Monday, November 02, 2009

Discuss 8

'For all of which, the complexity of our minds, or of our behaviour, is simply irrelevant to the question of whether our cognitive architecture evolved under selection pressure. I do think it's remarkable that nobody seems to have noticed this.'
Jerry Fodor

27 comments:

  1. So, erm, if we are still a long way from understanding how the mind works, or doesn't work, then there's no point holding out much hope that AI will get all that far, at least until we know a lot more. Or so say some. Others are mad for Hal and his pals. Is this the idea behind these discussion points?

    ReplyDelete
  2. One of them, Mark. My primary idea is to get you lot to come up with angles I haven't thought of. Wisdom of Crowds. I know it doesn'r work as such and there's not much of a crowd here, but I'm trying to get it to build somehow.

    ReplyDelete
  3. Yeah, apart from the wisdom bit and the crowd bit, it's working well.

    I initially took that statement to be a refutation of blind watchmaker style anti-evolution arguments based on the complexity of our poetry-scribbling, symphony-writing brains; but I suppose it could equally be rebuttal of the whole evolutionary psychology project.

    ReplyDelete
  4. it's the latter, Brit. Brilliantly done too by Fodor. Knocked it down for me

    ReplyDelete
  5. Well, I’ll throw in my two pence, not that it’s worth much.

    The question of how evolution could have sprung a trick as complicated as the human mind is perhaps parallel to how processors are developing. For a long time, there was the idea that processors would just keep getting bigger and bigger, more complex until they became like the human brain. I’m not sure that Moore’s Law is still holding up but recent processor designs have moved the other way, paradoxically, perhaps becoming more like the brain. Designers have reduced the complexity of the individual elements but make more of them. In Sony’s PS3, there’s the Cell processor which is 7 or 8 cpus. Fiendishly difficult to programme but more powerful when they work. I suppose this concept is behind recent advances: neural networks, bot nets, cloud computing, and even file sharing via Bittorrent. The next big thing will be probably involve some kind of distributed processing.

    There’s even a prototype games service called Gaikai being tested where you no longer have a machine in your own home, just streaming the results to your machine. This is probably not the place to mention this but many of the big advances in these areas such as AI are taking place in video gaming.

    ReplyDelete
  6. That, Willard, is JUST the type of contribution I was hoping for, thanks.

    ReplyDelete
  7. So what you're saying Willard, is that lots of simple bits interacting looks more like the human brain (and produces more complexity?) than does a few complex bits?

    ReplyDelete
  8. Brit, I'm no expert in this but from what I understand, that's something of the idea behind much of the recent research in processor design.

    Heat is the big problem for modern processors. They produce too much, too quickly, and can’t get rid of it. You’d have to look into the development of CPUs in the last ten years but I seem to recollect that one of Intel’s problems was that they tried to carry on produced these huge powerhouses of a processor. AMD took the route of simplifying the work but doing more of it. It accounted for their rise. These days, our PCs are becoming more powerful because each generation has more cores (i.e. more parallelism). The actual clock speeds of the processors (which used to be the main way of denoting how fast your PC was P75, P200, P333...) means very little since they’ve not increasing significantly. I think they’re all about 2 to 3 Ghz.

    ReplyDelete
  9. No idea whether this is relevant...

    Having decision-making distributed rather than centralised, made piecemeal rather than comprehensive, appears to produce better results in another man-made arena - the modern economy.

    Is this something that is common to many (all?) systems- that, to make efficient results more likely than not, a switch from central and complicated to distributed and simple is required once they reach a certain size and complexity?

    ReplyDelete
  10. I found this one of the most interesting quotes so far, without quite knowing what Fodor was on about. I'd like to know more about the debate with Dennett that I now see has been raging. I'm bound to be sympathetic to Jerry, Bryan and Brit on that. But, before that, a word of caution on the silicon and software front.

