Abstract

Most deontologists claim that it is more objectionable to do harm than it is to allow harm of comparable magnitude. I argue that this view faces a largely neglected puzzle regarding the moral relevance of an agent's past behavior. Consider an agent who chooses to save five people rather than one, where the one person's life is in jeopardy because of something the agent did earlier. How are the agent's obligations affected by the fact that his now letting the one die would retroactively make it the case that he has killed? I argue that the most promising deontological responses to this question are difficult to defend. Further, it is unclear how the deontologist can resolve a further set of cases, in which commonsense intuition indicates that an agent is required to terminate a threat she initiated more recently in preference to a threat she initiated less recently.