Dear Mat,
Interesting letter. It was a rollercoaster from my POV because the first few paragraphs appeared to be a dastardly warping of what I was saying, yet by the end of the letter I was nodding my head, utterly in agreement. How can this be?
This might seem lame but I think it’s easiest if I just respond to the relevant points inline. This is you:
I was saying adaptations reflect truths, which is to say they somehow represent (I was saying “encode”) some truth. Instead you say it’s more a “hack”, something that exploits the world without awareness certainly but also even without reflecting any truth. This is most obvious if you manipulate the environment so that the hack fails.
I never said “hack”. Bad start. Also “truth” here is iffy — is it something in the world the knower is trying to get at, or is it something the knower can obtain? Or is it a kind of harmony between what the knower has and how the world is? I should go even further now and admit that I not only consider correspondence theories of truth to be dead, but that truth is a concept that leads us down the wrong path, generally towards this picture/mirror/representational theory of knowledge. You go on:
The idea [of] a hack is informationless and so the knowledge lies externally is just the reverse of the more commonly held extreme that all the knowledge is in the genes and the world is inert. I’m convinced the answer is between these two extremes. Deutsch and Dennett explain the well-known problem of memes and genes both not being valuable by themselves. In each case you need a gene reader (a creature) and meme reader (a person with shared knowledge) which has its own knowledge/hacks.
More hackery! I agree, both extremes are wrong. Hence the example of Tinbergen’s pecking gulls: the bird has genes (info) that work in a given environment (more info); neither key nor lock can be informationless. I am agreeing with Dennett and Deutsch — in his ideas on better and worse explanations, not his idea of creativity ex nihilo which runs counter to this.
Which brings me to the problematic bit:
Just where knowledge resides depends of course on what knowledge is and neither of us are sure of that. If we were to start using the word “hack” to mean extreme complete uninsightful behaviour and “truth” the precise opposite of this, total omniscient fundamental insight, then I define knowledge as whatever’s in between these extremes and connects, like a ramp, one to the other. We don’t need a correspondence theory of truth for this. To avoid it completely we can define knowledge as anything more insightful than a hack, and that there are degrees of insightfulness with either no end point truth or an endpoint so far away it doesn’t matter.
I do have an idea about knowledge (of course I’m not certain about it) and I think most people are wrong about it precisely because of this misconception about “where knowledge resides”. We know about where information resides, but I say that in systems where the relevant info (that makes something happen) is in more than one place and over a stretch of time, it is involved in a process and that’s knowledge. What/where do you think knowledge is?
The truth-as-total-omniscience, even as an idealisation, is exactly what I’m saying is all wrong about theories of knowledge. The idea of unknowing hacks being the opposite pole is also classic correspondence theory of truth. Knowledge can’t be a ramp from one to the other: neither of them exist.
But now you bring it back to my analogy:
The other thing about knowledge is that it’s only determinable against a matching “problem situation”. Problem situations are fair tests analogous to your locks. The peck contains knowledge when tested against the problem situation of its natural upbringing, but it lacks knowledge when humans manipulate it in an experiment. If you ignore problem situations you’ll find yourself presuming a total lack of knowledge embedded in a Saturn V rocket because it fails to launch from inside an active volcano.
Totally. Problem situations are indeed like locks. In fact, change “determinable” into “usable” or similar and this is my theory. The rocket is a perfect example. Most of the knowledge isn’t in it. All the relevant rocket science, for instance, happened in laboratories, books, papers and universities around the world, over years and it’s extremely context-sensitive.
I agree with your idea about evolutionary biology being tantamount to finding the problem situations that organisms’ traits are for. You go on:
Hacks do not exist, they are illusions brought about from an ignorance of the problem situation. It is extremely hard to tell the difference between a trait whose selection pressure is yet to be discovered and a random unpressured mutation. It’s exactly the epistemological mechanism that limits our awareness of the knowledge in our theories without a valid problem situation to explain or to test them against.
You came up with “hacks” but again I agree, it is hard for a biologist to tell the difference between a random mutation and a yet to be discovered function and ditto for the epistemologist with her theory. But what if it’s not just hard but impossible, or better yet, inapplicable? This is what Dennett says: we can only tell in retrospect what a trait was for (i.e finding its function) because when it first appears the fact of the matter does not exist — yet.
I think you’re assuming a certain timelessness to knowledge. In an unfolding world where things are evolving in realtime, we shouldn’t even expect new theories to explain new things ahead of time… How could they? That’s not to say they can’t make predictions, but we can’t know if the knowledge in them will work going forward — we just can’t because it doesn’t exist yet because knowledge is something that happens (in a problem situation), that we do. Only if our theories were passive maps or reflections of a static, remote reality would that be the case (where knowledge simply is). Only if we were trying to approach some asymptote of truth would that be the case. But if theories are more like keys that allow us to unlock things — part of the process of doing knowledge — we should expect to need new ones and expect that some locks don’t have keys yet and some locks don’t exist yet and some keys change the locks. (This is related, I believe, to the so-called new riddle of induction, via Nelson Goodman, which is a kind of temporal mindfuck add-on to Hume’s problem.)
You’re assuming that problem situations exist independently and exhaustively. Maybe in particle physics they more or less do, but biology and almost everything else we care about is different. A useful example from Dennett is his discussion of the Baldwin effect, where learned behaviours interacting with the environment can change the genome. It sounds like Lamarckism, but really it’s just demonstrating how new problem situations for an organism can be incorporated into its genome in a fairly short amount of time and also how the problem situation (environment) is also changed by the organism’s responses. The notion of a self-fulfilling prophecy is poignant as well: a theory (key) about the world (illusory to start with) initiates a process that changes the world (problem situation or lock) to more closely fit the theory/key.*
We converge on the following point:
Theories/traits must reflect their problem situation, otherwise the latter would be irrelevant when considering the formers’ utilisable knowledge. As theories and traits reflect their problem situation, and because problem situations can exist at any scale and any level of emergence, real knowledge can be achieved at any level of abstraction. Pecking reflects the reality of the function of “red dots” the way a Saturn V reflects the reality of the function of a “launch pad” the way Newtonian physics reflects the reality of the function of “the force of gravity” the way that general relativity reflects the reality of “curved space time”. Knowledge is contextual, hence your lock and key metaphor – one needs the other. But the key matches the contours of the lock and both of these are aspects of reality.
Couldn’t have said it better myself (except I would omit the word “reflect” and I’m not ready to say any scale). I certainly never said the key doesn’t match the contours of the lock. It has to. I used the word “complementary” for that reason. Replace “theories/traits” with “key” and replace “problem situation” with “lock” and I’m pretty sure this paragraph is saying what I said…
So perhaps we have different notions of what knowledge is at a very fundamental, physical level, but we think it is doing the same job in practice? That’s philosophy I guess.
Jamie.
PS I can’t let the last letter pass without a comment that Sturgeons Law is that 90% of everything is crap, not wrong. 90% of pop science books certainly are crap. But they’re normally written by scientists afraid of moving outside their discipline so they’re generally pretty conservative on the science and flag wherever they’re getting unorthodox, for justified fear of rebuke from jealous colleagues. If you mean 90% of them are “wrong” in the Popperian sense that 100% of theories are “wrong”, well…
*For another example in biology, think of niche construction: a population changes its environmental conditions which in turn affect its members reproductive fitness and hence the makeup of genes in the gene pool.
Also published on Medium.