A deviant causal chain or wayward causal chain is any situation in which a casual relationship between two events has unexpected implications due to a deviant chain of intermediate events connecting them. Deviant causal chains were first defined by Donald Davidson in the sixties to provide counterexamples to certain naïve definitions of agency. Counterexamples to the tripartite theory of knowledge, called Gettier problems, may also be thought of as deviant causal chains.
Deviant causal chains may pose practical problems when interpreting legal liability. They can blur the line between negligence and intentional crime, and between attempted and completed crimes. Consider the case of a murderer who invites his victim to dinner, intending to serve her a poisoned bowl of noodles. The victim dies in a car crash while rushing to the murderer's home. The murderer's intention was for the victim to die, and his actions resulted in her death, yet the intervening causal chain does not appear intentional.
Deviant causal chains can also be used to pick holes in causal theories of perception. Such theories might hold that, if an experience matches the criteria that (a) it must correspond to a real object present to the perceiver, and (b) it must actually be caused by the presence of that object, then it cannot be a hallucination and must be a case of healthy perception. The (b) criterion of this definition seems strong enough to avoid the problem of veridical hallucinations, that is, experiences that happen to correspond with reality, but only as an accidental side-effect of hallucination, as these experiences would not be brought about as an effect of their objects being real.
However, it is still possible to devise a thought experiment that reintroduces this problem through deviant causal chains. Davidson gives the example of a wizard who is able to trigger unpredictable hallucinations in his victims. If you encountered this wizard, and he cast his spell on you, the same wizard would have a chance of reappearing within the ensuing hallucination. In this case, (a) your experience of a wizard would correspond to the real wizard in front of you, and (b) the experience would be caused by the real presence of the same wizard and his magical powers. So, the experience would fit the definition of healthy perception, despite still clearly being a hallucination. Thus some criterion should be added to the defintion to distinguish between deviant and appropriate causal chains.
Deviant causal chains can be found in several other contexts apart from those given here, usually as a contrivance or thought experiment to undermine an attempted causal defintion for some kind of action or state of being.