Eye Movements in Risky Choice

Abstract

We asked participants to make simple risky choices while we recorded their eye movements. We built a complete statistical model of the eye movements and found very little systematic variation in eye movements over the time course of a choice or across the different choices. The only exceptions were finding more (of the same) eye movements when choice options were similar, and an emerging gaze bias in which people looked more at the gamble they ultimately chose. These findings are inconsistent with prospect theory, the priority heuristic, or decision field theory. However, the eye movements made during a choice have a large relationship with the final choice, and this is mostly independent from the contribution of the actual attribute values in the choice options. That is, eye movements tell us not just about the processing of attribute values but also are independently associated with choice. The pattern is simple—people choose the gamble they look at more often, independently of the actual numbers they see—and this pattern is simpler than predicted by decision field theory, decision by sampling, and the parallel constraint satisfaction model. © 2015 The Authors. Journal of Behavioral Decision Making published by John Wiley & Sons Ltd.

Now or Later? Attentional Processing and Intertemporal Choice

“…Additionally, we examined whether the first AOI (Reutskaja et al, ) or last AOI (Stewart, Hemrens & Matthews, ; Stewart, Gächter, & Noguchi, ; and cf. Krajbich & Rangel, ) examined predicted choosing the immediate option in each of the decision tasks per delay amount.…”

Attention in risky choice

“…When buying a car, it is the favored Volvo one plans to purchase rather than the Dacia one expects to decline that captivates the mind during the buying process. Intuitions such as these receive empirical support from sophisticated laboratory experiments showing that people allocate more attention to the alternative they will later choose than to the one they will decline (Glaholt & Reingold, 2009;Krajbich, Armel, & Rangel, 2010;Shimojo, Simion, Shimojo, & Scheier, 2003;Stewart, Hermens, & Matthews, 2013). This is not surprising, because the favored alternative is usually more important than the non-favored one.…”

“…In particular, we tried to accomplish the following: (a) Use of a neutral task format: All pieces of information have equal distances both between and within gambles. This is crucial and has often been violated (e.g., Fiedler & Glöckner, 2012;Glöckner & Betsch, 2008;Glöckner & Herbold, 2011;Koop & Johnson, 2013; but see Stewart et al, 2013;Su et al, 2013). Attention is controlled to a large degree by visual properties of the stimulus, such as salience of or distance between pieces of information.…”

“…Therefore, task format can strongly influence the number and sequence of fixations and stimulus construction must take such properties of the oculomotor system into account; (b) counterbalancing of stimulus layout: Gambles are presented vertically and horizontally to control for attentional effects such as reading direction; (c) minimizing the effect of short-term memory: Visual attention and short-term memory are closely linked. Some studies presented only four pieces of information that could be held in short-term memory (Koop & Johnson, 2013;Stewart et al, 2013). In this case there is no need for refixations and one should not expect that participants pay different amounts of attention to the different pieces of information, since each piece could be remembered.…”

“…In this case there is no need for refixations and one should not expect that participants pay different amounts of attention to the different pieces of information, since each piece could be remembered. To overcome this problem, we used eight pieces of information; (d) concentration on two opposing reasons: In all problems used in our experiment the minimum gains (losses) were zero, leaving two opposing reasons-each favoring a different gamble (see also Koop & Johnson, 2013;Stewart et al, 2013). This allowed for an unambiguous test of the process; (e) comparison of three validity measures: We related eye-tracking data to choices and verbal protocols and thereby tested the validities of these measures against each other.…”