It is commonplace to think that the amount of force that can permissibly be used to repel culpable threateners is greater than that which can be used to repel innocent threateners (for instance child soldiers, innocently mistaken, brainwashed or coerced agents). However, for a threatener's culpability to be relevant to the amount of force that can be used to repel her, it is standardly thought that the threatener's culpable responsibility must appropriately relate to the threat she is harmed to avert:  whether a person who is now threatening you culpably injured someone else a week ago should not have bearing on how much harm you can inflict on her to avert the unrelated threat she is currently posing. I refer to this as the relevance constraint on culpability.
While granting the intuitive appeal of the relevance constraint, I will in this talk challenge its justification. I will argue that the most consistent view on the relevance of culpability to liability to preventive harm is to accept that non-related culpability can affect a person's liability to harm in a different situation from that in which the culpability arose. I argue that, in principle, a person can through culpable acts temporarily and partly exile herself from a moral community, during which her claim on protection from preventive harm is globally reduced.  I will address several objections to this draconian conclusion and argue that that the cost of rejecting it would be to reject the role of culpable moral responsibility to the justification of self-defence altogether.