Say that p is prior to q in the order of explanation provided that p enters into some explanation of q. One might think that a circle in the order of explanation is impossible. (Some background: Robert M. Adams has once constructed an elegant argument against Molinism that, among other premises, assumed that there were no circles in the order of explanation. William Lane Craig, however, has responded by arguing that circles in the order of explanation are quite possible, but the examples of his that I've seen I've found unconvincing.)
Suppose I promise that if this year you make a donation either to CRS or Caritas of Waco, I will this year make a donation to CRS. You make a donation to Caritas. I thus make a donation to CRS. This example inspires you, and you respond with a donation to CRS yourself. Let p be the proposition that you made a donation to CRS or Caritas. Let q be the proposition that I made a donation to CRS. Let r be the proposition that you made a donation to CRS. Then, p helps explain q, and q helps explain r. But r is a disjunct in p, so it seems q helps explain p. Thus, it seems, p is explanatorily prior to q and q is explanatorily prior to p.
Is this a good argument? I am not so sure. One problem is that the thesis that the truth of a disjunct a explains the truth of a disjunction a-or-b may not be true when a is itself explained by b.
But suppose one accepts the example. Is there still a way of capturing the intuition behind the idea that there can't be circular explanations? I think so, but it will take some hard work. And that I want to leave for another post.
Note, also, that the above example fits with a principle I've defended that if there is a circle of explanations, then the circle also has an explanation from beyond the circle. (If this principle is true, then Adams can probably regroup and defend his anti-Molinist argument.) In this case, your initial choice to make a donation to Caritas together with my promise are an explanation for the circle.