Abstract
Wolfe's (1994) Guided Search model was applied to a simple visual search task: looking for a store on a mall directory. The directory consisted of eight stores represented as rectangles arranged from left to right. Each rectangle contained the store name and was one of three colors. The project involved two parts: fitting the model parameters to the environment and using the resulting model to create optimal directory designs. For parameter fitting, 600 different mall directories were generated with random assignments of store names and colors to positions. Observers viewed the directory and a target store and made a speed judgment on whether the target was in the directory or not. Sixty-nine observers in four groups each viewed 150 trials. Average RTs for the correct identifications of target present across each group were used with the directory features to identify the 15 model parameters that best fit the data. The model did a good job of fitting the data (r=.721). Running the same parameter fit technique with RTs randomly assigned to trials with different directory features did not lead to good fits (average r=.053). Next, we explored how to use the model to design an optimal directory that minimizes visual search time. Given a distribution of frequency searching for different stores in the directory, what is the assignment of colors to stores that minimizes average visual search time? The answer was found by examining all possible color combinations and for each one computing the average visual search time. For the directories we used, the model makes a counterintuitive prediction: color doesn't matter. The parameter fitting data found large effects of target position on RT, with a target in the middle of the directory being found fastest (closest to a pre-stimulus fixation point). Any effects of color were miniscule compared to the larger effects of target position. This finding is useful to a designer who might otherwise fret over how to assign colors to store positions.