Individual psychological conflict within social group dynamics

Abstract

In this paper we investigate the implications of a conceptual-model for competition in decision-making, where as a result psychological conflict might arise for either individuals or social groups of artificial embodied agents. We propose a nonlinear dynamical system with control approach to compose a socio-cognio-physical model of embodied agents in our crowd dynamics simulator, with a Lewin field theoretic model of cognition, which in principal is capable of being applied to a wide range of concrete problems. For instance, elsewhere in this conference Millikan et al. (2013) applies these Lewinian principles to the problem of military suppression for dismounted infantry. Wong (2012) has also used these constructs in a study on crowd reactions to Non-Lethal-Weapons. Of Lewin's nine constructs of field-theory we will largely be concerned with the constructs of conflict (approach-avoid psychological conflict amongst the typology of Lewinian psychological conflicts) position, locomotion, force, goal and will briefly touch on power (inducing psychological forces on others to cause them to act; e.g. command). In particular we present a brief recount of psychological conflict simulation models, together with a qualitative analysis of conflicting attractor behaviour. Further, the consequences of the field shape have been considered for a simple nonlinear model of attraction and repulsion. There are a number of reasons why consideration of complex nonlinearity is required for modelling embodied cognition. Among them, we note, it is not possible to look at individual behaviours in complete isolation (classic controlled psychology experiments), without breaking the whole web of interdependencies (Lewin, 1997, pp 383-410). Methods of investigation for general social dynamics have been reviewed by Castellano et al. (2009) as being entropic, topological, dynamical systems (DS) and agent based models (ABM). Castellano's statistical-physics based approach, however, fails to identify explicit individual entity control or self-regulation as an important element; see (Conant and Ashby, 1970). Therefore the minimum modelling capability of an embodied agent must be effectively the equivalent of a dynamical system with control (DSC), irrespective of whether the model implementation is actually an ABM or alternatively a nonlinear DS with each agent possessing a cybernetic controller for its cognitive model. While there are other computational models of cognition ranging from the disembodied to the social, see (Sun, 2008) for summaries on SOAR, ACT-R, CLARION, they cannot cope with the change in scale of modelling fidelity, because, in our view, they do not see their psychological and physical worlds in terms of fundamental forces, be they physical, psychological or social, in some appropriate geometrical space. Neither can they seamlessly recurse between a representation of the psychological-present moment and either the psychological-past or a projected psychological-future, or indeed both; see (Ivancevic and Johnson, 2012). They also fail to represent an explicit model of control (Johnson et al., 2011), breaking Ashby's dictum that “Every good regulator of a system must be a model of that system” (Conant and Ashby, 1970). In brief then, our broad problem is to find predictive embodied models of cognition with explicit models of control or self-regulation. We are following Millikan et al. (2011), who have suggested that understanding the detailed behaviour of conceptual models is a key part of a wider effort toward holistic societal modelling. In particular, the conceptual model under investigation here is one of nonlinear psychological competition in decision-making, where the competition may lead to psychological conflict and vacillatory behaviour or stable equilibria. The specific question we will examine in this study is the effect of the dynamic spatio-temporal field-strength shape on the behaviour of artificial embodied-agents. Following a standard approach used in theoretical-physics we will progressively examine field-strengths which vary linearly, then cubically and finally quintically. Further, appropriate control geometries and are being sought, following Lewin (1936), to represent the dynamics of individual agents and social-groups decision making interactions.