Matt Dillahunty and disbelief by default

In his recent talk at the American Atheist convention on skepticism and atheism, Matt Dillahunty states (at about five minutes in) that skepticism does tell us what to believe in the case of untestable claims--that the default position is disbelief.

But no, the default position has to be nonbelief, not disbelief.  To disbelieve in a proposition is to believe in the negation of the proposition, to believe that the original proposition is false.  And Dillahunty already said that (a) we should proportion our belief to the evidence and that (b) the proposition in question is untestable, meaning there is no evidence for or against it.

The position he describes is logically inconsistent.

We know that there are untestable propositions that are true.  We shouldn't believe that they are false simply because they are untestable. We should only believe they are false if we have good reasons to believe they are false; in the absence of that we should be agnostic.

(Added 5:36 p.m.: What are the implications for the above argument if it is the case that untestability does not entail lack of evidence or reasons?  What about if we distinguish evidential from non-evidential reasons?  And if we take the latter course, what does that say about proposition (a), above? Left as an exercise for commenters.)