Just an observation to amplify something I mentioned before. The real problem in discussing these things is our language - and our habitual use of it.
The concepts of "self", "I", "me" etc., are self-referential. David's link suggests we should entirely set questions based on such references aside - I agree that would be the best course of action if we constrain ourselves to simply trying to resolve self-referential issues using combinatorial logic - that course of action is automatically doomed to (messy) failure.
However, what most people don't realise is that there are techniques in which you can meaningfully deal with such problems, as long as you identify their true nature. Self-referential logics tend to go into oscillation which means that if you try to resolve them into one single state or another the answer is nonsense. It's analogous to asking whether the swinging pendulum of a clock is swinging left or swinging right. The real answer is that it's doing both, but at any given single instant of
time it might be doing just one or the other. So the missing element is time. When you add the element of time into logics you get a different kind of logic which is sequential logic.
The key confusion due to misunderstanding of the sequential nature of certain statements is that what is universally true for the whole system all the time is not necessarily true for each of its constituent elements all the time. But may be true for both the system and the elements at specific moments of time. So the "correct" answer at any moment depends on the actual timing of the observer. Observers with different timings will come up with different answers that appear superficially contradictory - but in the bigger picture they're not.
For example, take the statement: everything I say is a lie. This is a classic self-referential statement. The statement cannot be resolved into one
constant state of truth or falsity. If I start by assuming the statement is true, I parse the statement and conclude it's false. On that basis I re-parse the statement from the assumption it's false and conclude that it's true. So it oscillates between true and false each time I parse it. The key element in the process is that each parsing of the statement
sequentially follows a previous parsing - in other words it is a process on-going in time. Therefore we can say that the statement has no single value of "truth" for all points in time.
Sequential logic of this type can be useful. It's actually used in computer logic circuits and computer hardware designers will immediately recognise that the above is a linguistic equivalent of a classic flip flop circuit formed by crossing the inputs and outputs of two logic gates.
Unfortunately, it's not so useful in everyday language it just leads to confusion. In the discipline of General Semantics which I find interesting in such contexts, they use the concept of "orders of abstraction". They resolve problems of this type by making a distinction between a thing, an abstraction of a thing and an abstraction of the abstraction (and so on ad-infinitum). They then apply a rule that says that in language you cannot meaningfully equate two things that lie at different orders of abstraction.
Taking the statement, "everything I say is a lie", they observe that it's a composite confusion of different orders of abstraction. The sub-statement, "is a lie" is an abstraction applied to a more fundamental abstraction "everything I say". In other words, what I say and what I say about what I say, are at different orders of abstraction and to equate them is meaningless.
How does this relate to the questions of "self" etc., in Buddhism? Well, any statement I make about "my self" is self-referential in the same way. "Self" is a label or abstraction applied to a collection of "things". What I
say about that abstraction is at a different order of abstraction, it's an abstraction of the first abstraction. So anything I say about "my self" is ultimately meaningless, because it is a confusion of two different orders.
Thus far, I've managed to avoid tying myself up in semantic knots in the discussion by referring largely to Ryokan when I have mentioned "self" - because it's not
my self I don't get caught up in the self reference. But anyone reading it can do if they try to think about it in terms of
their "selves".
The only reason I mention all this is because if anyone is unaware of it, it might help in understanding why Buddhism appears to be very cryptic, why some questions are answered with silence or why some questions are answered with cryptic statements, or why some questions get thrown back at the questioner. These are some of the issues raised by the commentator in David's link.
Personally though, I think we don't need to resort to such measures if we just remain aware of the real nature of the problem and make an effort to avoid getting caught up in it. It
can be done but it requires "skillful means" - which is another distinctly Buddhist concept.
