Chris has an excellent post at "Mixing Memory" on causal reasoning, looking at attempts in cog. sci. to model causal reasoning.
Some philosophical work on the topic that I've found at least thought-provoking is the generally unpopular (but also always misunderstood) manipulability theory. Manipulability theory cannot, I think, be a full account of causation, nor can it be a strict account of what the nature of causation is (this review of Woodward gives one reason why); nonetheless, I think it captures an important aspect of the way we reason about causes that is usually neglected, or, at least, used to be (it's recently become something of a hot topic due to people like Woodward). This (PDF) is an interesting defense of manipulability theory (a.k.a. agency theory) against some common objections.
I've previously recommended some of the work of Eric Hiddleston, especially his paper, Causal Powers (PDF). Hiddleston is an advocate of the probabilistic-contrast-based causal power theory Chris mentions in his post.
