Agent-Based Model for False Belief Tasks: Belief Representation Systematic Approach (BRSA)

EasyChair Preprint no. 50

Abstract

A meaningful social life relies on understanding others’ minds and behaviours. The ability to reason about an individual’s mental states such as beliefs and desires, and to understand and predict how these mental states shape an individual’s behaviour is called theory of mind (ToM). In order to examine an individual’s ToM ability, false belief tasks have been used widely in the literature.

This research is a novel attempt to clarify the basic cognitive processes shared across the different varieties of false belief tasks. For this purpose, an agent-based model has been implemented to evaluate how agents’ achievement of goals in a social context is dependent on the ability to understand others’ beliefs.

In our study, we offer a methodological framework for many belief-reasoning tasks called Belief Representation Systematic Approach (BRSA). BRSA is a simple and robust approach that breaks down false belief tasks into four fundamental cognitive phases, including Perception, Memory, Reasoning beliefs and desires, and Expressing others’ beliefs and desires in an action. BRSA identifies a network of indispensable resources for belief representation. It also clarifies that there is a difference between ‘understanding’ others’ beliefs and ‘using’ that understanding. BRSA demonstrates that false belief tasks, as a common decisive methodology for theory of mind competence, might involve more than understanding others’ beliefs. Moreover, the model demonstrates that agents’ understanding of others’ beliefs on the micro level will lead to significant improvements in their performances on the macro level.

Keyphrases: agent-based model, belief representation, False Belief Task, theory of mind

@Booklet{EasyChair:50,
  author = {Zahrieh Yousefi and Dietmar Heinke and Ian Apperly and Peer-Olaf Siebers},
  title = {Agent-Based Model for False Belief Tasks: Belief Representation Systematic Approach (BRSA)},
  howpublished = {EasyChair Preprint no. 50},
  doi = {10.29007/ttdg},
  year = {EasyChair, 2018}}