Holy Cyclops

In a comment to my post What’s Wrong with Modal Ontological Arguments, Kenny mentioned the Moorean Shift.  I want to take a few moments to look at it.

The Moorean Shift takes an argument whose form is modus ponens and converts it into one whose form is modus tollens, shifting premiss and conclusion in the process.  Thus, the argument

1.  p—>q
2.  p
Therefore,
3.  q

one of whose premisses is p and whose conclusion is q, becomes the argument

1.  ~q—>~p
2.  ~q
Therefore,
3.  ~p

one of whose premisses is ~q and whose conclusion is ~p.

The idea is that while accepting the premiss p—>q (which is rewritten in its equivalent contrapositive form ~q—>~p), the person making the Moorean Shift finds ~q more likely than he finds p, so instead of arguing from p’s truth to q’s truth, he argues from q’s falsity to p’s falsity. 

I see nothing wrong with this approach.  When evaluating an argument’s soundness, we must evaluate the truth-values of its premisses.  If one finds q more likely to be false than p is to be true, then he will be more inclined to view the second argument as sound than the first one; conversely, if one finds p more likely to be true than q is to be false, then he will be more inclined to view the first argument as sound than the second one.

For example, one might argue

1v.  If tigers are vegetarians, then tigers do not eat meat
2v.  Tigers are vegetarians
Therefore,
3v.  Tigers do not eat meat

But, while accepting premiss (1v), one might think that premiss (2v) is simply not true, and that the argument is therefore unsound; and if he also thinks that the conclusion (3v) is true, he might construct the new argument

1c.  If tigers eat meat, then tigers are not vegetarians
2c.  Tigers eat meat
3c.  Tigers are not vegetarians

Naturally, one finds the second argument sound but the first one unsound.

The difficulty with the Moorean Shift isn’t the Shift itself, which is entirely legitimate, but rather a linguistic problem in Moore’s use of it against philosophical skepticism that has nothing intrinsically to do with the Shift.  The philosophical skeptic thinks that one cannot know that he isn’t dreaming, or hallucinating, or a brain in a vat, or otherwise deluded about what appears to be true.  G.E. Moore argued against philosophical skepticism by holding up his hand and saying, “Here is a hand before me,” and claiming that since he knew his hand was before him, he knew something about empirical reality, and therefore philosophical skepticism was defeated.  He argued, in other words, in the following way (see Wikipedia entry Here Is a Hand):

Let S be an epistemic agent; let p be some skeptical possibility, like S’s dreaming or hallucinating or being a brain in a vat; let q be a knowledge claim about the world, like S’s hand being held before him.  Then the philosophical skeptic argues that

1s.  If S doesn’t know that ~p, then S doesn’t know that q  (If S doesn’t know that he isn’t dreaming, then S doesn’t know putative fact q about the world—in particular, S doesn’t know that his hand is held out before him)
2s.  S doesn’t know that ~p  (S doesn’t know that he isn’t dreaming)
Therefore,
3s.  S doesn’t know that q  (S doesn’t know putative fact q about the world—in particular, S doesn’t know that his hand is held out before him)

and Moore replies

1m.  If S knows that q, then S knows that ~p  (If S knows putative fact q about the world—in particular, that his hand is held out before him—then S knows that he isn’t dreaming)
2m.  S knows that q  (S knows putative fact q about the world—in particular, that his hand is held out before him)
Therefore,
3m.  S knows that ~p  (S knows he isn’t dreaming)

Moore holds out his hand in front of him and says, “Here is a hand.”  Since, he thinks, he knows that there is a hand before him (”S knows that q”), he also knows that philosophical skepticism is false (”S knows that ~p”). 

Put this way, it seems that a simple linguistic or conceptual mistake is being made.  The philosophical skeptic says that if one cannot know that he is not, say, a brain in a vat (or some other skeptical possibility, like being a dreamer or a self-generator of the appearances), then he cannot know any empirical fact; and then claims that one cannot know that he is not, say, a brain in a vat; and therefore one cannot know any empirical fact.  Moore says that one can know an empirical fact, and therefore can know that the skeptical possibility is false; but the philosophical skeptic’s use of the word know and Moore’s use of the word know seem to differ.  The philosophical skeptic’s use of the word seems intended to imply complete and utter certainty, beyond Cartesian doubt.  Moore’s use of the word seems only intended to imply everyday certainty, beyond everyday doubt.  (We don’t, after all, walk around muttering to ourselves, “Is this really my hand before me?”) 

It would be hard to believe that an acknowledged great philosopher like Moore would have missed this, so perhaps we can read his Moorean Shift, as applied to the philosophical skeptic’s argument, differently.  Perhaps all he means is that he finds it more likely that the everyday assertion that there is a hand before him is true than that the pathological assertion that, say, he is a brain in a vat, is true, and that he therefore accepts (1m)-(3m) rather than (1s)-(3s).  Unfortunately, that just seems like another way of saying that his use of the word know doesn’t imply complete and utter certainty, beyond Cartesian doubt, but only implies everyday certainty rather than philosophical certainty.  Lots of philosophical skeptics, I’m sure, would also accept (1m)-(3m), on that same everyday use of the word know—philosophical skeptics don’t go around muttering to themselves, “Gee, I wonder if this is a hand before me,” in everyday life, either.  The philosophical skeptic’s argument is directed at the conceivability of Cartesian doubt—at the conceivability of our being brains in vats, for example—but I imagine that most philosophical skeptics would nevertheless endorse Moore’s argument for the everyday sense of know.  So, it still seems that Moore and the philosophical skeptic are simply not addressing the same point.  Moore seems not to be addressing Cartesian doubt at all. 

But the Moorean Shift remains a perfectly reasonable way of choosing among deductively valid arguments, since what one wants to accept are sound arguments, which involves assigning truth-values, or at least likely truth-values, to a valid argument’s premisses.  And that’s all the Moorean Shift does:  It says, “I find this premiss more likely than that one, and therefore find this argument more likely to be sound than that one.”