    People have been aware for years that Moore's Law cannot forever deliver smaller, faster, cheaper CPUs to power our Turing (or von Neumann) inspired machines, with the C standing crucially for Central of course. For that reason there's been an enormous amount of research on parallel programming and architectures, for many years, not just recently. Some of the best fruit of that in the real world has to be Google's MapReduce and BigTable, helping to enable the fabulous response times for search we've all got used to from big G. But taken as a whole, parallel programming is proving incredibly difficult, at all kinds of levels. Many algorithms just don't seem to be amenable to being partitioned in a way that makes much difference to their performance. All those 'simpler' processors going to waste, way on into the foreseeable future. And, even more to the point, real world programmers have enough problems with their debugging right now. Parallel programming is exceptionally hard to get right and to test, except in narrow areas. I fully endorse Software Transactional Memory from Simon Peyton-Jones and the Haskell team at Microsoft Research in Cambridge, for example, as I write that - one of the few things that could really help. But still, even with STM and a very pure functional language ... very, very hard.

    It needn't affect the philosophical debate of course but the realist combined with the nerd combined with the pedant insisted that I say it.

    The wider points about AI I will also leave for now, though it would fair to mark me down as a sceptic on some of that too, much I'm sure to everyone's surprise!

    ReplyDelete
  11. The push for simplification is perhaps cyclic. In the '70s they thought processors would close the gap with high-level programming languages, which was just a daft idea squashed by the RISC revolution in the '80s. The big advantage there was learning to pipeline processors, increasing the internal concurrency.

    Now the concern is with power--if you can programme n processors to do the task in time then that will probably consume far less energy than running on a single processor running at n times the throughput.

    Anyway I think Fodor has an excellent point, and if I have understood Bryan right, I am pleased for the coherence of his project that he has understood it. Complexity is not the issue--its a basic conceptual one.

    ReplyDelete
  12. Not strictly apropos perhaps, but Steve Jones has a two-parter in which I believe he critiques evolutionary psychology, beginning tonight on Radio 4 (9pm).

    ReplyDelete
  13. What is meant by artificial intelligence? a machine that can replicate the thought processes, decision making processes, is it the intention that this machine will be intellectually superior because it lacks human frailty? It would seem to be an impossible dream.
    The present state of the computer processor is a workaround, they have put more hamsters in the wheel, but running side by side and in any case much of the time and effort spent increasing processing capability has been because of the demand for graphics, always more graphics. Willard mentions the application of computer technology in the games industry, does this give us an insight? billions spent on developing the computer to enable it to operate more complex toys? If this is correct what does this teach us about the future application of AI?
    Will AI merely be the controller of a new generation of electro-hydraulic-mechanical devices, or is there a more sinister purpose afoot.
    Recent history would caution us to beware, remember Oppie in his shed with a fag on and wearing that daft hat.

    ReplyDelete
  14. I'm not sure what is meant by 'artificial intelligence' but I suspect it might have less to do with having true intelligence and more to do with having the appearance of intelligence, in the same way that Wayne Rooney naming his son 'Kai Wayne' will fool some people into thinking it's the sophisticated choice.

    ReplyDelete
  15. Does this mean that the physical evolution of our brains is independent of the environment in which they evolved? i.e. if we all lived in some sort of cave with very little natural variability, would our brains still have the same structure?
    I would have thought (!) that a simpler world would have led to a simpler brain structure.
    I get the feeling I'm missing something (OK, a brain) - perhaps someone could explain it to me?

    ReplyDelete
  16. When Artificial intelligence meets its human counterpart on equal terms will the battle of Megiddo begin.

    ReplyDelete
  17. With Thanksgiving Day approaching and to thank for the customer,there are amazing ugg boots sale promotions on our company.We give you a fantastic gift that is buy one and get one free,once you buy one of ugg 30th anniversary,ugg 5359 nightfall,ugg 5245 ultra tall,ugg 5691lo pro button or 5818 sienna miller other ugg boots on our website, we will give you a belt, hat or a wallet as a gift. http://www.uggboots4buy.com/

    ReplyDelete
  18. Thank you for the great job!In order to meet Thanksgiving Day, recently,our company is on ugg boots promotional sales and the prices of all products such as UGG Sienna Miller or UGG 5803 Bailey Button are lower than any time!And the biggest surprise is that you can buy one get one free,that is once you buy an ugg classic short , ugg 5125 coquette, or uggs snenna mille,you can get an belt, wallet,or hat free.Don't hesitate, to get a free present, check it on http://www.uggsnowbootsbest.com/

