People have a keen ability to reason about others ’ mental states, which is central for communication and cooperation. A core question for cognitive science is what mental representations support this ability. We offer one proposal based on the framework of influence diagrams, an extension of Bayes nets that is suited for representing intentional goal-directed agents. We evaluate this framework in two experiments that require participants to make inferences about what another person knows or values. In both experiments, participants ’ judgments were better predicted by our influence diagrams account than by several alternative accounts.

A dynamic model of a multiagent system defines a probability distribution over possible system behaviors over time. Alternative representations for such models present tradeoffs in expressive power, and accuracy and cost for inferential tasks of interest. In a history-dependent representation, behavior at a given time is specified as a probabilistic function of some portion of system history. Models may be further distinguished based on whether they specify individual or joint behavior. Joint behavior models are more expressive, but in general grow exponentially in numberofagents. Graphical multiagent models (GMMs) provide a more compact representation of joint behavior, when agent interactions exhibit some local structure. We extend GMMs to condition on history, thus supporting inference about system dynamics. To evaluate this hGMM representation we study a voting consensus scenario, where agents on a network attempt to reach a preferred unanimous vote through a process of smooth fictitious play. We induce hGMMs and individual behavior models from example traces, showing that the former provide better predictions, given limited history information. These hGMMs also provide advantages for answering general inference queries compared to sampling the true generative model.

Factored models of multiagent systems address the complexity of joint behavior by exploiting locality in agent interactions. History-dependent graphical multiagent models (hG-MMs) further capture dynamics by conditioning behavior on history. The challenges of modeling real human behavior motivated us to extend the hGMM representation by distinguishing two types of agent interactions. This distinction opens the opportunity for learning dependence networks that are different from given graphical structures representing observed agent interactions. We propose a greedy algorithm for learning hGMMs from time-series data, inducing both graphical structure and parameters. Our empirical study employs human-subject experiment data for a dynamic consensus scenario, where agents on a network attempt to reach a unanimous vote. We show that the learned hGMMs directly expressing joint behavior outperform alternatives in predicting dynamic human voting behavior, and end-game vote results. Analysis of learned graphical structures reveals patterns of action dependence not directly reflected in the original experiment networks.

People are amateur social psychologists: they explain other people’s behavior, infer what other people are thinking and feeling, and predict how other people will act. I will refer to this sort of psychologizing as social reasoning in order to highlight the fact that it involves reasoning about people. Social reasoning often requires significant leaps of inductive inference: people infer others ’ mental states, such as their preferences, goals, and beliefs, from relatively sparse information, such as others ’ choices and actions. The capacity to reason about mental states and about how mental states relate to behavior is often referred