Learning looks different from doing

Whereas some actions are aimed at changing the world, others are aimed at learning about it. For example, someone might press on a door to open it, or to determine whether it is locked; someone might shake a container to shuffle its contents, or to figure out what’s inside. The distinction between ‘pragmatic’ and ‘epistemic’ actions is recognized in other fields—but do people naturally make this distinction when observing others? Here, 4 experiments show that naive observers infer the pragmatic or epistemic intent of another person’s action, in ways that suggest hierarchical organization of goal representations. We filmed naive subjects playing a ‘box-shaking’ game with either an epistemic goal (e.g., determining the number of objects inside an opaque box) or a pragmatic goal (e.g., shuffling the box’s contents). Then, separate subjects observed these videos and inferred the actors’ goals: Who was trying to learn something, and who was trying to do something? Regardless of whether they were informed (Experiment 1) or uninformed (Experiment 2) of candidate actions, observers distinguished pragmatic from epistemic goals, based purely on the box-shaking dynamics. Moreover, when asked to label each video’s corresponding action, ‘swapping’ errors revealed that pragmatic actions were more confusable with one another, as were epistemic actions—a hallmark of hierarchical processing (Experiment 3). Finally, these effects emerged with free responses (Experiment 4). These results suggest that observers are sensitive not only to an actor’s specific goal, but also that goal’s ‘type’: Learning looks different from doing — and people can tell.