From PhilPapers forum Philosophy of Mind:

2012-08-08
What would count as an explanation?
Reply to Daniel Davis
Explanation names a causal chain and identifies its events. Causally related physical events, and first, intermediate, and last events, cannot be identified except through the pragmatic needs that serve  "explanation".

So, for example, the causal chain that is named "our communication" identifies the events "my writing a post" and "your replying to it". Traditionally, these are called first and last events, but this imposes a temporal, ontological necessity above and beyond the atemporal necessity of the identifying condition ("our communication") that is an explanation.

Temporally necessitated causal chains have always been problematic (as Kant tried to show us; and modern science has yet to tackle the logic of a first event). They are a problem because they create new, hidden, causal chain branches. For example, "our communication" identifies the atemporal events "my writing a post" and "your replying to it". But if we want to construe these events as temporal "first" and "last" events then we seek a new explanation that identifies a new, branching, hidden chain whose event(s) is associated with either of the events "my writing a post" and "your replying to it" and which names either one of them "first" or "last". This naming establishes a new explanation.

I might summarise this by saying that explanation is an epistemological identifying (or transcendental) condition (not an ontological or psychological identifying condition) that names a causal chain and identifies its events. Explanations that are given through a temporal necessity are compound associations of explanatory chains.