Why cocaine addicts keep making bad decisions

Chronic cocaine use skews the brain's reward-related error management, according to a new study. Here, an anti-narcotics agent in Panama hacks open a package of confiscated cocaine.
(Arnulfo Franco / Associated Press)

Chronic cocaine use alters brain circuits that help us learn from mistakes, a new study suggests.

The study, published online Tuesday in the Journal of Neuroscience, could offer a biological marker for the cycle of destructive decisions that many addicts exhibit.

Researchers measured EEG signals from a region of the midbrain that has been associated with how the brain manages errors in reward prediction. Neurons there release and absorb more dopamine when things go better than expected and less when events fail to meet expectations, and remain unperturbed when outcomes match predictions.


That proportional feedback helps explain why we’re so pleasantly surprised at unexpected rewards, so sorely disappointed at unforeseen penalties, and relatively blase about the “predictable” outcomes in between.

“The brain learns from it -- whether you should go ahead with this experience the next time or you should stay away from it,” said the study’s lead researcher, Muhammad Parvaz, a cognitive neuroscientist at the Icahn School of Medicine at Mount Sinai in New York.

Among chronic users, said Parvaz, “the worse-than-expected response was not there.”

Researchers compared a range of EEG signals among people who don’t use cocaine with those from two chronic-use groups: one that tested positive for cocaine use within the past 72 hours and one that had tested negative.

All 75 participants played a computerized game in which they guessed which of four doors hid prizes. The trials offered randomly different odds of winning, from 1 in 4 up to 3 in 4, that were flashed onscreen for milliseconds before subjects made their choice.

Subjects also had to report whether they thought they had made the right or wrong prediction before they could see the result. A correct choice, marked by a green upward arrow, meant they had earned 60 cents, while a red downward arrow signaled they had lost 30 cents.

As expected, the feedback signal from nonusers was greater for unpredicted outcomes than for predicted ones.

Across both groups of chronic users, however, EEG readings showed no significant difference between expected and unexpected losses. Their management of negative reward error prediction was impaired.

This could explain why addicts will return to drugs despite the negative impacts of incarceration and loss of money, friends and family, according to the researchers. “They don’t learn from it,” Parvaz said. “They go back to the drug.”

Indeed, those who tested positive for recent cocaine use had relatively normal feedback for positive reward prediction error, according to the study.

“We think this is in line with self-medication hypothesis,” said the study’s principal investigator, Rita Goldstein, a neuroscientist who heads the Brain Imaging Center at the medical school. “Drug self-administration improves response to reward in drug addicted individuals.”

Chronic users who had not tested positive for recent cocaine use showed the worst of both worlds: impaired positive and negative feedback. That’s consistent with the known chemistry of cocaine use -- it tamps down production of dopamine, leaving the neurons unresponsive to just about everything but the drug.
That may explain why it’s hard for abstaining users to respond to incentives to quit, and easy for them to relapse, the researchers suggested.

Goldstein said the reductions in loss-related feedback, “could become important markers that can be used to predict susceptibility for addiction or relapse, or to develop targeted interventions to improve outcome in this devastating, chronically relapsing disorder.”

There are some limits to what can be drawn from the study data, however. Signals from other circuits that could be acting in tandem were not analyzed, the researchers said. And setup of the probabilities resulted in fewer unpredicted losses when chances of winning were just 1 in 4, and fewer unpredicted wins when those chances were 3 in 4.
Some researchers also have argued that the prediction of salience, not of reward, drives the measured feedback, as one study comparing reactions to pain and monetary reward seems to suggest.

Reward your curiosity. Follow me on Twitter: @LATsciguy