    ReplyDelete
  19. In order to enlarge and refresh our market,with the advent of Thanksgiving Day,we encourage both newcomers and loyal customers to browse our sale promotions.The price of our AF Men Trousers,ED Hardy Handbag or ED Hardy Women Suitsare lowest than anytime and anywhere,there are also lots ED Hardy Women Jeans and UGG 5818 Sienna Miller that you can choose of course high quality.If the goods you choose are more than $100,you can pick up a Belt,Cap,Jewelry,or Wallet Free.Time Limited, Don't be hesitation,pick up your favorate goods,Free Shipping & Fast Delivery. http://www.edfashionclothes.com/

    ReplyDelete
  20. Activity is the only road to knowledge . Do not, for one repulse, give up the purpose that you resolved to effect. air Jordan Love is ever the beginning of nike shox knowledge as ato shoes fire is of puma light. The sting of a reproach,is the creative recreation truth of it.A mothers love never changes. http://www.nikeaf1jordanshoes.com/

    ReplyDelete
  21. This is like Discuss 4 woven together with Discuss 1.

    I had mentioned that we are probably off the track on our cosmic model. And this discussion is showing how similarly, we are off track when it comes to AI. Our biases are preprogrammed into our computer systems. The great thing, is that if we keep massaging the modules, and keep mixing in enough abduction, and throw in some other biases we come up with, we might come up with a model of how we think. This would lead to theories on how and why we are the way we are, with proponents calling them the best humankind has ever come up with, which would lead into a massive movement for groupthink, similar to what we have in the Dawkins movement today.

    In Fodor's statement there is still a bias that psychology has no bearing on the question of evolution--and I read here not the word "question" so much as the word "assumption". It's a cavalier attitude. In the last 100 years, psychology, which used to be the study of the psyche, such as what Freud, Jung, and Assagioli did, became the study of behavior instead. Easy results were not forthcoming, so instead of looking where important information might be hidden, we decided to look only in the light. We decided to redefine psychology as the study of behavior, something which can be objectively observed, measured, and quantified. Fodor's attitude is simply to reassert this historic process of ignoring, or to remain ignorant of what the psyche holds for us. He even throws in complex behavior as being irrelevant. So now if the going gets tough with behavior, we can ignore that as well as ignoring the phenomena of the mental, even the spiritual processes.

    ReplyDelete
  22. Thank you for the message offer!There are amazing ugg sienna miller,ugg kid boots and ugg night fall sale promotions on our company to thank for the customer of http://www.uggboots2buy.com/ and to embrace the coming Thanksgiving.You can Get One belt or hat or wallet free once you buy an ugg boots,such as ugg anniversary or ugg bailey buttonon our website. Action is louder than speak. Hurry up and you may be the first one get the gift.

    ReplyDelete
  23. Thank you for your article!A light heart lives long. Early to air jordan women bed and early to rise, makes a nike shox man healthy, dior high boots wealthy and wise Sloth , like ed hardy rust, consumes faster than labor wears. gucci boots life is the art of drawing sufficient conclusions from insufficient premises. http://www.onestop-onlineshopping.com/

    ReplyDelete
  24. I was just reading a poem that made me associate the terms "artificial intelligence" with "fictional character". As soon as we start to objectify who we or someone else might be, we make up an artificial intelligence, a fictional character:

    Fictional Characters ---a poem by Danusha Laméris

    ReplyDelete
  25. How did Fodor measure complexity? On what scale does the complexity of the human brain exceed that of (say) the molecular-level interactions involved in a blade of grass converting sunlight to sugars etc.? Predicting the flight of a foot(soccer)ball when struck by a booted foot(foot), factoring in forces (actionary and reactionary), wind directions, air viscosity etc is hugely complex - way beyond our computational powers I think, even with machinery - but David Beckham does it very well and repeatedly.

    I think David Beckham refutes hard AI, not sure about Fodor - you'd have to ask David about that.

    ReplyDelete
  26. When Artificial intelligence meets its human counterpart on equal terms will the battle of Megiddo begin.

    I think that might have happened on this thread, Malty. The Uggnike Bot wins.

    ReplyDelete
  27. Parallel programming is used specifically to serve working software developers, not just computer scientists. It is a complete, highly accessible pattern language that will help any experienced developer "think parallel"-and start writing effective parallel code almost immediately. Instead of formal theory, it deliver proven solutions to the challenges faced by parallel programmers, and pragmatic guidance for using today's parallel APIs in the real world.

    ReplyDelete