to explore the influences of social learning on social cognition among individuals with Autism Spectrum Disorder (ASD) compared to the Typically Developing (TD) group. An experimental paradigm is designed to perceive and differentiate social cues related to real-time road and traffic light situations. The computational metrics such as sensitivity index ($d'$), response bias (c) and detection accuracy (DA) are recorded and analysed using machine learning classifiers. The results revealed that cognitive level is attenuated in ASD ($d'=0.427$, $c=-0.0076$ and $DA=51.67\%$) compared to TD ($d'=1.42$, $c=-0.0027$ and $DA=80.33\%$) with an improvement considering social influence as key factor ($S_f$) with best-fit quantitative value for ASD ($S_f=0.3197$) when compared to TD ($S_f=0.3937$). The automated classification with an accuracy of 96.2% supported the significance of the metrics in distinguishing ASD from TDs. The present findings revealed that social conformity and social influence imparted growth in ASD cognition.