No Evidence that Reversibility Affects Causal judgments in Late-preemption Cases

Abstract

Recently, Ross and Woodward (2022) argued that the reversibility of an outcome—that is, whether the outcome can be undone—affects causal judgments. One prediction of their account is that reversibility affects causal judgments in late-preemption scenarios, where people typically judge that events that produce the outcome earlier are more causal than preempted alternative events that would have otherwise produced the outcome. Ross and Woodward’s account predicts that when the outcome is reversible, people would judge these events similarly—as in standard overdetermination cases. In three experiments, we found no evidence that reversibility affects causal judgments in late-preemption cases. We discuss how these results affect the experimental philosophy and cognitive science of causation, causal modeling, and experimental jurisprudence.

Reasoning about actual causation in reversible and irreversible causal structures.

“…One reason why Henne et al (2023) might not have detected the predicted effect is that the test question about the second cause that they used in their experiments was not optimal to find an effect. In their Experiments 1 and 2, the statements for both causes that participants evaluated were “The purple light went on because David turned on the red [blue] switch.” As has been described earlier, Ross and Woodward (2022) assume that the second cause in reversible structures might be regarded as a maintainer, not as a trigger.…”

“…The present article addresses the empirical question of how lay people’s intuition about actual causation in preemption scenarios changes depending on causal reversibility. In standard scenarios of late preemption that play out in irreversible structures, people reliably have been found to say that only the event that occurs first makes an actual causal contribution to the effect (see, e.g., Gerstenberg et al, 2021; Henne et al, 2023; Stephan et al, 2020; Walsh & Sloman, 2011), while the one occurring second (or later in the case of more than two potential causes) is regarded as noncausal. How might reasoners’ causal intuition change if such a scenario were to play out on the stage of a reversible causal structure?…”

“…There is so far, to the best of the author’s knowledge, only one set of empirical studies that addresses the question of how reversibility might affect people’s causal intuitions in preemption scenarios. This set of studies, comprising three main studies and one supplementary study, was conducted by Henne et al (2023). In all experiments, the authors used the light switch scenario introduced above.…”

“…Does this mean that causal reversibility does not influence people’s causal intuitions in preemption scenarios? Henne et al (2023) were careful in the interpretation of their results, though. They report that they failed to find evidence for Ross and Woodward’s hypothesis, but they did not claim to have refuted it.…”

“…Across three main experiments and one supplementary study ( N = 590), it is shown that reversibility has the predicted effect: later causes are perceived to make an actual causal contribution to the effect. It is also shown that Henne et al (2023), in a first study, did not find evidence for Ross and Woodward’s hypothesis because they did not test whether people regard later causes in preemption-like sequences of reversible structures as maintainers and not as triggers of their effect. Because they used test questions that asked explicitly for triggering rather than maintaining or were at least ambiguous, their results seemed to show that people think that later events have no causal impact.…”