A detailed model of the outside world is an essential ingredient of human cognition, enabling us to navigate, form goals, execute plans, and avoid danger. Critically, these world models are flexible—they can arbitrarily expand to introduce previously-undetected objects when new information suggests their presence. Although the number of possible undetected objects is theoretically infinite, people rapidly and accurately infer unseen objects in everyday situations. How? Here we investigate one approach to characterizing this behavior—as nonparametric clustering over low-level cues—and report preliminary results comparing a computational model to human physical inferences from real-world video.

People often make decisions with the goal of gaining information which can help reduce their uncertainty. However, recent work has suggested that people sometimes do not select the most diagnostic information queries available to them. A critical aspect of information search decisions is evaluating how obtaining a piece of information will alter a learner’s beliefs (e.g., a piece of information that is redundant with what is already known is useless). This suggests a close relationship between information seeking decisions on one hand, and belief updating on the other. This paper explores the deeper relationship between these two constructs in a causal intervention learning task. We find that patterns in belief updating biases are predictive of decision making patterns in tasks where people must make interventions learn about the structure of a causal system.

Question asking is a powerful means by which humans learn. However, asking a question requires searching through a massive space of possible questions to find a single question that is relevant and informative. How do humans efficiently accomplish this task? Drawing on prior research on other decision problems, we propose that the search for new questions is constrained by those encountered in the past, so that people frequently reuse questions (or parts of questions) rather than generating new questions "from scratch." We find empirical support for this prediction, and we find that this "question reuse" has consequences for the informational value of people's questions. Taken together, this research sheds new light on the mechanisms behind human question asking abilities and, more generally, how we narrow down a large space of possibilities to find a single solution.

Inspired by notions of intrinsic motivation (Schmidhuber, 2010) and play as proposing and solving arbitrary problems (Chu and Schulz, 2020b), we report initial progress toward computational modeling of playful goal generation. We create an embodied, 3D environment resembling a child's room, and ask study participants to play in the environment and then create a scorable game. We propose to model games using a domain-specific language, which represents each game as a computer program. These programs act as reward-generating functions, mapping states visited by an agent as they play a game to the score they should receive. We then analyze our corpus of program representations to highlight four key aspects of human games that would contribute to constructing effective computational models of game generation: creativity, compositionality, common sense, and context sensitivity.

Helping is a universal human behavior, and is a core aspect of a functioning society. However, the decision to provide help, and what type of help to provide, is a complex cognitive calculation that weights many costs and benefits simultaneously. In this paper, we explore how various costs influence the moment-to-moment decision to help in a simple video game. Participants were paired with another human participant and were asked to make repeated decisions that could benefit either themselves or their partner. Several preregistered manipulations altered the cost each person paid for actions in the environment, the intrinsic resource capacity of individuals to perform the task, the visibility of the other player's score, and the affordances within the environment for helping. The results give novel insight into the cost-benefit analyses that people apply when providing help, and highlight the role of reciprocity in influencing helping decisions.

Learning traps are false beliefs that entrench themselves by discouraging the exploration required to correct them. In previous lab experiments, these learning traps have proven remarkably difficult to prevent. Here, we investigate whether learning traps remain stable in contexts in which both individual and social learning are possible. In two of our three experiments, we found that learners trapped by a false belief were significantly more likely to escape a learning trap when they were able to observe another decision-maker's choices (without observing their outcomes). However, social learning was not a panacea. Social learning was constrained by the challenge of inferring others' beliefs, and trapped learners struggled to learn from partners with sub-optimal decision rules, even when their partner's choices were informative. Collectively, these results suggest that while social learning can help overcome the limits of individual learning, learning from others comes with its own challenges and limitations.