People accurately predict the transition probabilities between actions

https://www.science.org/doi/10.1126/sciadv.abd4995

Social life is a complex dance. To coordinate gracefully with one’s partners, one must predict their actions.

Here, we investigated how people predict others’ actions. We hypothesized that people can accurately predict others’ future actions based on knowledge of their current actions, coupled with knowledge of action transitions.

To test whether people have accurate knowledge of the transition probabilities between actions, we compared actual rates of action transitions—calculated from four large naturalistic datasets—to participants’ ratings of the transition probabilities between corresponding sets of actions.

In five preregistered studies, participants demonstrated accurate mental models of action transitions. Furthermore, we found that people drew upon conceptual knowledge of actions—described by the six-dimensional ACT-FASTaxonomy—to guide their accurate predictions.

Together, these results indicate that people can accurately anticipate other people’s moves in the dance of social life and that the structure of action knowledge may be tailored to making these predictions.

We use these five datasets to estimate the “ground truth” probabilities of how actions actually transition from one to the next. In five preregistered studies, we then compare these ground truth measures of transition probabilities to people’s predictions of the same transition probabilities. In each study, participants rate the likelihood of a future action, given a current action. For example, “how likely is someone to start running, given that they are currently stretching?” or “how likely is some to start dancing, given that they are currently laughing?” We measure the accuracy of participants’ judgments by correlating their ratings with the ground truth measures of the same transitions.

We have previously proposed a theoretical framework for predictive social cognition, which suggests that social knowledge is organized by the goal of prediction (32). That is, the way in which people organize their conceptual knowledge of actions should allow them to make accurate action predictions. In previous research, we found that six psychological dimensions—together forming the Abstraction, Creation, Tradition, Food, Animacy, and Spiritualism Taxonomy (ACT-FAST; Table 1)—capture much of people’s conceptual knowledge of actions (33). Each action can be located on this “map” of actions based on its relation to each dimension. For example, the actions “stir” and “peel” load high on the food dimension; the action “dance” loads low on the food dimension, but high on the creation and animacy dimensions. Actions located closer to each other on these dimensions are more conceptually similar to each other. The ACT-FAST dimensions also describe how people infer the real-world properties of actions (e.g., where, why, when, how, and by whom different actions are performed) and predict patterns of brain activity elicited by perceiving actions (33, 34).