| Most driving-related eye-movement research completed before this present study has relied upon qualitative techniques to evaluate questions. These techniques primarily consist of descriptive statistics on distributional data. This is adequate for a single study, but can not be reliably compared to the results of another study having dissimilar conditions. This has led to a need for the development and testing of a new measurement technique that produces quantitative results that are scaleable across differing situations. The quantitative metric of entropy---derived from information theory---has been adapted toward this purpose. Visual entropy values were calculated from the eye movements of 28 young ( M = 24, SD = 5.3 yrs) and 14 older ( M = 75, SD = 5.6 yrs) participants who drove an instrumented vehicle along a predefined route while engaged in several functionally different cognitive tasks. Driver age and subsidiary task type (none, verbal, visual-spatial) were systematically varied in order to assess the sensitivity of the entropy metric to discriminate behaviors generated across these situational factors. Resulting entropy values were directly compared to several more common visual metrics such as pupil size, saccadic amplitude, and fixation dwell time. In all metrics, effects were most discernible in the visual-spatial condition which was further exacerbated in the older driver group. The finding that a visual attention task inhibited a visual scanning behavior the most supports Wickens' multiple resource theory. The finding that older individuals were the most affected in all cases supports the idea that older individuals may have diminished visual-cognitive resources available during driving compared to their younger counterparts. Upon comparison of the different visual metrics it was found that the entropy metric proved to be more sensitive to attentional demands than all alternative visual metrics assessed (e.g. pupil size, dwell times, and saccadic amplitudes). The findings of this study strongly suggest that global measures of eye movement behavior captured by the entropy metric are useful for understanding the correlation between normal adult aging and task-induced cognitive demands within the context of real-world driving. |
