Greg Detre
Wednesday, February 12, 2003
common sense as common conclusions?
most of Piaget�s conclusions have been borne out, despite only having 3 subjects, even that children don�t have a real sense of number till 5 or 6
just because birds get agitated if they return to their nest and there�s only 2 rather than 3 eggs left doesn�t mean that they can count
he had read that children don�t recognise faces until c. 6 months (recognise people�s speech by 3 months), until saw an experiment
infants of 1 week get excited when their mothers come into the room, but it turned out to be just the hairdo
the emotion machine never mentions neurology
doesn�t want to cite evidence that will probably be discredited soon
ch on credit assignment � part of the reason that people can learn so much faster than other animals
he�s been trying to convince NASA to have unmanned space stations � 1/10 of the cost � 3 robots, so that they can fix each other, do experiments, and send back results � but they�re convinced that the public won�t support unmanned space exploration
string � �you can�t push things with a string�
qualitative physics � ben kuijpers(sp???) �
emotions as selecting out and restricting thought � it�s rationality that�s hard
why don�t we call �hunger� etc. an emotion?
emotion as a big suitcase of ideas
he does view emotion as flexible behavioural strategies, I think
anger isn�t just an inhibition of censors and restriction of ways of thinking � it does make you think new things� but how could you tell if those �new things� are just unconscious all the while�???
improved reactions�
but emotions aren�t that low-dimensional, and they needn�t really be global!!! hmmm. maybe.
free will as a pernicious idea � free will is not a way to make a decision � it�s more like �my decision process has come to an end. it selected this particular option. I have no idea about how that was done.�
a smart robot will need to be able to:
remember what you�ve recently done
make descriptions of those events
reflect on what you should have done
make and using models of yourself
describe yourself as an entity
evaluate your various options
envision what would happen if
formulate new goals and plans
hear conversations in your mind
make descriptions of your ideas
recognise your level of wakefulness
notice some things, but not others
represent your emotional state
represent your body�s condition
he wouldn�t acknowledge what we might mean by conscious experience/qualia � what could there be besides the computation, he says?
but I can
point to what I mean by consciousness by pointing to the boundary outside me of
things I�m not conscious of
the thing about emergent properties is that you can predict all of the emergent properties � the barrier is epistemological, one of difficulty
Hameroff conjectures that neurons have global quantum state
�consciousness is the popular name for our single most serious limitation: that our top-level representations depict such a very small fraction of each situation�
he considers that Hume, Kant and Descartes� ideas on the workings of the mind are the sort of thing that you�d give a freshman an �A� for and a senior a �D� :)
can�t account for why Piaget�s discoveries were so late � did everyone just agree that children were little people???
in contrast, Galileo lacked the tools to formulate his evidence as an inverse square law
if you�re going to blame religion for holding science back, why didn�t anything happen between Archimedes and 430�
Twain: �The way to make money is to buy stocks when they�re cheap and sell them when they�re expensive. If they don�t go up, don�t buy them.�
unlike animals, we think a lot about internal problems:
some obstacle is in the way
my goal didn�t achieve its supergoal
I don�t have access to knowledge I need
one of my predictions has failed
two of my goals seem to be incompatible
I cannot get this method to work
more self-reflective
this problem is too hard
I have too many goals
I don�t have any worthwhile goals
he wants to know if anyone has thought about the kind of problems
see questions doc
what does he say about mental effort???
a kind of currency between agents
why do they need currency if they work in parallel??? that�s just the implementation though