Behavioural Abstraction of Agent Models Addressing Mutual Interaction of Cognitive and Affective Processes

Abstract

Abstract. In this paper the issue of relating a specification of the internal processes within an agent to a specification of the behaviour of the agent is addressed. A previously proposed approach for automated generation of behavioural specifications from an internal specification was limited to stratified specifications of internal processes. Therefore, it cannot be applied to mutually interacting cognitive and affective processes described by interacting loops. However, such processes are not rare in agent models addressing integration of cognitive and affective processes and agent learning. In this paper a novel approach is proposed which addresses this issue. The proposed technique for loop abstraction is based on identifying dependencies of equilibrium states for interacting loops. The technique is illustrated by an example of an internal agent model with interdependent processes of believing, feeling, and trusting.

Abstraction of an Affective-Cognitive Decision Making Model Based on Simulated Behaviour and Perception Chains

“…The abstracted agent model ensures an increase of the computational efficiency from more than twice for smaller numbers upto at least 3 to 4 times for larger numbers, at the cost of approximately 1% behavioural error. The abstraction mechanism employed can be used for a wide range of agent models with internal loops limited only by the requirements identified in [14]. Further abstraction steps can be made by dynamic clustering of agents in groups considered as higher-order agents; this will be addressed in future research.…”

“…In this section the process of abstraction of the affective-cognitive internal agent model from Section 2 is described, making use of the procedure from [14], which allows elimination of the 'as if body loops' from the model. To apply this loop elimination procedure, the following representation of a loop is assumed:…”

“…This is illustrated in this paper for a variant of the model instantiated for an emergency case. To address this problem, a loop abstraction technique from [14] has been used. This technique is based on identifying how equilibrium states for loops in an internal agent model depend on inputs for these loops.…”

Biological and Computational Perspectives on the Emergence of Social Phenomena: Shared Understanding and Collective Power

“…However, under certain assumptions such internal models can be abstracted to behavioural agent models providing more efficient processing, which is important especially when larger numbers of agents are simulated. In [55], [56] it is addressed how more complex internal agent models can be abstracted to less complex behavioural models. Agent models used for collective social phenomena traditionally are kept simple, and often are specified by simple reactive rules that determine a direct response (output) based on the agent's current perception (input).…”

“…In [56] an automated transformation is introduced from an internal agent model to a behavioural model, abstracting from the internal states. Within this transformation, techniques for loop abstraction are applied by identifying how equilibrium states depend on inputs for these loops.…”