Michael Matessa

ACT-R: a theory of higher level cognition and its relation to visual attention

The ACT-R system is a general system for modeling a wide range of higher level cognitive processes. Recently, it has been embellished with a theory of how its higher level processes interact with a visual interface. This includes a theory of how visual attention can move across the screen, encoding information into a form that can be processed by ACT-R. This system is applied to modeling several classic phenomena in the literature that depend on the speed and selectivity with which visual attention can move across a visual display. ACT-R is capable of interacting with the same computer screens that subjects do and, as such, is well suited to provide a model for tasks involving human-computer interaction. In this article, we discuss a demonstration of ACT-Rs application to menu selection and show that the ACT-R theory makes unique predictions, without estimating any parameters, about the time to search a menu. These predictions are confirmed.

Eye tracking the visual search of click-down menus

Click-down (or pull-down) menus have long been a key componentof graphical user interfaces, yet we know surprisingly little abouthow users actually interact with such menus. Nilsens [8] study onmenu selection has led to the development of a number of models ofhow users perform the task [6, 21. However, the validity of thesemodels has not been empirically assessed with respect to eyemovements (though [l] presents some interesting data that bear onthese models). The present study is an attempt to provide data thatcan help refine our understanding of how users interact with suchmenus.

Explorations of an Incremental, Bayesian Algorithm for Categorization

An incremental categorization algorithm is described which, at each step, assigns the next instance to the most probable category. Probabilities are estimated by a Bayesian inference scheme which assumes that instances are partitioned into categories and that within categories features are displayed independently and probabilistically. This algorithm can be shown to be an optimization of an ideal Bayesian algorithm in which predictive accuracy is traded for computational efficiency. The algorithm can deliver predictions about any dimension of a category and does not treat specially the prediction of category labels. The algorithm has successfully modeled much of the empirical literature on human categorization. This paper describes its application to a number of data sets from the machine learning literature. The algorithm performs reasonably well, having its only serious difficulty because the assumption of independent features is not always satisfied. Bayesian extensions to deal with nonindependent features are described and evaluated.

Modelling Focused Learning in Role Assignment.

ACT-R is a general theory of cognition (Anderson, 1993; Anderson & Lebiere, 1998) which is capable of learning the relative usefulness of alternative rules. In this paper, a model utilising this implicit procedural learning mechanism is described which explains results from a concept formation task created by McDonald and MacWhinney (1991), a role assignment task for artificial languages created by Blackwell (1995), and a new role assignment experiment. By focusing learning on one cue of role assignment at a time, the model predicts a blocking phenomenon where certain cues can dominate and partially block the learning of other cues. In all of the experiments, subjects’ trial-by-trial use of cues is better predicted by the ACT-R model than by a pure learning-on-error model that learns all cues simultaneously.

An incremental Bayesian algorithm for categorization

Publisher Summary This chapter discusses an incremental Bayesian algorithm for categorization. A rational analysis is an attempt to specify a theory of some cognitive domain by specifying the goal of the domain, the statistical structure of the environment in which that goal is being achieved, and the computational constraints under which the system is operating. The predictions about the behavior of the system can be derived assuming that the system maximizes the goals it expects to achieve while minimizing expected costs, where expectation is defined with respect to the statistical structure of the environment. This approach is different from most approaches in cognitive psychology because it tries to derive a theory from assumptions about the structure of the environment rather than assumptions about the structure of the mind. This approach is applied to human categorization and an effective algorithm for categorization is developed. The analysis assumes that the goal of categorization is to maximize the accuracy of predictions about features of new objects.