Signs and Learning

How do we learn to swim?  This example that pops up on several occasions throughout Deleuze's work, and here on page 23.  

The movement of the swimmer does not resemble that of the wave, in particular, the movements of the swimming instructor which we reproduce on the sand bear no relation to the movements of the wave, which we learn to deal with only by grasping the former in practice as signs. That is why it is so difficult to say how someone learns: there is an innate or acquired practical familiarity with signs, which means that there is something amorous - but also something fatal - about all education. We learn nothing from those who say: 'Do as I do'. Our only teachers are those who tell us to 'do with me', and are able to emit signs to be developed in heterogeneity rather than propose gestures for us to reproduce.

I think he chooses the example of swimming in particular to illustrate the general failure of the idea of imitation to account for learning.  Someone shows you the movements are involved in the crawl stroke.  You mime them on the beach.  It's actually a very simple pattern.  You imitate the instructor till everything looks exactly the same.  Then you get in the water and try to reproduce those movements and you sink like a stone.  You can't learn to swim by imitating the motions of someone who knows how to swim.  If you could, there wouldn't be any process of learning to swim at all; someone would tell you or show you how to do it, and you'd just do it.  

What you're actually doing when you learn to swim is learning to fit the pattern of your motions together with the patterns of how the water moves.  You're not copying an external model like the swim instructor.  That is just providing you with a set of signs that will hopefully enable you to grasp how you can interact successfully with the water.  A good teacher is not just someone who knows how to do something well, and so provides a good model, but someone who knows how to explain it in a way that triggers your ability to fit your motions together with the world.  If they are transmitting some sort of information to you, it is at best in the manner that a seed transmits information to the next generation of tree -- not like a copy, but like a recipe.

I think it's easy to acknowledge this distinction in the case of learning a new physical skill like swimming or playing tennis.  But what about areas where we seem to be able to just learn that as opposed to learning how?  Aren't there plenty of situations where we learn the right answer more or less instantaneously just by someone telling us what it is?  We do seem to learn information or facts this way, and certainly a lot of our schooling seems to revolve around force feeding people these facts.  I think these situations may actually be the exception that proves the rule though.  Inevitably the new information we acquire so quickly by copying it is actually just a small new modification within a very large framework that took us a long time to build up, just like acquiring physical skills.  

You might call this sort of thing "propositional learning".  Where you learn, say, the definition of something, or what formula to use to calculate the internal rate of return on an investment, or even what Plato said about the concept of repetition.  We tend to think of a lot of "higher" learning as propositional learning, perhaps because it's only this type of learning that can end with a right or a wrong answer.  You either correctly copy the instructor, or you don't.  

In fact, a lot of philosophy is taught as if it were a form of propositional learning.  Was Descartes "right" when he said that the pineal gland was the seat of the soul?  Was Kant "correct" when he said that we can never know a thing-in-itself?  Obviously, in philosophy the answer is up for debate in a way it usually is not in physics.  But the discipline is still mainly taught as if there were a right answer that we will somehow eventually come to.  So the other reason I think Deleuze invokes his swimming example early on in the book is to alert us to the fact that he does not see philosophy in this mold.  We are meant to "do it with him" and not just "do it as he does".  Perhaps this helps us appreciate his unique writing style a little more?  It's a lot easier to simply agree or disagree with Plain English than it is with French Philosophy, where you have to struggle over a sentence just to figure out what it might mean.  I don't think the goal is obscurity, so much as to try and unfold the complexity of the problem, instead of simply offering a ready made solution.

And we're back

Okay, enough about Stengers and the War Machine and the philosophy of science.  It's time to get back to Difference & Repetition.

Last episode on FPiPE we were using embryogenesis as our central metaphor to get a grip on what Deleuze has in mind when he talks about true repetition and its relationship to difference in the introduction.  I got a lot of mileage out of that analogy, and I generally find it useful to think of Deleuze modeling his philosophy after biology rather than after physics or mathematics.  But it's probably time to back up a step and and think a little more abstractly about the point we've arrived at and not get locked in to one metaphor.

