A paradox?

[Genady]

Stanislas Dehaene is an important psychologist and cognitive neuroscientist, but I think that he is grossly mistaken in genetics, in the following passage from his book, How We Learn: Why Brains Learn Better Than Any Machine . . . for Now:

He goes on suggesting his idea for the paradox resolution, but I think that there is no really a paradox, just a mistake in the assumption on how genome works.

Edited December 28, 2021 by Genady

[CharonY]

My guess is that what he describes is what folks might think how genes work when they hear about DNA being the blueprint (which, in many ways is of course not accurate) and uses it as a starting point to explain how it actually works. In fact, I think fairly early on the brain has been used as one of the examples how you can have higher complexity from (relatively) simple instruction sets.

[Genady]

22 minutes ago, CharonY said:

My guess is that what he describes is what folks might think how genes work when they hear about DNA being the blueprint (which, in many ways is of course not accurate) and uses it as a starting point to explain how it actually works. In fact, I think fairly early on the brain has been used as one of the examples how you can have higher complexity from (relatively) simple instruction sets.

Unfortunately, he does just the opposite. He takes this starting point as a correct description and goes to conclude that actual brain structure is a result of learning rather than genes. Of course, the mistake is to see genes as a "blueprint", while they rather are an "instruction set". I was surprised to read this in a very recent book, written by a well-known author.

Edited December 28, 2021 by Genady

[CharonY]

3 hours ago, Genady said:

Unfortunately, he does just the opposite. He takes this starting point as a correct description and goes to conclude that actual brain structure is a result of learning rather than genes. Of course, the mistake is to see genes as a "blueprint", while they rather are an "instruction set". I was surprised to read this in a very recent book, written by a well-known author.

The way you describe it does not sound like a contradiction to me. Most of the structural events in the development and maintenance of the  brain involves genes, it requires post-genetic interactions and more of an emergent property. I think instruction set is not quite the right analogy, either. The instructions are fairly trivial, "produce RNA". All the whens, ifs and other interactions are not as such controlled by the genes. In a way blueprint is perhaps more fitting after all, but on the most technical level it is not a blueprint of the final product, but mostly a blueprint of the individual elements. Perhaps something like the blueprint of legos, but not necessarily of the desired item to be built. However, if all the bits and pieces interact in a biochemical favourable environment, the desired item kind of self-assembles.

[Genady]

6 minutes ago, CharonY said:

The way you describe it does not sound like a contradiction to me. Most of the structural events in the development and maintenance of the  brain involves genes, it requires post-genetic interactions and more of an emergent property. I think instruction set is not quite the right analogy, either. The instructions are fairly trivial, "produce RNA". All the whens, ifs and other interactions are not as such controlled by the genes. In a way blueprint is perhaps more fitting after all, but on the most technical level it is not a blueprint of the final product, but mostly a blueprint of the individual elements. Perhaps something like the blueprint of legos, but not necessarily of the desired item to be built. However, if all the bits and pieces interact in a biochemical favourable environment, the desired item kind of self-assembles.

DNA is neither a blueprint, nor an instruction set, and I am not a fan of metaphors /analogies. The point is that it, as you say, is not a "map" of the final product. The paradox he describes does not exist, this is my point. He says that it is a paradox, that the final product contains more details than a source used to build it. But this happens all the time, e.g. with algorithms. For example, a number that represents square root of 2 contains infinite number of details (infinite non-periodic sequence of digits), but it is produced by a short simple algorithm, which knows nothing about them.

[TheVat]

Yep.  Not sure Dehaene is paying attention to complexity theory there.  And the human body is not an isolated system, so any supposed paradox would have to ignore the fact that it interacts massively with a complex environment and biochemical history of the planet, as it develops.  This all goes back to guys like Warren Weaver and "organized complexity" and emergentism.   

The house analogy doesn't apply, either.  A stud is a complex artifact, with complex structure produced by a tree and its ecosystem, but the blueprint doesn't have to code for all that.  It just says basically "go buy a pile of studs, 92.5 inches long (USA)."

[CharonY]

I think we need more context to establish what the author means at this point. For all I know it could be a throwaway comment which basically only tries to establish the rather simple fact that a lot of biological processes arise from interactions rather than a top-down program.

[Genady]

1 hour ago, CharonY said:

I think we need more context to establish what the author means at this point. For all I know it could be a throwaway comment which basically only tries to establish the rather simple fact that a lot of biological processes arise from interactions rather than a top-down program.

Yes, it appears so. He quickly moves on to deeper advantages of flexible learning compared to pre-wiring.