Proceedings of the Annual Meeting of the Cognitive Science Society

People have a remarkable capacity to infer others’ goals and intentions based on how they behave. Yet, humans are also motivated to ensure that others can infer their mental states easily and accurately. Past work has shown that people achieve this by introducing inefficiencies to their behavior, which reveal its underlying goal (e.g., exaggerating one’s movements so as to make their purpose obvious). We hypothesized that inefficiency is not a constitutive feature of signaling, and that people will often signal their goals and intentions solely through efficient action. We test this idea in a signal-design experiment where participants need to reach an instrumental goal while also making that goal as inferable as possible. In line with our hypothesis, people shape their behavior to increase inferability without jeopardizing efficiency (Experiment 1). Using a computational model, we show that these efficient signals are well-designed to guide observers’ inferences about the relevant instrumental goal. Moreover, observers’ intuitions about which paths were produced to signal correlate with the proportion of times that the paths were generated in the signaling condition of our first experiment (Experiment 2). Our results show that humans not only exploit opportunities to reveal their goals without deviating from efficient action, but that these signals allow observers to understand the instrumental and signaling goals underpinning the movement.