Academic journal article Psychonomic Bulletin & Review

Decision-Tree Analysis of Control Strategies

Academic journal article Psychonomic Bulletin & Review

Decision-Tree Analysis of Control Strategies

Article excerpt

Published online: 15 October 2014

© Psychonomic Society, Inc. 2014

Abstract A major focus of research on visually guided action is the identification of control strategies that map optical information to actions. The traditional approach has been to test the behavioral predictions of a few hypothesized strategies against subject behavior in environments in which various manipulations of available information have been made. While important and compelling results have been achieved with these methods, they are potentially limited by small sets of hypotheses and the methods used to test them. In this study, we introduce a novel application of data-mining techniques in an analysis of experimental data that is able to both describe and model human behavior. This method permits the rapid testing of a wide range of possible control strategies using arbitrarily complex combinations of optical variables. Through the use of decision-tree techniques, subject data can be transformed into an easily interpretable, algorithmic form. This output can then be immediately incorporated into a working model of subject behavior. We tested the effectiveness of this method in identifying the optical information used by human subjects in a collision-avoidance task. Our results comport with published research on collision-avoidance control strategies while also providing additional insight not possible with traditional methods. Further, the modeling component of our method produces behavior that closely resembles that of the subjects upon whose data the models were based. Taken together, the findings demonstrate that data-mining techniques provide powerful new tools for analyzing human data and building models that can be applied to a wide range of perception-action tasks, even outside the visual-control setting we describe.

Keywords Math modeling and model evaluation · Computational modeling · Perception and action · Perceptual attunement

(ProQuest: ... denotes formulae omitted.)

Introduction

An animal's perception of the environment and ability to control action is made possible by the availability of information contained in ambient energy arrays (Gibson 1986). An understanding of the components within these energy arrays that make them informative often yields useful insights into how animals perceive and act in a particular context. In the study of visually guided action,thisoften entails an analysis of optic flow fields, with the goal of identifying optical variables that specify action-relevant properties and prescribe how to act to bring about a particular goal. This undertaking has led to the formulation of numerous control laws for steering, fly ball catching, braking to avoid a collision, and a variety of other tasks (see Fajen (2005b); Warren (1998) for reviews).

Discovering new sources of information through the analysis of optic flow fields alone is extremely challenging. The aim of this paper is to introduce a novel approach to identifying sources of optical information for visual control. This approach makes use of a technique from data mining called decision-tree learning and differs from previous approaches in that behavioral data are used at an earlier stage of the process. Rather than relying entirely on the researcher's intuitions to identify candidate informational variables, the method we describe lets the behavioral data tell the researcher what those informational variables may be. As we show, this can lead to new insights into the information-based control of action. Furthermore, although we demonstrate the method within the context of visual control, it is hardly limited to this domain. In principle, essentially any experimental variable from multiple domains of interest can be investigated in the manner we describe here.

Information and laws of control: A brief review

Control laws (or control strategies) often take the form of mappings from information variables that specify actionrelevant states of the environment to action variables that are controlled by the actor (Warren 1988; 1998). …

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.