Modeling how menu search strategies develop with experience

To find an item in a menu, users can follow different visual search strategies, such as scanning items one by one (serial search) or trying to remember where the item was (recall search). However, building predictive models of search behavior has turned out to be challenging, because these strategies evolve with practice. To address this challenge, we study theory-inspired models of visual search in linear menus and propose a novel arbitration mechanism to coordinate the adoption of such visual search strategies. Given a menu design and the user’s previous experience with it, our approach predicts when different search strategies (serial, recall, random) will be adopted and which menu item will be fixated next. Our results (1) describe empirical data plausibly with psychologically valid and interpretable models, (2) provide new insights about how search strategies evolve with practice, and (3) show how to infer search strategy from eye tracking data. To sum up, the models provide a foundation to better understand how users learn to scan linear menus.