Representation of Value in the Primate Brain
published: Aug. 9, 2010, recorded: July 2009, views: 2908
Report a problem or upload filesIf you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.
Pigeons really like millet seed, monkeys crave juice, and humans get a kick out of winning money. While all animals don’t enjoy the same rewards, Paul Glimcher has discovered some common features in the way animal brains learn to recognize and pursue something of value.
Glimcher is one of the founding fathers of the young field of neuroeconomics, in which economic theories help inform investigations of brain function. It’s not surprising then, that his approaches include game theory as well as measuring the firing of single neurons. Glimcher’s talk details his research from the past 15 years, what he describes as an attempt to “add something” to the classic studies on the basal ganglia circuit conducted by fellow symposium speaker Okihide Hikosaka. From Hikosaka’s data and other research, Glimcher came to believe that neurons of the substantia nigra (part of the basal ganglia) were coding for something of worth to an animal, but that these neurons were “responding not to reward per se, but to deviations to expectation.” For instance, if a pigeon expected a delivery of millet seed following a conditioned cue, no neurons fired, but if the reward was delayed, then suddenly delivered, the pigeon would find its initial prediction in error, and its neurons burst into action.
Various models emerged to capture the ways in which these neurons, energized primarily by the neurotransmitter dopamine, enabled animals to adjust expectations about and predict rewards. But Glimcher found fault with others scientists’ “conditional parameters.” He says, “As an economist, this is frustrating.” So he developed three mathematical axioms for testing the so-called Reward Prediction Error (RPE) models. His work “suggested a way of unifying the data,” with the notion that the basal ganglia learns “the values of actions in a quantitative way … from the dopamine neurons and the incoming stimulus.”
Glimcher hypothesized that dopamine neurons take the value of a reward just received, “and subtract it from a weighted exponential average of previous rewards, and if there’s no mismatch, there should be no firing of…dopamine neurons.” Human, monkey and pigeon studies -- based on gambling, juice, and seed rewards, respectively -- solidified his notion that dopamine neurons are part of an RPE encoding system where they convey the differences between rewards expected and rewards received. This has led Glimcher to believe that “one of the principle functions of the basal ganglia is to learn the values of our actions, represent them, and pump out the data to produce choice.”
Link this pageWould you like to put a link to this lecture on your homepage?
Go ahead! Copy the HTML snippet !