Top-down Planning and Bottom-up Perception in a Problem-solving Task Enkhbold Nyamsuren (e.nyamsuren@rug.nl) Niels A. Taatgen (n.a.taatgen@rug.nl) Department of Artificial Intelligence, University of Groningen, Nijenborgh 9, 9747 AG Groningen, Netherlands Color, Number, Shape and Shading. Each attribute can have one of three distinct values: Red, Green and Blue for the Color attribute; Open, Solid and Textured for the Shading attribute; One, Two and Three for the Number attribute; Oval, Rectangle and Wiggle for the Shape attribute. The gameplay for SET is relatively simple. At any moment in the game, 12 cards are dealt open (Figure 1). Players should find any combination of three cards, further referred to as a set, satisfying the main rule stating that in the three cards the values for a particular attribute should be all the same or all different. The number of different attributes in set cards is further referred as a level of the set. As such, the set, in which only one attribute is different, is level 1 set. Correspondingly, there can be levels of 2, 3 or 4. Figure 1 shows examples of level 1 (different shape) and level 4 sets (all attributes are different). In the regular game, if a player finds a set, he or she picks up the three cards that form a set, and replaces them with new cards from the deck. After the deck runs out the player with most cards wins. Abstract In this paper we study the roles of top-down planning and the bottom-up elements in problem-solving tasks. We investigate how factors, such as conceptual understanding, perceptual representation and previous experience with the task, influence the action selection. The cognitive and perceptual aspects of problem- solving task are studied within the environment of card game SET. The discussion is provided on cognitive and perceptual demands on the game, and the difference between novice and expert players is analyzed with respect to two types of processes. The hypotheses proposed in this paper are tested on data obtained through an eye tracking experiment. Based on findings the ACT-R model of human player is implemented and compared to human performance. Keywords: cognitive architecture; visual attention; cognitive control; games; ACT-R, problem solving. Introduction Human performance in complex tasks is often a combination of internal planning and responding appropriately to the environment. Nevertheless, cognitive models of complex tasks typically focus on the mental planning aspects, and fail to consider possible influence of an external world on the control of behavior. The role of the environment was first recognized in robotics (Brooks, 1991) but was later extended to human cognition in the embodied cognition approach (e.g., Clark, 1997; Kirsh & Maglio, 1994). The challenge is to understand how control is shared between goal-driven planning and processes that are driven by perceptual input. The approach we will take is to assume two parallel processes: a bottom-up visual process that scans the visual field on the basis of saliency and similarity, and a top-down planning process that tries to achieve the goal, but also biases the bottom-up process. Finding an appropriate task to study the cognitive aspects of human behavior in real-life situation is not easy. However, games provide environments that often require the same type of complex processes that are usually involved in real-world situations. This has the advantage that behavior of a player can be studied in a controlled environment. These qualities make games on a computer an ideal tool for studying complex cognitive processes. One such game is the card game SET 1 . The SET card deck consists of 81 cards. Each card differs from other cards by a unique combination of four attributes: Figure 1: An example array of 12 cards. The cards with the solid highlight form level 4 set (all attributes are different), and cards with dashed highlight form level 1 set (Shape is different). There are several advantages of choosing SET as a target game of study. SET has very simple rules to follow and relatively static game environment. Despite the simplicity, SET requires complex cognitive processes including pattern recognition, visuospatial processing and decision making. It is our hypothesis that in SET both cognitive and perceptual processes are equally important to play the game. As such, SET provides an excellent opportunity to study the dynamics of such processes in a relatively simple game SET is a game by Set Enterprises (www.setgame.com)
[1]
Paul P. Maglio,et al.
On Distinguishing Epistemic from Pragmatic Action
,
1994,
Cogn. Sci..
[2]
R. A. Brooks,et al.
Intelligence without Representation
,
1991,
Artif. Intell..
[3]
Helge J. Ritter,et al.
Comparative visual search: a difference that makes a difference
,
2001,
Cogn. Sci..
[4]
A. Clark.
Being There: Putting Brain, Body, and World Together Again
,
1996
.
[5]
John R. Anderson.
How Can the Human Mind Occur in the Physical Universe
,
2007
.
[6]
Robert W. Kentridge,et al.
Eye movement research : mechanisms, processes and applications
,
1995
.
[7]
K. Rayner.
Eye Movements and Cognitive Processes in Reading, Visual Search, and Scene Perception
,
1995
.
[8]
K. Cave,et al.
Spatial Attention in Visual Search for Features and Feature Conjunctions
,
1995
.
[9]
N. Taatgen,et al.
How to construct a believable opponent using cognitive modeling in the game of set
,
2003
.
[10]
Shaul Hochstein,et al.
Set recognition as a window to perceptual and cognitive processes
,
2008,
Perception & psychophysics.
[11]
Dario D. Salvucci,et al.
Threaded cognition: an integrated theory of concurrent multitasking.
,
2008,
Psychological review.
[12]
R. Baayen,et al.
Mixed-effects modeling with crossed random effects for subjects and items
,
2008
.