Let us say that we are contemplating some future event. Obviously there are some current factors that we can single out as causes of that event. Are there also past events that we can pick out as causes, or are we limited strictly to talking about what currently exists when we discuss causation?*
The answer is yes and no. Specifically it depends on how we wish to deal with causation. It is generally accepted that causation can be dealt with through counterfactuals; usually of the form: if it is the case that without some factor, X, Y wouldn’t occur then we say that X is a cause of Y. Of course things can get a little more complicated when several causes are coming together to produce some result (a discussion of which is here), but such difficulties have no bearing on our discussion here. And when we investigate causation through counterfactuals we also assume that everything else is held constant, and this is where the difficulty with historical causes arises. One way to interpret talk of historical causes is to assume that only that past moment in time is altered in our counterfactual considerations, with subsequent moments remaining unchanged. Obviously such an interpretation would mean that the historical events could never be considered causes, since the current state of affairs wouldn’t be affected, and thus the future event, Y would not be affected either. Or we can understand talk of historical causes as not holding the rest of the past constant. Instead we can assume that what we mean by our counterfactual claim is that the event in question and the events that resulted from it are allowed to change. In that case some historical events can legitimately be said to be causes, since a change in that historical event would result in a sequence of changes, possibly including a change in the future event under consideration.
I think the second interpretation is often the better one, since we usually accept that if A causes B, and B causes C then A can be said to be a cause of C. What is important to note though is that a historical event can only be said to be a cause of some future event if it is part of a causal chain that includes some present events. In general we can say that if an event at time t1 is to be a cause of an event at t3, and if t2 is between t1 and t3, then the event at t1 must be a cause of some events at t2. This conclusion is not a metaphysical thesis about the nature of causation, certainly I can imagine universes with laws that allow an event at one time to cause effects at future times without intermediate effects. However, our world is not like this, as far as we can tell future moments depend exclusively on the immediately previous moment.
And this makes talking about historical causes a bit tricky. Specifically, when wondering whether historical causes are epiphenomenal. As I pointed about above there is a valid way of talking about them in which they aren’t, but that is because they take part in a chain of events. So precisely speaking it is the causal chain that is not epiphenomenal, not the historical events themselves. If we were considering the historical events alone, outside of the causal chain, then we would have to say that they are epiphenomenal. And this is relevant to a particular kind of externalism, that identifies mental content with certain historical factors resulted in adaptations that allowed the creature to represent the world. Certainly the causal chain including these factors is not epiphenomenal, but the historical factors by themselves are. We can imagine a spontaneously generated duplicate of some individual, and since it would act in the same way as the original, we can conclude the historical factors by themselves have no effect. Properly speaking then it is only the current state of the individual that is a cause, historical factors can only said to be a cause in virtue of their effects upon this current state. And thus an externalism that ties representation to historical factors is still epiphenomenal, since those historical factors only have an effect in virtue of current properties.
(This is a more detailed discussion of one part of my response to Dretske’s externalism, here.)
* For simplicities sake I have framed this discussion in the “current tense”, i.e. looking at the causes of a future event from some present moment, with consideration given to past events. It is trivial to generalize this to talk about effect at some moment, causes in the immediately previous moment, with consideration given to possible causes at previous moments.