Workload is Bad, Except when it's Not: The Case of Avoiding Attractive Distractors

Workload is Bad, Except when it’s Not: The Case of Avoiding Attractive Distractors Christopher W. Myers, Wayne D. Gray, & Michael J. Schoelles Cognitive Science Department Rensselaer Polytechnic Institute [myersc, grayw, schoem] @rpi.edu Abstract information is all that is necessary for consistent saccades. The two effects taken together suggest that the saccadic mechanism relies on spatial properties of stimuli when determining saccadic endpoints. The amount of influence deliberate, top-down strategies have on saccadic endpoint location is still unclear. However, it is unlikely that humans solely rely on purposeful, top- down strategies when determining saccadic endpoints. He and Kowler (1991) propose a serial, two-stage process for determining saccadic endpoints that incorporates both automatic processes and intentional strategies. The two- stage process involves an initial intentional target selection, followed by an automatic weighted averaging of the shape or stimuli to determine the saccadic endpoint. Shen, Reingold, and Pomplun (2000) demonstrated that in a conjunctive search task visual search is also affected by the cost structure of the search environment. When few same-color distractors were present, saccadic selectivity was biased towards color. However, as the number of same- color distractors increased, saccadic selectivity shifted from same-color to same-shape stimuli. This suggests that visual search may be sensitive to the soft constraints of the search space. Hard constraints arise from the types of stimuli built into the search environment, and the types of interactive behavior permitted (such as searching by color or shape). Hence, hard constraints determine which microstrategies are possible (Gray & Boehm-Davis, 2000). In contrast, soft constraints determine which of the possible microstrategies are most likely to be selected (Gray & Fu, 2004). When selection is non-deliberate or automatic the least effort microstrategy is chosen. Searching same-color targets when they are the majority distractor leads to higher movement latencies, higher manual response times, and more fixations than searching the minority, same-shape distractors (Shen et al., 2000). Our research has focused on where a participant is likely to initially fixate. Initial fixations are the dwells located at the endpoint of the initial saccade. This work has uncovered an effect of stimuli density on initial fixation locations, or the pro-density effect (Myers, Gray, & Schoelles, 2003, 2004). As the density of stimuli increases (inter-stimulus distances become smaller), the probability of initially fixating the dense group also increases. Our work in conjunction with Shen et al. (2000) makes it apparent that stimulus features are not the only aspects of the search space considered. Rather, we have found that stimulus configurations are also important. It is likely that the results of Shen et al. (2000) and Myers et al. (2003; 2004), are solely attributable to neither data–driven nor purposeful, Increased cognitive workload is typically considered to hinder task performance. The current study presents an example where increased workload aided a visual search task. Increased workload, via a secondary task, provided participants extra time to avoid distracting stimulus configurations. Furthermore, initial fixations on distracting densities occurred at higher frequencies when initial saccades lasted less–than 400 milliseconds. We conclude that the combination of the primary visual search task and the secondary task create an environment where the secondary task was beneficial to the visual search task. Introduction There is a rich literature demonstrating how visual stimuli affect visual search patterns (Findlay, 1982, 1997; He & Kowler, 1991; McCarley, Kramer, & Peterson, 2002; Pomplun, Reingold, & Shen, 2003; Rayner, Liversedge, White, & Vergilino-Perez, 2003; Wolfe, Cave, & Franzel, 1989; Zelinsky, 1996). However, few studies have focused on how stimulus configurations influence eye movements. An example of a stimulus configuration is differences in inter-stimulus distance, or density. Stimulus density can be easily manipulated. Increasing the inter-stimulus distance decreases density, and vice-versa. There is also little research describing the effects of increased workload on visual search. Do visual search strategies change as a function of workload? In this paper, we address workload and stimulus configuration effects on visual search. Previous research suggests that saccades are programmed and targeted in an automatic, data-driven fashion. Data- driven processes shape overt behavior via environmental factors, and are typically considered unconscious processes. There are two striking examples that suggest data-driven processes determine saccadic endpoints. The first example is the global effect (Findlay, 1982, 1997). The global effect occurs when saccadic endpoints land at intermediate target positions during abrupt onset tasks containing at least two stimuli. That is, when two stimuli appear to the right or left of an initial fixation point, saccadic endpoints tend to be located between the stimuli. The global effect provides evidence that global target configurations influence saccadic amplitude. It appears that saccadic processes use stimulus attributes such as spatial properties in determining endpoints. The second example is the center–of–gravity effect (He & Kowler, 1991). The center–of–gravity effect indicates that saccades directed toward a shape land at consistent locations near the center of the shape, and that the shape’s contour