Changing Minds ― Epistemic Interventions in Causal Reasoning

Abstract

Did Tom’s use of nuts in the dish cause Billy’s allergic reaction? According to counterfactual theories of causation, an agent is judged a cause to the extent that their action made a difference to the outcome (Gerstenberg, Goodman, Lagnado, & Tenenbaum, 2020; Gerstenberg, Halpern, & Tenenbaum, 2015; Halpern, 2016; Hitchcock & Knobe, 2009). In this paper, we argue for the integration of epistemic states into current counterfactual accounts of causation. In the case of ignorant causal agents, we demonstrate that people’s counterfactual reasoning primarily targets the agent’s epistemic state – what the agent doesn’t know –, and their epistemic actions – what they could have done to know – rather than the agent’s actual causal action. In four experiments, we show that people’s causal judgment as well as their reasoning about alternatives is sensitive to the epistemic conditions of a causal agent: Knowledge vs. ignorance (Experiment 1), self-caused vs. externally caused ignorance (Experiment 2), the number of epistemic actions (Experiment 3), and the epistemic context (Experiment 4). We see two advantages in integrating epistemic states into causal models and counterfactual frameworks. First, assuming the intervention on indirect, epistemic causes might allow us to explain why people attribute decreased causality to ignorant vs. knowing causal agents. Moreover, causal agents’ epistemic states pick out those factors that can be controlled or manipulated in order to achieve desirable future outcomes, reflecting the forward-looking dimension of causality. We discuss our findings in the broader context of moral and causal cognition.

No privileged link between intentionality and causation: Generalizable effects of agency in language

“…Perceived agency has previously been found to influence people's causal judgments (see e.g., Kirfel & Lagnado, 2021b;Lombrozo, 2010;Rose, 2017;Schwenkler & Sytsma, 2020), suggesting that reasoning about how much agency was involved is part of how people understand what qualifies as cause of a given outcome. Yet here we find that this phenomenon-the effect of perceived agency-may actually be far more general than causal cognition.…”

“…The extent to which people think that Tom (or any animate agent) caused a particular outcome depends in part on whether or not Tom was exercising his own agency in the way that he affected the outcome (see e.g., Kirfel & Lagnado, 2021a, 2021bLagnado & Channon, 2008;Lombrozo, 2010;Rose, 2017;Schwenkler & Sytsma, 2020). Here, it seems straightforward to say that "Tom caused train delay" only if Tom was acting intentionally when he ended up being involved in the train delay.…”

“…There are different theories as to what exactly explains this effect of perceived agency, but these theories share a key perspective: We should seek to understand the role of perceived agency in shaping people's thinking in order to better understand causal judgments. Accordingly, research has discovered many important dimensions to these effects on causal judgments: for instance that perceived agency affects evaluation of causal judgments that involve physical contact and those that don't (Lombrozo, 2010), that intentional actions are judged to be both more causal and more blameworthy (Lagnado & Channon, 2008), and that reasoning about agents' mental states may factor into how people identify relevant counterfactuals to causal statements (Kirfel & Lagnado, 2021a, 2021b. On these and related views, there is a clear motivation for why one might be interested in understanding the effect of perceived agency-in order to better understand causal cognition.…”

The Pervasive Impact of Ignorance

“…The vignette leaves open whether there was an opportunity for the captain to learn about the secret passengers in the cargo compartment. This has likely led to assumptions about the general immutability of the agent's ignorance in a number of scenarios (Kirfel & Lagnado, 2021b), and subsequently affected people's counterfactual reasoning about epistemic states.…”

“…One challenge for these theories, however, is to explain how and why these epistemic states play the correct role in shaping counterfactual reasoning. Some progress on this has been made in the domain of causal judgments (Kirfel & Lagnado, 2021b;Kominsky & Phillips, 2019). Kominsky and Phillips (2019) argue that the epistemic state of a norm-violating agent modulates the perceived abnormality of their action.…”