In their 1994 paper, “Conditioning and Intervening,” Christopher Meek and Clark Glymour use the interventionist approach to causal modeling to articulate a novel way of understanding the rift between causal decision theorists and evidential decision theorists. Specifically, they claim that we can redress the dispute such that “[t]he difference between the two does not turn on any difference in normative principles, but on a substantive difference about the causal processes at work in the context of decision making—the causal decision theorist thinks that when someone decides… an intervention occurs, and the ‘evidential’ decision theorist thinks otherwise.”

In this paper, I adopt Meek and Glymour’s framework and argue that neither causal decision theory nor evidential decision theory issues rational verdicts across all decision-making contexts. More specifically, I argue that in addition to contexts in which agents should regard their decisions as interventions, there are decision-making contexts in which agents should regard themselves as incapable of intervention, as well as contexts in which agents should be uncertain about whether their decisions constitute interventions. My strategy is to argue (i) that the details of Newcomb’s Problem can be specified such that the subject should regard it as impossible for her to intervene, (ii) that the details of Newcomb’s Problem can be specified such that subject should be uncertain whether she is in a position to intervene, and (iii) that each of these specifications constitutes a genuine decision problem. If I am right, then it is important that our decision theory issues rational verdicts not only when an agent is certain that she can intervene, but also when an agent is certain that she cannot intervene, and when an agent has some degree of belief in her capacity to intervene. My last task in this paper is to propose and defend a decision rule within the interventionist framework that delivers such verdicts.