The broad sweep of the Introduction seems to be:

1. How can true perfect repetition exist in an objective rule based world?  
2. If true repetition is an illusion, what's happening when we see something that looks like repetition?  It must be that we're seeing a general form or concept get repeated, with some of the particular details changed.  This general form would either be some objectively existing Platonic thing out there in the world, or some concept existing in our head.  Either way, the phenomenal world would be understood as never repeating.  Each moment of each thing is completely distinct and unique.  Our sense of true deja vu (as opposed to general) repetition would only arise when we see two things that are so close that they should share one concept, but somehow there are mysteriously two of them, like a left and a right hand.  These must be some sort of "degenerate" case (as the mathematicians like to say).
3. But wait, nature shows us all kinds of things like left and right hands that seem to be exact repetitions of the same concept.  This presents a major problem if we think that there should be a 1-to-1 correspondence between conceptual forms and things in the world.  The difference between left and right hands doesn't seem to map onto a conceptual difference.  And yet that difference also doesn't seem to fall into the form of a particular variation of a general form (like, say, longer or shorter fingers might).
4. So there seem to be some cases of true repetition that aren't general but that don't correspond to conceptual differences.  This is what real repetition is -- difference, but without a concept for it.
5. This kind of repetition can be explained by a repeated or ongoing process that has some difference within it that leads it to produce two forms as an end product. This is where embryogenesis came in.  The process, as an algorithm or recipe, can be repeated exactly, but the outcome can be slightly, or even radically, different every time.  The difference between the final forms isn't an external conceptual difference (as there is none in the case of symmetrical objects), but a difference internal to the process that creates the forms.  And the difference is also not a particular difference that falls within some general limits.  The difference between left and right can't be explained on the basis of how close they are to one another or to some general model.  The "level" at which the difference happens isn't the same level of the two distinct forms, it is somehow before or beneath them.

I've collected a few of the quotes that step 5 translates: 

We are right to speak of repetition when we find ourselves confronted by identical elements with exactly the same concept. However, we must distinguish between these discrete elements, these repeated objects, and a secret subject, the real subject of repetition, which repeats itself through them. Repetition must be understood in the pronominal; we must find the Self of repetition, the singularity within that which repeats.

It is true that we have strictly defined repetition as difference without concept. However, we would be wrong to reduce it to a difference which falls back into exteriority, because the concept embodies the form of the Same, without seeing that it can be internal to the Idea and possess in itself all the resources of signs, symbols and alterity which go beyond the concept as such.

The interior of repetition is always affected by an order of difference: it is only to the extent that something is linked to a repetition of an order other than its own that the repetition appears external and bare, and the thing itself subject to the categories of generality.

So, basically, there are two kinds of repetition -- repetition of process, and repetition of form.  These operate like subject and object.  

For the object, all the differences between instances lies outside the form, external to its completed individuality.  That's why when we think about two repeated objects, we find that the specifying their difference requires invoking some other level of explanation that doesn't have to do with the objects themselves.  We need ideas like "number", or "reflection" or "translation" or some other mathematical transformation that would apply equally to all objects.  The difference between objects isn't itself an object.  We have no concept or form for this repetition like we have for the objects, and when we reach for it, we immediately reach for descriptions that involves process and movement.  

The subject of repetition -- that is, the repeating process -- actually has an inside to it.  The difference that gives rise to the repeated forms is built into the process.  I think this is Deleuze's answer to the question, "what is a subject"?  The subject is a form of interiority.  It has a "what it is like from the inside" to it.  Only processes have an inside.  Finished forms do not.  Sure, we may take them apart and find other forms within them, but that's exactly when we start to say that the larger form is an illusion that "is really" reducible to smaller forms following laws which are also external to them.  

Now we find ourselves at a strange moment though.  It seems to me there's an ambiguity in the notion of a repeating process that I've already alluded to a couple of times.  Is it really right to say that the process repeats, in order to produce two forms?  In the case of embryogenesis it was more like the process bifurcated or differentiated based on some internal difference.  In other words, the different outputs were products of one ongoing process, albeit one that can spawn new subprocesses.  This is the paradox, as it were, of true repetition -- it's always yet again for the first time.   This brings us back to where we started with the introduction.  Repetition is an infinite series just like the repeated celebration of the 4th of July.  Because the event is historic we repeat it, and repeating it makes it historic.  Without the repetition, it is no longer a singular event, but just some stuff that happened a long time ago.  To understand the inner and subjective form of repetition, we have to understand the singular process that gives rise to the outward repetition of forms.  

Asymptotically Objective

So, I finished reading Stenger's book of essays about slowing science down.  I was thinking of writing some more about it when I looked back at the two earlier posts and realized that I would mostly be repeating myself.  There are some interesting details along the way, but taken together I've mostly covered her basic critique and suggestion.  Scientist should drop the philosophically dubious idea that they are producing objective, authoritative knowledge, and instead focus on the particularly interesting situation they really do create -- a human system that takes into account what matters for  other parts of the cosmos in reliably useful ways.  Science is a human system with human goals, but it cares deeply about what the rest of the world "thinks".  Which is to say that the world can prove scientists wrong, which is the only way one can learn from another entity.

