Abstract
When given the opportunity to choose a visual search strategy (e.g. for color or shape), people often make suboptimal choices (i.e., yielding lower response time or accuracy). We have hypothesized this reflects a trade-off between minimizing effort and maximizing performance. If the most efficient strategy is cognitively demanding, some will choose a less optimal strategy, to reduce cognitive effort. It follows that increased motivation should increase one’s willingness to expend effort to perform optimally. We previously reported tentative support for this (Irons & Leber, 2017, VSS); participants performed a modified version of the Adaptive Choice Visual Search (ACVS; Irons & Leber 2016), a task designed to assess attentional control strategy. Specifically, participants receiving performance-contingent monetary reward – in which faster reaction times conferred higher payoffs -- were significantly more likely to adopt an optimal strategy, compared to a random-reward control group. However, subsequent experimentation yielded smaller effect sizes and nonsignificant results. Here, to determine with confidence whether reward improves search strategy, we ran a high-powered, high-N study, via Amazon Mechanical Turk (MTurk). Interestingly, a first experiment revealed floor-level rates of optimal performance, making it impossible to assess the effect of reward and also suggesting that MTurk participants are less strategic than in-lab participants. We ran a second experiment to boost baseline rates of optimal performance, by including instructions for the optimal strategy and a “preview” of the search array (without targets) to provide more time to choose the optimal strategy (see Hansen, Irons & Leber, 2019). Results confirmed that performance contingent reward significantly increased optimal strategy usage, compared to those in the random reward group. These results support that monetary reward exerts a motivational influence to push people away from effort minimization and toward performance maximization. Additional work will futher compare strategy use in MTurk vs in-lab participants.