[ExI] predictive neurons?

Mike Dougherty msd001 at gmail.com
Fri Jan 29 02:09:43 UTC 2010


My thought below the quote

from: 'How We Decide' And The Paralysis Of Analysis [Fresh Air 2010-01-22]
http://www.npr.org/templates/transcript/transcript.php?storyId=122854276

[ begin quote - about 1/3 into the article ]
But dopamine's even more important than that. It also seems to
modulate many of our feelings, from the pleasure of eating chocolate
cake or a crack high to feelings of fear and disgust.

So it, in many respects, is one of the key neurotransmitters of
emotion in general. And I think when you talk about decision-making,
that's part of what makes it so important. And I think there's also
been a lot of really interesting work, much of it done by a scientist
named Wolfram Schultz at Cambridge University, that have shown how
dopamine neurons react in detail to the real world.

GROSS: Can you tell us about one of those studies?

Mr. LEHRER: His experiments observe a really elegant protocol. He'll
monitor individual dopamine neurons in the brain of a monkey, and
he'll show that at first, these neurons respond to a reward, to a
squirt of juice.

So if you give the monkey a squirt of juice, these dopamine neurons
will fire, and the monkey experiences the pleasure of getting a squirt
of apple juice. That's the reward. But these neurons quickly adapt to
the pleasure. So they quickly stop firing.

That, you know, that makes perfect sense. You have an iPod. It makes
you happy for a day or two, and then it stops, you know, giving you
squirts of joy every time you look at the iPod. We adapt to these kind
of hedonic pleasures.

But what Wolfram Schultz found is that if you then play a bell before
giving the monkey a squirt of juice, the dopamine neurons will fire
whenever you play the bell. And if you flash a light before playing a
bell before giving a squirt a juice, they'll fire whenever you flash a
light. And if you play a song before flashing a light before ringing
the bell, etc., etc., the dopamine neurons will always try to predict
the reward.

So they're called prediction neurons. Their job is to predict the
first event that signals a reward is coming, a squirt of juice is
coming. And so you can begin to understand how these neurons are so
important in terms of allowing us to make sense of reality, in terms
of finding the patterns and correlations and causations that allow us
to actually figure out what's going to happen, and most importantly,
from the perspective of evolution, figure out when our squirts of
juice are going to arrive - try to make sense of those rewards.

You know, and so you can begin to understand why they're so important
for decision-making in terms of allowing us to make decisions that
allow us to maximize our rewards.
[ endquote ]

  Is it possible that over a lifetime of event chaining that some
people do build some twitchiness that successfully predicts events in
their local environment from a complex number of cues prior
experience?


And for those interested:

Wolfram Schultz at Cambridge University:
http://www.neuroscience.cam.ac.uk/directory/profile.php?Schultz

Tobler PN, O’Doherty JP, Dolan R, Schultz W (2007), “Reward value
coding distinct from risk attitude-related uncertainty coding in human
reward systems” J Neurophysiol 97:1621-1632
http://www.neuroscience.cam.ac.uk/publications/pubInfo.php?foreignId=pubmed%3A17122317

Temporal Difference Model Reproduces Anticipatory Neural Activity
http://portal.acm.org/citation.cfm?id=1120460


More information about the extropy-chat mailing list