Why do we dream?
Many Minds
The Overfitted Brain Hypothesis
Dr. Eric Coel has written a paper about the overfitted brain hypothesis. He argues that learning is always essentially a trade-off between memorization and generalization. Dreams are kind of like a combative force to learning itself, he says.
You may not remember much about it, but chances are last night you went on a journey. As you slept, your brain concocted a story—maybe a sprawl of interconnected stories. It took you to some unreal places, gave you superpowers, unearthed old acquaintances, and twisted your perceptions. Meanwhile, billions of brains all around you, up and down the tree of life, were probably doing something very similar—dreaming, that is. But why do we do this? What could possibly be the function of these nightly ramblings?
My guest today is Dr. Erik Hoel. He is a writer and a neuroscientist at Tufts University. In a paper published earlier this year, Erik presented a new theory of why we (and other creatures) dream. It's called the “over-fitted brain hypothesis”; the basic idea is that dreaming helps us stay cognitively limber, adaptable—less tied to the particulars of our previous experiences.
Erik and I discuss how he came to this new theory. We talk about how his account develops an analogy between the "overfitting" problem in machine learning and the "overfitting" problem that biological brains face as well. We discuss how his hypothesis can account for the bizarre nature of dream experience. And we consider Erik's provocative suggestion that dreams are really just one type of fiction—biological fictions, if you like—and that other types of fiction may serve similar purposes.
Erik is a fascinating, wide-ranging thinker (there aren’t a lot of neuroscientists who also write novels). And this is a conversation I'll be chewing on for some time. It takes on one of those timeless questions about human experience—why we dream—from an angle that feels fresh and energizing.
Alright friends, on to my conversation with Dr. Erik Hoel. Hope you enjoy it!
The paper we discuss is available here. A transcript of this episode is available here.
Notes and links
4:00 – Dreams have been in the news recently, with reports of an uptick in strange dreams during the pandemic.
9:30 – An early study on “dream deprivation.”
11:00 – An article on the idea that dreams serve memory consolidation.
23:00 – A study showing that we don’t dream about reading or writing.
27:30 – An attempt to solve a Rubik’s cube with a robot hand.
32:00 – An influential paper articulating the synaptic homeostasis hypothesis.
38:30 – A recent paper on the question of whether animals like octopuses dream.
42:00 – We’ve discussed Pinker’s “music is like cheesecake” analogy in previous episodes, most recently in our discussion of the evolution of music.
46:00 – For more on these ideas, see Dr. Hoel’s essay ‘Enter the Supersensorium’—and be sure to check out his new novel The Revelations!
You can find Dr. Hoel on Twitter (@erikphoel) and subscribe his newsletter on Substack.
Many Minds is a project of the Diverse Intelligences Summer Institute (DISI) (https://disi.org), which is made possible by a generous grant from the Templeton World Charity Foundation to UCLA. It is hosted and produced by Kensy Cooperrider, with help from assistant producer Cecilia Padilla. Creative support is provided by DISI Directors Erica Cartmill and Jacob Foster. Our artwork is by Ben Oldroyd (https://www.mayhilldesigns.co.uk/). Our transcripts are created by Sarah Dopierala (https://sarahdopierala.wordpress.com/).
You can subscribe to Many Minds on Apple, Stitcher, Spotify, Pocket Casts, Google Play, or wherever you like to listen to podcasts.
We welcome your comments, questions, and suggestions. Feel free to email us at: manymindspodcast@gmail.com.
For updates about the show, visit our website (https://disi.org/manyminds/), or follow us on Twitter: @ManyMindsPod.
The AI-powered Podcast Player
