|
Abstract
|
Evaluating and modeling human performance on even simple tasks requires a great deal of attention to millisecond-level cognitive and perceptual-motor operations. Modeling human performance in a task often requires that special care be taken to understand how these millisecond level operations are interleaved and how they evolve during the execution of the task. In modeling a simple decision-making task, we found that human subjects improved their routine speed as they became more familiar with the task. Modeling was conducted using the ACT-R architecture (Anderson & Lebiere, 1998). Refinements of the model indicated that interleaving of millisecond-level perceptual-motor and cognitive operators was crucial in accounting not only for the strategy shift as per soft constraints, but also in the marked speedup in performance over the course of several trials.
|
Author Information
|
|
Download
|
|
|
|
download publication