My mother is a more patient human being after having raised a child who incessantly asked, “Are we there yet?” That information, often out of reach for a frustrated toddler, carries with it a feeling of reward. The majority of us are all too familiar with the urge to know more about the future, whether it is an exam grade, an experimental result, or the status of a new job.  Prior knowledge frequently has no effect on the actual outcome of the event – we’ll get the same grade regardless – and yet we still desperately want to know. This leads to what scientists refer to as “information-seeking behavior” – our mind craves relevant information. The neural basis behind this seemingly universal desire has eluded scientists for some time, but the wait is over.

Contemporary theories of reinforcement learning are rooted in the dopaminergic reward system.  Dopamine neurons in parts of the midbrain, such as the ventral tegmental area and substantia nigra pars compacta, play a vital role in the expectation of reward. Most of what is known about these neurons comes from electrode recording experiments with rhesus monkeys. Not surprisingly, these neurons respond to primitive rewards, such as food and water. They signal a monkey’s expectation of rewards, but what was not known until now is whether these same neurons might also signal expectation of information. To test for this preference for information, which is a cognitive reward, a new paradigm needed to be put in place. Ethan Bromberg-Martin and Okihide Hikosaka, both at the National Eye Institute, developed a brilliant behavioral task that opened the door. 

In the experimental design, monkeys were placed in front of a computer screen and were trained to perform a saccade task, in which they learned to direct their gaze at specific areas. The monkeys were first given the option of choosing between one of two colored targets. One of these targets would give the monkey advance information about its future reward. The advance information came in the form of visual cues, one representing a large reward and the other a small reward.  Choosing the other initial colored target revealed cues that were randomly associated with reward size, thus possessing no informative value.  After only a few days of training, the monkeys showed a clear preference for choosing the informative colored target.

The researchers then tested to see when the monkeys wanted the information.  In this scenario, the monkeys were again initially presented with two colored targets.  One of these targets had informative value while the other did not.  The difference was that the monkeys always received informative cues just before their rewards.  The choice each monkey had to make was whether to see an earlier informative cue.  Despite always having a delayed informative cue, regardless of which initial target they selected, the monkeys preferred to have advance information as soon as possible.  Like high-school seniors waiting on their SAT results, the monkeys wanted to know, and they wanted to know right now.

As humans who experience anticipation almost on a daily level, we can easily understand why monkeys would prefer immediate information. What exactly in our brains, though, might be responsible for such a behavior? Working off the hunch that the midbrain dopamine neurons are heavily involved in reward expectation, such as the arrival of a squirt of juice, the researchers hypothesized these same neurons might also be signaling the expectation of information.  As it happens, when colored targets were presented individually, the dopamine neurons responded more strongly to the informative target than the random target. This result supports the notion that midbrain dopamine neurons are coding for both primitive and cognitive rewards.

So why do dopamine neurons treat information as a reward? It’s easy to see how treating information this way might be a useful evolutionary adaptation. For many animals, each day consists of numerous decisions that pertain to eating, reproducing and socializing. Obviously, having access to more relevant information – such as knowing where the food is located - allows animals to make better decisions. Furthermore, having access to such information might give us better control over our environment, thus increasing our chances of survival.

When asked if their work might have clinical applications, Drs. Bromberg-Martin and Hikosaka point out that deciphering the neural code behind these systems gives insight into basal ganglia disorders. Parkinson’s disease is characterized by the death of dopaminergic neurons in areas of the midbrain. While the disease is most often associated with motor dysfunction, in some cases there are also cognitive impairments. One such impairment is a patient’s gradual loss of motivation and ability to learn from rewards. Since the same dopamine neurons are encoding rewards and information, the death of these neurons may reduce the drive to seek information, as well.

Dr. Hikosaka is also enthusiastic about where the research is going next.  It has long been thought that there are two levels of decision-making: a conscious level taking place in the cerebral cortex and an unconscious level in the basal ganglia. The story is not so simple, he says, because these two systems are connected via the midbrain dopamine neurons.  Perhaps future work will reveal how our conscious and unconscious decisions are influencing one another, all due to this very busy population of dopamine neurons. 

It’s often remarked that “ignorance is bliss.” However, when you look at ignorance from the perspective of the brain a very different picture emerges. Our brains, and the brains of other animals, have evolved to find information rewarding. In fact, not knowing is stressful, which is why we strive to decrease that uncertainty whenever possible. We want the information and we want it now!