Cue transparency leads to a myopic decision habit: Optimal training ratios to avoid overshadowing

Abstract

Environmental cues to optimal decisions abound. Some are highly predictive and immediately transparent (e.g., amounts) whereas others may either be harder to discern or take time to obtain (e.g., probability and delay). In the present experiment, we manipulated the availability of a particularly salient, and thus dominant, cue regarding when a participant should cash-in when an outcome’s value is increasing over time – a target cube’s size grew as its value grew. For some targets, this expanding cube cue was visible but for others it was not. In contrast, a highly predictive but less obviously relevant color cue was omnipresent, and the recent delay-reward relationship also signaled the reward dynamics. The research addressed a simple question: would more training with transparent targets that reveal the dominant cue produce greater sensitivity to a changing contingency or would more training with opaque targets produce greater sensitivity? Participants’ contingency sensitivity benefitted more when the majority of the training involved opaque targets. Although extensive training with transparent targets produced a strong understanding of reward dynamics, this dominant decision cue undermined attention to secondary cues that are more important in less transparent environments.