To glean information from social networks, people must be adept at distinguishing real evidence from hearsay. Here, we investigate human inferences within a simulated social network. We introduce a novel social learning task in which participants infer their surrounding social network structure while also using social communications to aid judgments about the shared environment. We find that the majority of adults correctly identified independent and acyclic network structures, though they struggled to identify cyclic communications. Comparison of judgments with several social inference models supports a naïve social learning account that integrates social evidence with the learner’s own observations in a structure insensitive way. This suggests that people are capable of using communication patterns to identify social influences, but they may still be misled by the distortions of evidence that these network dynamics can produce.

As AI systems like language models are increasingly integrated into decision-making processes affecting people's lives, it's critical to ensure that these systems have sound moral reasoning. To test whether they do, we need to develop systematic evaluations. We provide a framework that uses a language model to translate causal graphs that capture key aspects of moral dilemmas into prompt templates. With this framework, we procedurally generated a large and diverse set of moral dilemmas---the OffTheRails benchmark---consisting of 50 scenarios and 400 unique test items. We collected moral permissibility and intention judgments from human participants for a subset of our items and compared these judgments to those from two language models (GPT-4 and Claude-2) across eight conditions. We find that moral dilemmas in which the harm is a necessary means (as compared to a side effect) resulted in lower permissibility and higher intention ratings for both participants and language models. The same pattern was observed for evitable versus inevitable harmful outcomes. However, there was no clear effect of whether the harm resulted from an agent's action versus from having omitted to act. We discuss limitations of our prompt generation pipeline and opportunities for improving scenarios to increase the strength of experimental effects.

In both professional domains and everyday life, people mustintegrate their own experience with reports from social networkpeers to form and update their beliefs. It is therefore importantto understand to what extent people accommodate the statis-tical dependencies that give rise to correlated belief reportsin social networks. We investigate adults’ ability to integratesocial evidence appropriately in a political scenario, varyingthe dependence between the sources of network peers’ beliefs.Using a novel interface that allows participants to express theirprobabilistic beliefs visually, we compare participants against anormative Bayesian standard. We find that they distinguish thevalue of evidence from dependent versus independent sources,but that they also treated social sources as substantially weakerevidence than direct experience. The value of our elicitationmethodology and the implications of our results for modelinghuman-like belief revision in social networks are discussed.

Humans are remarkably adept at inferring the causes of events in their environment; doing so often requires incorporating information from multiple sensory modalities. For instance, if a car slows down in front of us, inferences about why they did so are rapidly revised if we also hear sirens in the distance. Here, we investigate the ability to reconstruct others' actions and events from the past by integrating multimodal information. Participants were asked to infer which of two agents performed an action in a household setting given either visual evidence, auditory evidence, or both. We develop a computational model that makes inferences by generating multimodal simulations, and also evaluate our task on a large language model (GPT-4) and a large multimodal model (GPT-4V). We find that humans are relatively accurate overall and perform best when given multimodal evidence. GPT-4 and GPT-4V performance comes close overall, but is very weakly correlated with participants across individual trials. Meanwhile, the simulation model captures the pattern of human responses well. Multimodal event reconstruction represents a challenge for current AI systems, and frameworks that draw on the cognitive processes underlying people's ability to reconstruct events offer a promising avenue forward.