How the brain decides what to learn

To learn more about the world, an animal must do more than just pay attention to its environment. He must also know the views, sounds and sensations of his environment that are most important and watch how the importance of these details changes over time. Yet the way humans and other animals follow these details remains a mystery.

Now, Stanford biologists report October 26 in Sciencethey think they understand how animals sort details. Part of the brain called the paraventricular thalamus, or PVT, plays the role of guardian, ensuring that the brain identifies and tracks the most salient details of a situation. Although the research, funded in part by the Wu Tsai Neuroscience Institute's Neurochoice Initiative, is currently limited to the mouse, the results may one day help researchers better understand how humans learn or even treat the disease. addiction, said the main author Xiaoke Chen, assistant professor of biology.

The results are a surprise, Chen said, partly because few people thought that the thalamus could do something so sophisticated. "We have shown that thalamic cells play a very important role in monitoring the behavioral significance of stimuli, which no one else has done before," said Chen, also a member of Stanford Bio-X and Wu Tsai Neurosciences Institute.

Decide what to learn

In its most basic form, learning is about feedback. For example, if you have a headache and are taking a medication, you expect the medicine to clear your headache. If you are right, you will take this medicine the next time you have a headache. If you make a mistake, you will try something else. Psychologists and neuroscientists have studied this aspect of learning in depth and have even made the connection to specific parts of the brain that process feedback and stimulate learning.

Still, this image of learning is incomplete, Chen said. Even in relatively simple laboratory experiments, not to mention life in the real world, humans and other animals need to understand what to learn – essentially, what about feedback and noise . Despite this need, it is a problem that psychologists and neuroscientists have not paid enough attention to.

To begin to remedy this, Chen and his colleagues taught mice to associate particular odors with good and bad results. One smell signaled the arrival of a sip of water, while another signaled to the mouse about to blow a breath of air to the face.

Later, the researchers replaced the air puff with a slight electric shock, which would probably attract a little more attention. The team discovered that PVT neurons were tracking this change. During the air swell phase, two-thirds of the PVT neurons responded to both odors, while an additional 30% were activated only by odor-signaling water. In other words, during this phase, the PVT reacted to both good and bad results, but it was more effective to the good ones.

During the electric shock phase, however, the scale has shifted. Almost all PVT neurons responded to shock, while about three quarters of them responded to both good and bad outcomes.

A similar change occurred when the mice had been filled with water. Now that water was less important for mice, PVT was less sensitive to water and more sensitive to air flushes, which means that it reacted better to poorer results and less to the good ones. Taken together, the results showed that the PVT was monitoring what was most important at the moment – the good result when it outweighed the bad, and vice versa.

A new place to look and edit

The results suggest several broader conclusions, Chen said. Perhaps the most important point is that other researchers now have a place to look at – the PVT – when they want to study how the attention to different details affects how animals learn and what 'they learn.

Neuroscientists also have a new way of controlling learning, Chen said. In additional experiments with genetically engineered mice so that the team could control the activity of PVT with light, the researchers found that they could inhibit or enhance learning – for example, they could more quickly learn how to mice that a smell that no longer reliably signaled the coming of water the smell had gone from water signage to the signaling of a shock.

These findings may indicate new ways of modulating learning – in the mouse at the moment – by stimulating or suppressing the activity of PVT as needed. They also highlight, in the long run, ways to help treat addiction, he added, helping addicts to no longer understand the connection between taking a drug and the drug. subsequent peak.

The other authors are Liqun Luo, Professor Ann and Bill Swindells of the Faculty of Humanities and a researcher from the Howard Hughes Medical Institute; postdoctoral fellows Yingjie Zhu and Gregory Nachtrab; and graduate students Piper Keyes and William Allen.

The research was funded by the Whitehall Foundation, the Ajinomoto Innovation Alliance Program, a Terman Scholarship, the Firmenich Next Generation Fund, the Brain and Behavior Research Foundation, the National Institute on Drug Abuse, the Fannie and John Hertz Foundation, and the National Science Foundation. .