Pavlovian conditioning is the classic example of associative learning. A dog that always hears a bell ring immediately prior to being fed will eventually salivate at the mere sound of the bell. At the heart of this type of learning is the feel-good neurotransmitter dopamine, which helps animals make positive associations to stimuli that herald pleasurable outcomes. But there’s a flip side to dopamine signaling: the development of addictive behaviors. Columbia University neuroscientist Eleanor Simpson discusses a study that pushes the boundaries of what’s known about dopamine, associative learning, and addiction (Nature, 469:53-57, 2011).
The Scientist: How does dopamine help the brain form associations between signals and the rewards that follow?
Eleanor Simpson: One of the major theories of how dopamine is involved in learning is by encoding reward prediction error. This is based on the observation that dopamine cell firing rates increase when an unexpected reward occurs and decrease when an expected reward does not occur, suggesting that dopamine codes the difference between what is expected and what actually happens. The theory suggests that this difference provides the teaching signal.
An alternative hypothesis is that rather than dopamine teaching the reward prediction error, it drives something called “incentive salience.” Meaning it drives the motivation to attend to or pursue stimuli that come to predict reward. It informs us about something important. So the other idea is that dopamine tells you when something is relevant.
TS: In this study, University of Michigan neurobiologist Huda Akil and colleagues measured dopamine levels in the brains of rats as they underwent classic Pavlovian conditioning: being presented with a signal a few seconds before receiving food. Did their results surprise you?ES: The authors showed that animals learned that the cue predicted the reward, but they learned it in different ways. Some animals, which they name “goal trackers,” paid attention only to the reward itself. During the cue presentation, their attention was focused not on the cue, but actually on the reward. On the other hand, some of the animals were “sign trackers.” That means that in the presence of the cue they actually attended to the cue, instead of looking for the reward itself. What’s really important about this differentiation is that these animals learned to the same extent. Their learning of the association between the cue and the reward was the same, it’s just that their focus was either on the cue or on the reward. But in the goal-tracking animals, despite the fact that they learned the association with the same speed and their learning curve looked the same, they actually didn’t need dopamine at all to do this. So this shows that some forms of associative learning can occur without dopamine signaling.
TS: To measure dopamine in the brain, the researchers used fast-scan cyclic voltammetry—a technique not commonly employed for these types of experiments.ES: A lot of theories on the role of dopamine in learning have come not from measuring dopamine release, but from recording the activity of the cells that release dopamine. In animals, the firing rates of dopamine-releasing cells are recorded, and how those changes in firing rates would affect dopamine release is inferred, without actually measuring levels of the neurotransmitter. But in this paper they are directly measuring the release that occurs, and that’s because they’re using an electrochemical technique which the authors, and others, have developed for use in awake, behaving rodents. I think it’s a big step forward that you can follow, in subsecond timing, how dopamine is released during these different behavioral tasks.
TS: How do these findings mesh with our understanding of reward learning and addiction?ES: The initial study was done in lines of rats specifically bred for their response to novelty. So, in that way, there’s been a genetic selection for this type of behavior, and it seems that animals that are highly responsive to novelty sign-track, while the animals that are less responsive to novelty goal-track. One of the implications of this is that animals that pay attention to the signal become much more behaviorally energized by a signal rather than the actual reward. And that’s something that people consider a mechanism of addiction. So people who are addicted to certain substances or certain activities become much more energized to cues which predict or are associated with those reinforcers than people who are not addicted. In the human population there are differences between people who like to take risks, people who don’t, and people at a higher risk for forming addictive behaviors. In this simple animal behavioral paradigm of goal tracking versus sign tracking, individuals who become focused and energized by the cue which is signaling the reinforcement would be more susceptible to addictive behavior.
Eleanor Simpson works on modeling dopamine dysregulation in patients with schizophrenia at the Columbia University Medical Center.