(September 12, 2015 at 10:02 am)Nestor Wrote:(September 12, 2015 at 8:48 am)Alex K Wrote: Ok, so from my somewhat understanding of neural networks, and my complete layman's understanding of philosophy - we understand how neural networks process information and dynamically divide incoming signals into categories (reinforcement learning for example), and I simply think that old philosophers who could not possibly imagine such mechanisms at work in our heads, felt compelled to explain why we know categories of things, to explain the discreteness of categories - well, we recognize a chair for a chair independent of its detailed properties (material, color... it's always examples with chairs), so there must be some abstract idea of chairness which is a thing in idea space. Today, I would think, even our rudimentary understanding of the technical details of the mind make this much less compelling. A chair is a chair whenever the neural networks in our heads sort the incoming data into the category chair, which they have learned culturally. and that's it. My gut feeling is the abstract objects Nestor mentions, such as numbers, ideal straight lines, etc. can also be explained in this way as a property of our brains, not of the world, because what is a straight line but a category that the neural network in our heads has learned. I'm sure Nestor, who knows much more about philosophy than me, can comment whether I completely misunderstand things here.Interesting... I would probably agree that Plato's forms are much less compelling with regards to explaining how particulars relate to universals, or vice versa, given the far more elaborate understanding of different cognitive functions - which, as you say, includes reinforcement learning - that we possess against the ancients. Of course, it seems to some extent that how our brains process the signals it translates to be appearances of three-dimensional objects, and then the mental cues it associates with them, will be somewhat arbitrary, i.e. determined by the series of accidents that comprise so much of evolutionary history. I doubt bats, for example, would experience objects - like chairs - similar to how we do... But I'm not so sure we can say quite the same thing about the apparently intrinsic values expressed in arithmetic and geometry. They would appear, to my mind anyway, to relate to the world in a way that is the complete opposite of arbitrariness or convention... almost as if there is a necessity to their properties both in relation to world and with respect to themselves, and I cannot see how neutral networks could account for that.
This reminds me of an idea I encountered long ago (so long ago that it hardly matters now if it was true then) that most mathematicians were platonists with respect to mathematics. The Internet Encyclopedia of Philosophy states "Mathematical platonism enjoys widespread support and is frequently considered the default metaphysical position with respect to mathematics." But it would be interesting to see a poll of mathematicians.
Thinking back on the expressed attitudes of my mathematics teachers, I do not recall any of them saying they had any other position, though I do not recall all of them talking about this.
"A wise man ... proportions his belief to the evidence."
— David Hume, An Enquiry Concerning Human Understanding, Section X, Part I.