John Locke and David Hume had entirely different views on the issue of personal identity. The former took the existence of identity for granted and sought to define it in terms of memory, using memory of past action as essential criteria both in defining and pointing out cases of continued personal identity in time; the latter, on the other hand, was highly sceptical of the notion of personal identity itself and argued that, if anything, memory contributed a necessary part of the illusion of continued identity in time, but that memory is not sufficient for sustaining such a notion. In this paper I will explain how this fundamental difference persists despite two initially misleading statements made by Hume.

Hume says that "memory not only discovers the identity, but also contributes to its production, by producing the relation of resemblance among the perceptions" (page 170, lines 5-7). In order to understand why the first part of this sentence does not represent an agreement with Locke, we must first understand what is meant by a "relation of resemblance." Hume says we are lead to the belief in identity not by the things we perceive but by relations we create between the perceptions we have. These relations can be of three types: resemblance, contiguity, or causation (page 165 line 9). Thus if an object persists through time with only minimal physical changes, if the object resembles itself at a different time, we are apt to attribute identity to it. Hence, resemblance causes a belief in identity. Similarly, if an object changes drastically, but does so very slowly and gradually, we do not haste in attributing identity to it so long as it has this "contiguity". Finally, if an object changes completely, and even if it ceases to exist for a time, but has a "common end or purpose" (Page 166 3rd paragraph), we are still willing to give this object the same identity (Hume cites the example of a church that is knocked down and rebuilt with different design and materials). In this case, it is causation which leads to the attribution of identity.

In the case of personal identity, Hume says only resemblance and causation apply. By creating a relation of resemblance between our perceptions, memory helps produce the idea of personal identity. Hume says that when we remember past perceptions, the memory itself is a perception. Since, according to Hume, perceptions resemble their objects, one memory of differing perceptions provides a link of resemblance, since all these past perceptions must resemble this one memory. In other words, memory itself is a perception that resembles each perception it recalls. This resemblance allows us to create a relation among impressions that otherwise would have nothing in common with each other.

It is important to remember that for Hume in this case memory is only contributing to an illusion, which is unlike Locke’s views, in which memory constitutes a real phenomenon. Furthermore, whereas Locke sees memory as the only possible criterion for personal identity, Hume discards the notion that memory is sufficient to give the idea of personal identity any substance. To see this, let us examine how the idea of causation provides another relation amongst ideas, which in turn gives us the the other necessary component in building a notion of personal identity.

Hume uses the idea of causation not only to denote a relation of "common cause", but also, when it is applied to something like personal identity, to explain a number of impressions and ideas that are linked together by a relation of mutual causality. He says "our impressions give rise to their correspondent idea; and these ideas, in their turn, produce other impressions." (page 170, lines 14-15) Thus even if a person changes completely from T1 to T2, both physically and in terms of personality, we can still trace the charcateristics at T2 to impressions and ideas sustained by the person at T1 and show how these caused the changes to take place.

In this case, memory is still important, for it allows us to witness these changes, remember different states, and then attribute appropriate causes and effects. And, in this case, memory does not produce personal identity; instead, it merely makes it possible to see it, for memory does not constitute causation. Hence, Hume calls memory "the source of personal identity." However, this leads Hume to a view that directly clashes with Locke’s, for once memory has aided us in establishing this sort of causation, the relation of causation can be applied to instances not in memory’s reach. In other words, even if one doesn’t remember a particular idea or impression, one can deduce the idea that he existed at a particular time by assuming the existence of certain impressions, based entirely the causal relations memory has helped forge.

As already stated, Hume thinks that the idea of personal identity is an illusion, and in explaining how memory functions to create this illusion through a system of relations, Hume is trying to explain the phenomenon of the idea of personal identity, not personal identity itself. Thus, he explains that one cannot, as Locke does, hold a view of personal identity in which identity does persist when memory does not, for this blatantly disregards the fact that one’s idea of his own personal identity extends to times he cannot remember. It is in this sense that "memory does not so much produce as discover personal identity", since if we extend personal identity to span times not accessible to memory, then it must be the case that that identity existed prior to the memory that may well have aided in seeing it.

In conclusion, we can now a) reconcile Hume’s two statements in which he says first that memory contributes to the production of personal identity and, second, that memory does not so much produce as discover personal identity, and b) understand why neither of these statements constitute a version of Locke’s understanding of personal identity. It is simply that in the first statement Hume is talking about the relation of resemblance, and in the second he is talking about the relation of causation. Furthermore, in neither case is he saying that memory either entirely constitutes personal identity or that it suffices in providing evidence for its existence.