I don't have time to find caselaw, but the impression I've always had is that the answer is at least conditionally yes. This is because the specifications and terms came from one original application.
So if in B a claim term like "widgiedigit" was argued by the applicant to always mean "a generally linear articulated structure capable of bending at least 45 degrees at two locations along its length", then the public should be able to rely on this absolutist statement for the definition of "widgiedigit" in the claims of B and A.
But in freedom to operate practice you seldom find absolutist statements hanging like ripe cherries low on a branch, just waiting for you to pluck.
Contrast instead where the specification for A and B describes multiple embodiments of the invention, and where "widgiedigit" has slightly different uses or meanings depending on differing embodiments of the invention. Then, if in A the claim containing "widgiedigit" was generally directed at embodiment-1, but in B the claim is directed more to embodiment-2, the patentee at least has wiggle room to argue that the construction of "widgiedigit" in A is not limited to that definition given during prosecution of B.
As always, corrections, amendments and catcalls from the Patentnut Gallery are welcome.
This is my understanding as well. Also keep in mind that IC charges can carry over from a divisional to a parent application.