Emulating a Visuospatial Memory Field Using ACT-R

We define visuospatial working memory (VSWM) as the set of cognitive processes used to visualize the locations of things. VSWM is not permanent visual memory; it is a temporary visual workspace used to solve spatial problems. VSWM is ubiquitous in everyday reasoning (for example, visualizing different ways of arranging furniture in a room), and is thought to be particularly important for certain occupations (for example, engineers, architects, and pilots). Both subjective experience and empirical studies suggest that the capacity of VSWM is limited (c.f. Logie, 1994). Our goal is to model the underlying sources of this capacity limitation. To measure VSWM capacity, we use a new method called Path Visualization (PV, Lyon, 2004). PV is similar to some existing methods (e.g. Brooks, 1968; Attneave & Curlee, 1983; Kerr, 1987, 1993; Diwadkar, Carpenter & Just, 2000; Barshi & Healy, 2002) in which participants must visualize relative positions or movements (up, right, down...) in a two-dimensional or threedimensional array of locations. However PV also requires participants to make continuous timed spatial judgments about intersections in a path defined by such movements. This yields accuracy and response time data that help define the VSWM representation of an imaginary path.