Areas of Interest as a Signal Detection Problem in Behavioral Eye‐Tracking Research

Abstract

Decision researchers frequently analyze attention to individual objects to test hypotheses about underlying cognitive processes. Generally, fixations are assigned to objects using a method known as area of interest (AOI). Ideally, an AOI includes all fixations belonging to an object while fixations to other objects are excluded. Unfortunately, due to measurement inaccuracy and insufficient distance between objects, the distributions of fixations to objects may overlap, resulting in a signal detection problem. If the AOI is to include all fixations to an object, it will also likely include fixations belonging to other objects (false positives). In a survey, we find that many researchers report testing multiple AOI sizes when performing analyses, presumably trying to balance the proportion of true and false positive fixations. To test whether AOI size influences the measurement of object attention and conclusions drawn about cognitive processes, we reanalyze four published studies and conduct a fifth tailored to our purpose. We find that in studies in which we expected overlapping fixation distributions, analyses benefited from smaller AOI sizes (0° visual angle margin). In studies where we expected no overlap, analyses benefited from larger AOI sizes (>.5° visual angle margins). We conclude with a guideline for the use of AOIs in behavioral eye‐tracking research. Copyright © 2015 John Wiley & Sons, Ltd.

Effects of salience are both short- and long-lived

“…To account for inaccuracy in recording of fixation locations, the margins of the AOIs were approximately .5°larger than the label. For different views on optimal AOI sizes see Holmqvist et al (2011) and Orquin, Ashby, and Clarke (2015).…”

The pyeTribe: Simultaneous eyetracking for economic games

“…A second indication of the data quality is the relation of the fixations within AOIs versus non-AOIs (but see Orquin, Ashby, & Clarke, 2016, on flexibility in the definition of AOIs and the consequences for data analysis). Our feedback screen consisted of eight AOIs with numeric information and six AOIs with written text (see Fig.…”

Missing the best opportunity; who can seize the next one? Agents show less inaction inertia than personal decision makers

“…Next, it was divided by the number of experimental stimuli (i.e., 45), reflecting the average number of fixations 1 . A 2 (role: personal decision maker or agent) Â 2 (price: previous price or original price) repeated-measures ANOVA on the average number of fixations yielded a main effect Note that results of eye-tracking research may be influenced by AOI size (Orquin et al, 2016). Therefore, AOI size was changed to test whether our results were robust.…”

“…Hence, BeGaze was used to physically draw AOIs around the stimuli (drawing method). This approach allows for a large degree of flexibility when defining the shape and size of AOIs (Orquin, Ashby, & Clarke, 2016). As recommended by Holmqvist et al (2011), the AOIs were defined as larger than the stimuli.…”