This is an archived story; this page is not actively maintained. Some or all of the links within or related to this story may no longer work.
For the latest University of Minnesota news, visit Discover.
Applying learning theory and studying brain signals should reveal triggers of addictions, says U of M neuroscientist David Redish.
By Brenda Hudson
From eNews, April 21, 2005
Addiction can be disruptive and devastating. Those whose lives have been affected by alcoholism or drug abuse don't need a theory to show that addiction is irrational. Yet a working model that explains addictive processes could be used to better understand and to make predictions about addictive behavior, perhaps leading to new, more effective treatments.
U of M neuroscientist David Redish has developed a computational theory of addiction. Addiction, he believes, lies partly in the brain and partly in the way we learn. "Addiction messes up your system," he says, by mentally and physically altering the way your body responds to a particular stimulus. Redish's new way of looking at addiction combines two well-established areas of research--learning theory and the effects of drugs on dopamine.
"We already know that drugs of abuse affect dopamine," neurotransmitters that respond to drugs and other stimuli, he explains. Cocaine, for instance, is known to produce an increase in dopamine at the synapse. But Redish, whose expertise includes learning and memory, took this knowledge one step further by incorporating the learning model known as temporal-difference reinforcement. According to Redish, dopamine serves as a "reward-error learning signal" in the brain. That is, if a response (or reward) is better than predicted, more dopamine is released; if worse than expected, less.
Redish provides an example of how this works. "Say you're at a vending machine and you put in a dollar. But instead of one bottle of soda, as you'd expect, you receive two bottles. You think, 'This is better than I thought.' As a result, your dopamine response goes up. If you put a dollar into the machine but don't receive your bottle, your dopamine goes down. On the other hand, if you put a dollar into the machine and receive a bottle, as expected, there is no change in the dopamine. In this way, we learn to make predictions about how much reward to expect in future situations." It's a phenomenon that has been directly observed in animals, he says. "As the animal's prediction changes, its dopamine changes."
But what does this mean in terms of addiction?
In the model above, the soda is the reward; the prediction is what you expect to receive when you put a dollar into the vending machine (i.e., one bottle of soda). But with addictive drugs, the release of dopamine seems to trigger an "error signal," tricking the body into responding as if the drug was "better-than-expected." This is what leads to addiction, says Redish.
"Every time you take the drug, the brain gets a signal that the prediction wasn't good enough. So your brain learns to value the drug more and more." This irrational assignment of value to the drugs explains a number of phenomena commonly associated with addiction, such as a person's willingness to increasingly pay for drugs. "Every time a person takes cocaine," says Redish, "the value of the reward goes up; therefore, the person's willingness to pay goes up--to infinity."
Redish hopes to test his theory further by collaborating with colleagues in other disciplines. "There's a lot more to do," he says. "[Because] many of these drugs affect other neurotransmitters in the brain."
Redish presented his computational theory of addiction in a paper published in the December 2004 issue of Science.