It's a deceptively simple suggestion, and in the later essays you can really hear how it resonates with her thinking about Whitehead and James (who she thinks Whitehead was elaborating on).  To follow all of it would be a very long conversation about Whitehead's idea that everything (including, say, elementary particles) is actually a "society" defined by its own "values", by what "matters" for that society.  The main difference between societies for Whitehead is the way their values incorporate those of other societies by way of contrast.  Which is to say the way they incorporate the possible into the real.  Like I say though, without a much longer discussion, that's not going to sound convincing or even intelligible, so after putting several everyday words in scare quotes, I'm stopping here.

I did have one final thought though, inspired by a comment from Dr. CC that has been rolling around in my head for a while now.  Is science "asymptotically objective"?

This idea seems like a natural and appealing fallback in the face of something like Kuhn's thinking about the possibility of a paradigm shift.  Sure "we once thought" that gravity "was really" the point-to-point attraction of massive bodies, "but now we know" that it's really about curved space-time.  What we thought was "objectively true" has proven to be a complete abstraction that merely worked well to help us do what we wanted.  And we're conscious that our new abstraction is also going to someday fall victim to this same pattern.  Still, though, we know that with each revolution we are somehow getting closer to the one true real explanation.  The objective world at the end of the asymptote.

This is a strange image though.  Because how is a progression of what we now freely admit are abstractions created for human purposes supposed to be magically transmuted into an "objective description" at the end of the asymptote?  And how exactly can a description be objective anyhow?  It's certainly not just in someone's head, because many of us share it.  And it's certainly not just a collective delusion because it allows us to effectively do all kinds of things.  But at the same time a description of the object is clearly not the object itself, but only one aspect of it; the one that matters for our purposes.  Whitehead calls the problem the "fallacy of misplaced concreteness".  An electron is pretty clearly an idea, not a thing, so we must be pulling a fast one when we claim that the world is made up of electrons.  And no matter how much further or smaller we go in our abstractions, we're not going to ever magically hit the "objective" world with them because this same problem is going to apply.

What content would be left to the claim of asymptotic objectivity then? Isn't it really about creating ideas that allows us to do more and more with the world?  Explanations that allows us to do some new stuff on top of all the old stuff we used to be able to do?  Seems like a noble goal and a good candidate for the content behind what we'd like the claim to do.  If you believe A, you'll be able to do X and Y.  But believing B is better (slip the word "more objective" in here) because you can do X, Y, and Z.

Unfortunately, this claim doesn't seem to have anything to do with objectivity, though it does sound a lot like our idea of progress.  Once you drop the claim of objectivity though -- which seems at first may seem costless and merely semantic  -- you will also have to subtly modify your definition of progress.  Because now there's no such thing as "general progress".  There's only specific progress along a particular dimension you happen to value.  But not all of these dimensions point in the same direction and some may even be opposed (as in predictive accuracy and mathematical elegance in the standard model vs. string theory debates).  If you wanted to do Z, then believing B was definitely better.  But what if you wanted to do Q?  

This "progress" then begins to look a lot more like Brownian motion.  Or maybe an amoeba moving up a sucrose gradient at best.  It may have a direction, but there is no a priori reason to think that it's the right or unique one.  The metaphor of an asymptote only makes sense if there is some value to converge to.  If we start to see our direction as inevitably defined by our value space (so to speak), then it seems like we're wandering in a very, very large space indeed, and we're going to need to ask much more complex questions about how our values might overlap with other possible sets of values (say, alien or computer values).

A final thought occurs to me here.  One I don't completely understand  In some sense, Brownian motion does have a sort of asymptote that we call "equilibrium" -- the diffusion of something to occupy a volume at a uniform concentration.  Maybe I've inadvertently defined something exactly like general progress?  Perhaps, yes, science always has a particular direction at a particular time, but maybe somehow, on average over time, it expands in every direction to completely describe everything that could matter about the world?  This is a wildly ambitious hypothesis that makes the physicist's dreams of a final theory-of-everything look like a silly footnote.  Because were they to announce the current version of this theory tomorrow, it still wouldn't describe a fraction of the things that matter just to me -- like why Deleuze is so great or even whether it will rain April 3, 2056452.  But maybe, someday, there could be an ever bigger ToE that would explain those things?

Tough question.  I'm going to go with Stuart Kauffman on this one though, and posit that the universe is actually non-ergodic.

Consider next the number of proteins with 200 amino acids: 20 to the 200th power. Were the 10 to the 80th particles in the known universe doing nothing but making proteins length 200 on the Planck time scale, and the universe is some 10 to the 17th seconds old, it would require 10 to the 39th lifetimes of the universe to make all possible proteins length 200 just once. But this means that, above the level of atoms, the universe is on a unique trajectory. It is vastly non-ergodic. Then we will never make all complex molecules, organs, organisms, or social systems.

 And if life and matter is wandering through this non-ergodic universe, then so are our explanations of it.