Abstract
Attainment of long-term goals requires good plans. Given limited cognitive resources, challenges arise as planning can be extremely complex and even infeasible when facing an uncertain environment and multiple conflicting goals. To explain how humans make plans in such cases, we demonstrate that intention, an essential mental state for decision-making under multiple conflicting goals, constrains human planning hierarchically. We designed a task requiring hierarchical planning, a process to achieve the final goal by executing a serial of sub-plans (e.g., getting a key to unlock a box to seek treasures in it). In a 2D navigation game, participants were instructed to open one of two treasure boxes with the prerequisite of obtaining a key to the box, while the key had a possibility of malfunctioning. In cases where the key failed, participants had to re-plan and choose a new key to fulfill the task. If sequential actions are constrained by intentions, or to say, a commitment was made to one’s choice through planning, participants would tend to acquire a key that matched their initially intended box, no matter whether the choice is optimal (getting another key to the alternative box could be a better choice). Across three experiments (N = 150), people favorably chose the key with consistency towards their intended box (Exp.1). This bias remained after controlling the color similarity between the keys (Exp.2). Furthermore, an increase in re-planning time did not significantly alter humans’ planning pattern (Exp.3). Collectively, these findings show that humans construct follow-up sub-plans with consistency, revealing a spontaneous commitment to intentions in hierarchical planning with conflicting goals. The study may provide insights into understanding human planning as automatic intention-constrained beyond optimal resource-constrained.