Greg Detre
Wednesday, March 05, 2003
I guess
even if he�s right that you could model a brain on a pentium (which I still
find highly implausible), you�ve got to speed that up hundreds of times because
we�re not prepared to let that simulation run for 10 years
he thinks
that there must be a serial controlling process somewhere that sends datasets
to the other bits, and tells them to do so many cycles of hill-climbing or
whatever�
attacks
Fodor for not being a thinker
not thinking hard enough about how it works
considers common sense to be an undecomposable module
both Sloman
and Dennett were students of Ryle
Minsky:
what good ideas have Nagel and Block had?
talks about
MNs as allowing animals to imitate
but actually, can chimps imitate???
he thinks you have to be smarter to imitate than to learn for itself
do autistic
children have trouble with expression + face recognition???
why do we
forget so much???
re Damasio
� but the body is so dumb
�emotion
as a context of cognition�
emotion as a type of thinking, rather than something separate that
affects thinking
someone
said that there�s a stage of child development where they treat themselves as
the third person (at least when they talk) � can that be true???
children appear to divide the world into things that move of their own volition and things that don�t
the sense of a single self is so compelling because we use that
self-model as a means of predicting
does Turing actually say that the test tests for intelligence???
he thinks that anyone who thinks the Turing test is valuable is not worth talking to � the Turing test test J !!! pah
if we
had a theory of problem types, we could completely do away with the Turing test
the mystery about using associations with familiar items in your bedrooms is that if you do the same thing the next day, you don�t get them confused
remembering numbers � each number is connected with two or three consonants � most strings of 3 or 4 numbers have a good word that you can use to chunk them together
the slots in your STM are probably of different sizes
does the LTM of anterograde amnesics degrade???
if you carry on talking to them for ages, do they remember the details you�ve told them
g-factor � Eysenck showed that speed of identifying longer lines was a better correlate of overall IQ than any single one of the IQ test components
intelligence is based on credit assignment � there might be a single central bit that decides how and what to store � there might be 3 or 5 copies that know/remember a lot
is he arguing that these are blackboards??? I thought that translation between representations was by a series of steps � see micromemes
Braitenberg
hippocampus as cache
common sense helps you with credit assignment because you have expectations about what�s probably relevant that help shortcut
apparently you can learn to wiggle your ears with biofeedback