Academic journal article Attention, Perception and Psychophysics

Exploring the Mental Space of Autonomous Intentional Agents

Academic journal article Attention, Perception and Psychophysics

Exploring the Mental Space of Autonomous Intentional Agents

Article excerpt

Published online: 18 October 2011

© Psychonomic Society, Inc. 2011

Abstract The ability to infer the intentions of other agents on the basis of their motion is a critical psychological faculty. In the present study, we examine a key question underlying this process, namely: What are the psychologically natural categories of intentional agents and actions? To investigate this question empirically, we use displays containing a number of autonomous, independently programmed agents moving about a two-dimensional environment and interacting with one another. Each agent behaves according to its own simple program, controlled by a small number of parameters that define its "personality." We probe participants' impressions of the similarities among the behaviors of the various agents, and then use multidimensional scaling in an attempt to recover the subjective mental space of agent types. An important variable underlying this space turns out to be a parameter that determines how the agent reacts to a nearby agent at one critical distance. A follow-up experiment suggests that variation along this parameter ultimately contributes to modulating a more fundamental perceptual dimension that reflects how "hostile" or "friendly" the agents appear to be.

Keywords Motion: Biological . Visual perception . Motion: *Other

(ProQuest: ... denotes formula omitted.)

Intelligent agents can and must distinguish between animate and inanimate objects they encounter in the world. Even infants make this distinction, apparently possessing a naive theory of other beings' mental states and intentions (Gergely, Nádasdy, Csibra, & Bíró, 1995; Johnson, 2000; Keil, 1994; Leslie, 1984). Socially intelligent agents naturally conceive of other humans as animate, mentalistic agents with independent perceptions and motivations. We further benefit from being able to infer the intentions of other agents in the environment. This is essential for understanding and predicting others' behavior, a prime skill whether for chess players contemplating their moves or for gazelles and lions engaging in mutual scrutiny on the African plain. The understanding of intentions is also critical for parsing of the perceptual flow of observed activities into meaningful, segmented events (Zacks, 2004; Zacks, Kumar, Abrams, & Mehta, 2009).

One particularly salient cue for intention is motion, although it is admittedly only one cue among many (Gelman, Durgin, & Kaufman, 1995). Ever since the famous study of Heider and Simmel (1944), it has been well known that participants will readily ascribe intentionality even to simple moving geometric figures moving about sparse environments. A handful of studies have shown that varying the motion of simple geometric figures along certain parameters (e.g., speed, trajectory) can influence the perception of animacy and intentions (Dittrich & Lea, 1994; Gao, Newman, & Scholl, 2009; Gao, McCarthy, & Scholl, 2011; Tremoulet & Feldman, 2000, 2006). Animate motion captures visual attention (Pratt, Radulescu, Guo, & Abrams, 2010). But the factors creating the impression of animacy, and the computational mechanisms of intentional interpretation, are still poorly understood.

In the present study, we investigate how adult human observers use visual motion to infer an agent's goals, behavioral dispositions, or other temperamental or psychological characteristics, given that the observer attributes animacy and intentionality to this agent. In any computational treatment of intention estimation (e.g., Baker et al., 2006, 2009; Feldman & Tremoulet, 2008), assumptions must always be made about the class of intentional agents to be considered as candidate models. For example, in a Bayesian formulation of the problem, priors must be defined over some hypothesis space of intentional agent types. But the empirical bases for such assumptions are still very much open questions. What are the natural psychological classes of agent types? …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.