Using modified incremental chart parsing to ascribe intentions to animated geometric figures

Abstract

People spontaneously ascribe intentions on the basis of observed behavior, and research shows that they do this even with simple geometric figures moving in a plane. The latter fact suggests that 2-D animations isolate critical information-object movement-that people use to infer the possible intentions (if any) underlying observed behavior. This article describes an approach to using motion information to model the ascription of intentions to simple figures. Incremental chart parsing is a technique developed in natural-language processing that builds up an understanding as text comes in one word at a time. We modified this technique to develop a system that uses spatiotemporal constraints about simple figures and their observed movements in order to propose candidate intentions or nonagentive causes. Candidates are identified via partial parses using a library of rules, and confidence scores are assigned so that candidates can be ranked. As observations come in, the system revises its candidates and updates the confidence scores. We describe a pilot study demonstrating that people generally perceive a simple animation in a manner consistent with the model.

Agency and rationality: Adopting the intentional stance toward evolved virtual agents.

“…What does an object's behavior reveal about its mental state? These questions have historically received a great deal of attention in developmental psychology (e.g., Gelman, Durgin, & Kaufman, 1995;Gergely, Nádasdy, Csibra, & Bíró, 1995;Johnson, 2000;Kuhlmeier, Wynn, & Bloom, 2003;Onishi & Baillargeon, 2005;Williams, 2000) and philosophy (e.g., Goldman, 2006;Heal, 1996;Nichols & Stich, 1998;Stich & Nichols, 2003), and are receiving increasing attention in psychophysics with adult subjects (Barrett, Todd, Miller, & Blythe, 2005;Blythe, Todd, & Miller, 1999;Gao, McCarthy, & Scholl, 2010;McAleer & Pollick, 2008;Pratt, Radulescu, Guo, & Abrams, 2010;Pantelis & Feldman, 2012;Tremoulet & Feldman, 2000, 2006Zacks, Kumar, Abrams, & Mehta, 2009) and in computational modeling (Baker, Saxe, & Tenenbaum, 2009;Burgos-Artizzu, Dollár, Lin, Anderson, & Perona, 2012;Crick & Scassellati, 2010;Feldman & Tremoulet, 2008;Kerr & Cohen, 2010;Pantelis et al, 2014;Pautler, Koenig, Quek, & Ortony, 2011;Thibadeau, 1986). Many of these past studies have relied on the direct parametric manipulation of the physical qualities of stimulus objects (e.g., their velocity or acceleration), and measurement of the resulting subjective percepts (such as perceived animacy).…”

An Authoring Tool for Movies in the Style of Heider and Simmel

“…As in previous research [3,12], we aim to develop algorithms that can automatically interpret and narrate observed behavior, in much the same fashion as the subjects in Heider and Simmel's original study. This authoring tool allows us to collect hundreds or thousands of movie and narration pairs from volunteers, which can be used both to evaluate the performance of our algorithms and to model the relationships between observed action, intentions, and the language used to narrate interpretations.…”

Commonsense Interpretation of Triangle Behavior

“…Thibadeau (1986) takes a symbolic approach, representing the coordinates of each object in each frame of original film, which are matched to defined action schemas, such as opening the door or going outside the box. Pautler et al (2011) follows a related approach, beginning with object trajectory information from an animated recreation of the Heider-Simmel film. An incremental chart parsing algorithm with a hand-authored action grammar is then applied to recognize character actions as well as their intentions.…”