In his classic article Hume on Personal Identity, Terence Penelhum charges that Hume, in maintaining that we are always mistaken in ascribing identity to persons, has made "an elementary error" and fallen victim to a conceptual "muddle". 1 Hume's error, Penelhum says, is in thinking that a succession of different objects cmnot be counted as one object; Hume's muddte consists in thinking it a contradiction to say that an object is both "numerically the same" and "changed". But this line of criticism ought to make us suspicious. Hume is a great philosopher and it seems unlikely that one of the central theses of the Treatise is merely the result of blunders.
In this paper I will argue that Penelhum has misconstrued Hume's argument and Hume's enterprise. Hume is presenting a fundamental metaphysical problem about identity through change (a problem as old as Heracleitus), not trying to analyze the ways we talk about change, as Penelhum seems to believe. In Sections I and II I will show that Hume's disturbing conclusion about personal identity does not depend upon either of the mistaken theses that Penelhum attributes to him. In Section III I explicate Hume's argument which, I argue, raises problems so serious and so general that Hume's own positive account of identity falls prey to them.
I
Penelhum observes that Hume defends the conclusion that persons do not persist through change by advancing a general thesis about identity. Penelhum writes Hume's thesis turns on one central point, and stands or falls with it. This point is his contention that it is, "to a more accurate method of thinking", a confusion to call an object that changes the same. The "idea of identity or sameness" is the idea of an object that persists without changing. The fact that the parts of a changing thing may be related to one another does not, after all, alter the further fact that they do change; so in this case we do not have identity or sameness, and it must therefore be due to some ingrained tendency of the mind that we talk as we do.~