Merleau-Ponty starts “RM” with a pertinent reminder:

Skepticism has two sides.  It means that nothing is true, but also that nothing is false.  It rejects all opinions and all behavior as absurd, but it thereby deprives us of the means of rejecting any one as false.  Destroying dogmatic, partial, or abstract truth, it insinuates the idea of a total truth with all the necessary facets and mediations.  If it multiplies contrasts and contradictions, it is because truth demands it.

The reminder here is that there is a form of skepticism, Montaigne’s, that serves the demand of truth.  As such, Montaigne’s skepticism is two-sided:  it has its expected negative side (“nothing is true”) but also its unexpected positive side (“nothing is false”).  The point of the two sides is not skeptical self-stifling, but rather an all-the-more gladsome servitude to truth.  Dogmatism, partiality, abstraction all ill-serve the truth, providing only one facet among many or excluding required mediations.  Think of this as skepticism with a finally positive valence, a skepticism that approaches truth by various refusals of truths.

As I mentioned before, in his “In Praise of Philosophy”, Merleau-Ponty insists that great philosophers thematize ambiguity, but that their so doing “contributes to establishing certitudes rather than menacing them”, and so he distinguishes between good and bad ambiguity–the one, I take it, establishing, the other menacing, certitude.  What Merleau-Ponty says about Montaigne’s skepticism concretizes the claim about thematizing ambiguity.  In destroying dogmatic, partial or abstract truth, Montaigne thematizes good ambiguity, an ambiguity calling for nuance, mediation; an ambiguity indicating the shape of the total truth.

Montaigne contradicts himself, when he does, out of the extremity of his servitude to the truth.

The first and most fundamental of contradictions is that by which the refusal of each truth uncovers a new kind of truth.

This sentence is the nervus probandi of the Merleau-Ponty’s introduction to “RM”.  Notice that Merleau-Ponty here describes (materially) metalinguistic negation.  (Laurence Horn, who has done the most to clarify this form of negation, understands it as a metalinguistic device for registering objection to a previous utterance on any grounds whatever (including even the way it was pronounced).  In the thanks-footnote to his classic article from 1985, Horn provides a gracious and humorous example of the phenomenon:  after mentioning the folks he is indebted to, he notes, “Their contributions were not important–they were invaluable.”)  I understand Merleau-Ponty to see Montaigne’s negations as informed by a recognition that a particular utterance is dogmatic, or is partial or is abstract.  That utterance is then negated on the grounds that it is either dogmatic, or…   This is the way in which Montaigne’s refusals of truths “uncover a new kind of truth”, where “new kind” does not mean that we have, as it were, shifted from a truth-predicate at say, the zero level, to one at the first level (and so on) but rather that we are moving from a partial truth to a less partial truth–and the partial truth is not negated in the sense that it is false, but rather in that it is partial.  This shifting can be seen not only at the level of particular lines of Montaigne’s essays, but even in the essay’s basic structure, in the way paragraph follows paragraph, shifting from the partiality of the previous paragraph to a subsequent paragraph that renders what is being said less partial.  The earlier paragraph is not to be refused totally, but rather refused in the interests of the total truth.  A formalist example:  If I say “It is not exactly incorrect to say that p”, I do not mean that It is correct to say that p but I also do not mean It is correct to say not-p.  Rather, saying “p” stands in need of further saying, of a further less partial saying, perhaps; or, of one less dogmatic or less abstract.  A new kind of truth is a less dogmatic or a less partial or a less abstract truth.  (Some of you will recognize metalinguistic negation from its very common use in J. L. Austin’s work, where it informs not only the content of his presentation of, e.g., performatives, but where it informs the very style of Austin’s prose.  Jean-Philippe Narboux has recently written a wonderful paper on this and related matters, “There’s Many a Slip between Cup and Lip:  Dimension and Negation in Austin”.)  Often, then, what take the form of contradictions in Montaigne are best understood as pairings of p and not*-p, its metalinguistic negation.  This has an important shaping effect on Montaigne’s skepticism and his prose, and helps to reveal that it is, as Merleau-Ponty says, two-sided, insisting on facets and mediation, multiplying contrasts and contradiction, welcoming ambiguity as a helpmeet—because truth demands it.