I'm currently teaching Modern Philosophy, and we are reading Berkeley's Three Dialogues.  Philonous's presentation of what Gallois (1974 Phil Review) calls Berkeley's "Master Argument"  (MA) was always particularly striking to me.   The majority of my concern about the argument was along Bertrand Russell's line of resistance: there is a confusion between imagining, say, a tree independent of mind and holding that the tree, in imagination, is dependent on the mind.  So I'm not sure the argument is sound.  I still have that concern, but yesterday morning I found myself having another problem with the argument.  Maybe NonSequitur readers can help me out here, because I, now, don't even think the argument is valid.  But it can be revised.  Here goes.

Here's Philonous's presentation of the MA (edited for space):

I am content to put the whole upon this issue.  If you can conceive it possible for … any sensible object whatsover to exist without the mind, then I will grant it actually to be so….

Is it not as great a contradiction to talk of conceiveing  a thing which is unconceived?…

Now, I've always understood the MA to be one that establishes the falsity of materialism (or non-mentalism) and the truth of idealism (as the two being mutually exclusive and exhastive options).  That is, if you show the falsity of one, you've established the other.  The MA is a demonstration that materialism is false.  But as stated, it doesn't do so validly.  Here's a formalized version:

P1) If it is possible to conceive of a sensible object w/o a mind, then it is possible that those things exist without minds. 

P2) It is not possible to concieve of a sensible object w/o a mind.

C) It is not possible that there are sensible objects without minds.

Again, most folks object to P2.  That still seems right to me.  But even if you grant P2, the argument doesn't go through, because it's a fallacious form of inference: Denying the antecedent.  That is, the form of the argument is:

P1) If P, then M

P2) not-P

C) not-M

This is craziness.  Now, I think that there are two options for Berkeley-defenders to go here.  The first is to say that C, because it's not explicitly stated, isn't the conclusion.  But Philonous certainly seems to be convinced that he's shown not-M in the follow-up with Hylas.  And in the Principles, Berkeley takes it that MP on P1 with P2 establishes the falsity of idealism:

[I]f you can but conceive it possible for … any thing like an idea, to exist otherwise than in a mind perceiving it, I shall readily give up the cause. (PHK  22)

Now, this seems clear that he does think that P2 establishes certain truths, specifically, the truth of materialism and the falsity of idealism.  So I'm unsure that this first option is really a good interpretive one for Berkeley-defenders.  But it seems  plausible philosophically.

The second option is to read all those IF-clauses as ONLY IF-clauses. And so P1 should be, rather:

P1*)  I will agree that Materialism is true ONLY IF it is possible to concieve of a thing existing independently of a mind.

P2, then, works just fine to show that Materialism is false, by Modus Tollens, now.  So the other interpretive strategy is to read Berkeley's argument as needing a switch of antecedents and consequents.

This, by the way, seems even more philosophically plausible than the first option, as I think that P1* is much more plausible than P1.  Just because materialism is concievable, it doesn't mean that materialism is true (as P1 runs); but if materialism isn't even consistently conceivable, then that counts as very good evidence that it's false (as P1* runs). 

So this strategy saves the MA from the fallacy of denying the antecedent and has it with a more philosophically plausible first premise.  The only problem: you have to take it that Berkeley mixed up necessary and sufficient conditions.  When he's doing metaphysics.  That's uncharitable, to say the least.  Oh, and it also doesn't save it from the old Russell objection that P2 is just false.