Intention Modelling Clause Samples

Intention Modelling. ‌ Perception of the social world in terms of agents and their intentional relations and belief attribution is a funda- mental cognitive ability that lays the foundations of our social awareness [Woo+09]. This skill enables reasoning about other agents in the environment and performing appropriate decision making. Mentalising, mentalisation, or theory of mind refers to this ability to read the mental states of other agents [FF06]. The psychology, cognitive science and artificial intelligence community has been investigating computational mod- els for artificial intention reading for many years. Mentalising approaches have been used to perform social sim- ulations of multiple artificial agents [PM05]. A recent common approach to model reasoning about other agents’ desires and beliefs based on their actions is the Bayesian theory of mind (BToM) model [Bak+17; PK19], that considers the problem of inferring the mental states of agents through probabilistic inverse-planning. In contrast, [Rab+18] designed a neural network which uses meta-learning to build such models of the agents, able to predict the behavior of multiple agents in a false-belief situation given their past and current trajectories. The generator could make use of some of these methods to generate believable attributions for agents’ intentions in the simulated scene.