Showing posts with label Pinker. Show all posts
Showing posts with label Pinker. Show all posts

Friday, June 19, 2009

What's the Big Idea?

More from Steven Pinker. I was going to take issue with a line about birds not being able to evolve propellers what with all the shouting about how pure natural selection could not have brought about the flagellum that certain organisms have, but in a later paragraph there is a qualification which maintains that propellers cannot evolve in vertebrates which is much better, though I am obviously still trying to come up with a reason for even that not to be true. There are the cow-like animals in The Amber Spyglass by Philip Pullman, which use circular seeds clasped between specially adapted limbs to form wheels, though they are only of use in a particular environment where the geology has created long straight and smooth lines of rock along which the creatures travel. Obviously this is not true natural selection (and I do realise that it is only fiction) but it shows that many such problems are easily surmountable in the mind at least.

Pinker also talks about Natural Selection having no foresight - it cannot reach out and pluck ideal solutions to problems out of the air - in short it cannot do any design. The whole of this chapter seems to hinge around the idea that intelligence is not the pinnacle of evolution - it is just a solution, but it struck me that intelligence allows natural selection to become its own engineer and to have that missing foresight. Intelligence allows us to examine the mechanism by which the characteristics of organisms are passed on to offspring and, once we have enough understanding and tools to manipulate that mechanism, also allows us to change the design through construction rather than by blind luck and survival of the fittest offspring. I suppose that this is a sort of meta-evolution. I make no moral arguments here but the evolution of intelligence in humans allows us the beauty of self-reference, the ability to understand everything about ourselves and to engineer our future.

I remember a short story in Omni magazine, in which a group of scientists were attempting to determine if mice could affect a reward mechanism telekinetically so that is produced the rewards at a higher frequency than that set by the scientists. Two project leads had had to resign after being discovered tampering with the equipment to increase the rate of reward. The end of the story described how the mice, being totally unable to affect the machine telekinetically, were actually tapping into the minds of the experimenters telepathically and getting them to increase the reward rate. This is like meta-evolution. Lamarckian ideas about acquired traits being passed on to offspring have been rightly derided but our minds allow us to circumvent this and to actually add traits to new generations - things which would never evolve because there is no pressure for them to. We can imagine many things that would be useful for humans to be able to do but for which there is just no possible mechanism for the evolution to occur. However, any trait already existing in an organism can ALWAYS be explained by natural selection. Most traits are easy to explain - indeed most can actually be simulated. Some, like the flagellum, are very hard to explain but eventually our knowledge of the mechanisms will uncover the truth of the matter.

Most (though not all) of these mechanisms take many generations and consequently much time, to evolve and this is where the supporters of Creationism and Intelligent Design often bail out, when the sheer idea that billions of generations are required to bring about all the diversity we see becomes too much for them. This partners the idea that there are actually billions of people in the world and that some of them might not have the same views as you. Get over it!

Wednesday, May 27, 2009

Alice in Wonderland



We were watching The Incredible Human Journey last night (delayed by the wrong kind of snow from Sunday or something like that). At one point the discussion turned to the first encounters between modern humans and the incumbent European humans - the Neanderthals. The learned professor who Double Doctor Alice was talking to, suggested that the meeting would have been an profoundly interesting moment and this got me wondering if it will ever be possible to re-construct the exact moment that it occurred. There is a device in Stephen Fry's book, Making History, which allows the viewing of places in past time but the mechanics are never really explained.

I also remember a letter that was published in a list of various "would-be-usefuls" that members of the public had sent to Tomorrow's World, which said that as energy was never destroyed, just converted why couldn't we work backwards to discover what exactly had happened in the past. The correspondent suggested that they would love to learn what Henry V had really said just before Agincourt. The letter was illustrated with a cartoon with the suggestion that the answer was "Good luck lads! Wish I was coming with you." This is sort of like the idea of a Newtonian computer that calculates the motions and masses of all particles and can therefore work out the future of everything. If it works forwards it just as easily works backwards. Anyway we all know that those pesky little Quanta put paid to that idea but I still wonder if we could pick up something from somewhere.

Genetics already has something akin to this idea in the ability to work backwards to determine single ancestors of entire populations but I cannot see of any scientific discipline which would be able to reconstruct moments like this. It would be the ultimate in forensics.

Isaac Asimov wrote a short story about the creation of a time machine and how its inventor tried to keep it secret simply because he realised that rather than using the device to view the great moments in history it would be used as a shabby reason to determine the truth behind trifling domestic incidents which is of course the truth. I'm currently reading How the Mind Works by Steven Pinker which starts with a loose discussion of robots and general intelligence. Pinker makes a good argument against the necessity of Asimov's laws, saying that no machine really needs to be instructed against evil because it is only humans who have the ability to program evil in to things which is a nice lead-in to the truth that human minds are still many thousands of times more complex than even the best-performing artificial mind. I'm not entirely sure of the argument against the laws of robotics - it must be relatively for a machine to accidentally injure or kill a human and putting some simple overriding fail-safes into the system would guard against this. Maybe Pinker says this later or maybe I fell asleep.

Control - April 2019

‘It had nothing to do with me, but I couldn’t help wondering what it would be like, being burned alive all along your nerves.’ I. The Dispos...