A Computational Dual-Process Model of Social Interaction
Abstract
Dual-process models postulate two distinct modes of information processing, one automatically invoked, highly parallel, and not under conscious control, and the other consciously invoked and limited to serial processing. Stereotype-driven behaviors have been identified as an example of an implicitly activated and automatically invoked behavior. In this effort, we describe the modeling of stereotypes and prejudice, principally with respect to race and religion, in social interactions. We developed a computational model of these implicitly activated behaviors and how they are sometimes refined, even overridden, by often concurrent, explicitly-driven processes. The test-bed for the work reported on here is the Operator Model Architecture (OMAR), a simulation framework for agent-based modeling. OMAR was used to facilitate the building of the computational models of agents, visualized as avatars, which pursue goals that drive their behaviors in social interactions. In carrying out their proactive agendas the OMAR agents respond to perceptual stimuli that activate implicit attitudes which may, in turn, be overturned by stimuli that activate explicit attitudes leading to alternate behaviors. Our goal has been to model both the proactive and reactive processes, and their concurrent implicit and explicit components, in social interactions from perceptual input through behavioral consequences.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 30, 2014
- Accession Number
- ADA612453
Entities
People
- Michael Young
- Stephen Deutsch
Organizations
- BBN Technologies