Skip to Main Content

A version of this Q&A first appeared in STAT’s Morning Rounds. You can subscribe to the free newsletter here.

If you wonder why we forget our dreams, new research in mice might answer that question and others related to memory. U.S. and Japanese scientists discovered that when certain neurons fire during REM sleep — when most dreams occur — they control whether the brain remembers new information.


Thomas Kilduff, director of the Center for Neuroscience at SRI International and a co-author of the paper, talked with STAT about the study and its implications. This interview has been condensed and edited.

What prompted you to study forgetting while dreaming?

Well, we didn’t start out to study dreaming, although that ends up being an implication of our results. There were four threads that motivated this study. There is a lot of societal concern about neurodegenerative diseases such as Alzheimer’s disease, there is a lot of research interest in the neural bases of learning and memory, there is research controversy about the role of sleep in learning and memory, specifically whether REM sleep, slow wave sleep, or both types of sleep are important for memory consolidation. And there is a lot of recent research in animal models of Alzheimer’s disease suggesting sleep disruption may accelerate the accumulation of beta-amyloid in the brain and, conversely, if you enhance sleep, that may slow this process.


In collaboration with my Japanese colleagues Akihiro Yamanaka and Akira Terao at Hokkaido University, we published a paper in 2014 that described a transgenic mouse in which we could eliminate the melanin-concentrating hormone neurons in the hypothalamus. These mice showed a partial insomnia. Because of the links between sleep and memory, we wanted to evaluate whether learning and memory were affected in these mice.

What did you find?

The very surprising result was that mice without MCH neurons performed better on learning and memory tasks than mice with intact brains. And intact mice performed more poorly on the memory tasks when MCH neurons were activated — as they are during REM sleep — and better on the memory tasks when these cells are turned off during REM. These results suggest that activation of the MCH neurons during REM sleep may interfere with memory consolidation — that is, MCH neurons may facilitate forgetting.

So melanin helps forgetting? And it promotes REM? Also, appetite?

Actually, we studied hypothalamic melanin-concentrating hormone neurons, not melanin. Which is in skin – that is another story.  

How do you test memory in a lab mouse?

There are standard behavioral tests of learning and memory you can use. One is novel object recognition, one is contextual fear, and another is navigating a maze. You can look at whether the animal learns better or retains information better than control animals.

Why would mice — or people — need a mechanism to forget?

The brain has an information storage problem:  We can’t — and need not — remember everything.

Scientists have made good progress in identifying the neural bases of learning and memory and have found that a brain region called the hippocampus plays a particularly important role. Less effort has been spent studying forgetting than learning, but forgetting is important to eliminate useless information. It’s also an active process. Some memories are consolidated during sleep but other information must be forgotten to prevent information overload. 

Most dreams have informational content that is not useful for waking activity and are probably best forgotten. We and others have shown that most MCH neurons are active only during REM sleep, which is when most dreams are reported in humans. Since we have shown that activation of MCH neurons interferes with memory consolidation, activation of this neural circuit during REM sleep may indeed make it more difficult to remember our dreams.

What’s new in what you found?

The mice in which the MCH neurons had degenerated actually performed better on learning and memory tasks than intact mice. Our results are consistent with previous studies that supported a role for REM sleep in forgetting, but now we identify the particular neural circuit that is responsible.

What’s next?

Understanding the neural basis of learning and memory is a huge area of neuroscience research because of its implications for our everyday lives, as well as conditions with cognitive dysfunction such as Alzheimer’s disease. Prior to this work, no one suspected that MCH neurons had anything to do with forgetting.

What about making the leap to humans?

Selective activation of the MCH neuron-hippocampus neural circuit could result in elimination of unpleasant memories as occur in PTSD. 

What are limitations of your work to keep in mind?

The study was conducted in laboratory rodents, not humans, so we can’t ask them about their dreams. Mice are nocturnal (active at night) with sleep and wake throughout the day and night, whereas most humans have a single “consolidated” sleep period that usually occurs at night.