Joined: June 2008
|Quote (midwifetoad @ Aug. 22 2012,15:59)|
|You know the driving rules are smuggled into the algorithm.|
Seriously, I have a question for evilutionists. What is the problem with learning as a metaphor for evolution?
I know there's a huge difference in the supporting substrate, but the process seems to be the same. When an organism is learning, There is no invocation of magic regarding where the information comes from.
I see all these battles over the word information, when it seems to me that however loosely the term is used, it can be a useful metaphor. The battle should be to understand that learning is an accumulation of information, and that evolution is learning.
I don't think it is a good metaphor, but I think the discussion sounds fun!
What kind of learning are we talking about? High level human learning as in figuring out how to be more attractive to women? Or neural network back propagation?
In learning, there is no population, and no crossing/mutating of 'memes'. (I think that Dawkins' use of that term referred to cultural change.)
In back propagation in particular, there is a goal. You are training the neural network with a set of examples. The experimenter isn't smuggling information, she is delivering through the front gate.
An IDist might try to dismiss GA results with "That isn't evolving, that is just learning!" (And Bob Marks could spin that line better than anyone, given his strong background in NN research.) I think the paradigms are different enough that there are clear distinctions.
I’m referring to evolution, not changes in allele frequencies. - Cornelius Hunter
I’m not an evolutionist, I’m a change in allele frequentist! - Nakashima