How Do People Generalize Causal Relations over Objects? A Non-parametric Bayesian Account

Abstract

How do people decide how general a causal relationship is, in terms of the entities or situations it applies to? What features do people use to decide whether a new situation is governed by a new causal law or an old one? How can people make these difficult judgments in a fast, efficient way? We address these questions in two experiments that ask participants to generalize from one (Experiment 1) or several (Experiment 2) causal interactions between pairs of objects. In each case, participants see an agent object act on a recipient object, causing some changes to the recipient. In line with the human capacity for few-shot concept learning, we find systematic patterns of causal generalizations favoring simpler causal laws that extend over categories of similar objects. In Experiment 1, we find that participants’ inferences are shaped by the order of the generalization questions they are asked. In both experiments, we find an asymmetry in the formation of causal categories: participants preferentially identify causal laws with features of the agent objects rather than recipients. To explain this, we develop a computational model that combines program induction (about the hidden causal laws) with non-parametric category inference (about their domains of influence). We demonstrate that our modeling approach can both explain the order effect in Experiment 1 and the causal asymmetry, and outperforms a naïve Bayesian account while providing a computationally plausible mechanism for real-world causal generalization.

Actively learning to learn causal relationships

“…Consequently, they may have resorted to the same deterministic-and disjunctive-favoring prior in the next task, as is consistent with the No-Transfer model. Such behavior is also consistent with how people generalize causal laws across several tasks (Zhao et al, 2022). The alternative behavior would be to expect a high degree of similarity in our experiment's training and transfer tasks and thus find it useful to transfer overhypotheses between these tasks, as is consistent with our full model.…”

Actively Learning to Learn Causal Relationships

“…Consequently, they may have resorted to the same deterministic-and disjunctivefavoring prior in the next task, as is consistent with the No-Transfer model. Such behavior is also consistent with how people generalize causal laws across several tasks (Zhao et al, 2022). In other words, they may not have taken the high degree of superficial similarity between our experiment's training and transfer tasks to be a strong indicator that the underlying causal relationships would have the same form, contra our full model.…”

Counterfactuals and the logic of causal selection.

“…A large literature on causal cognition has explored how people learn causal facts about the world (e.g., Bramley et al, 2015, 2017; Cheng, 1997; Gopnik et al, 2004; Griffiths & Tenenbaum, 2005, 2009; Lucas & Griffiths, 2010; Zhao et al, 2021) or how they infer whether an event would have happened in the absence of another event (Ahn et al, 1995; Gerstenberg et al, 2021; Kelley, 1973; Stephan et al, 2020). Here, we are concerned with a different problem.…”