Some neuroscientists are confused

This summer, I wrote a series of stories about the Thousand Brains Theory of Intelligence which postulates that the neocortex is subdivided into hundreds of thousands of cortical columns which each build models of the world and work together to retrieve knowledge by voting. If you missed these essays, I highly encourage you to check them out.

The Thousand Brains theory is still unproven and, while backed by some empirical evidence, is largely supported by logical deduction and mathematical models—not observational data from brains.

However, I saw a story from The Atlantic a few weeks ago that I think actually connects to what we’ve been exploring here on Synapse and I wanted to share it with you.

The story is about how neuroscientists Carl Schoonover and Andrew Fink have confirmed a rather puzzling phenomenon in the mouse brain called representational drift. Essentially, the research team recorded how specific neurons in a mouse’s brain respond to a smell sensation over the course of several weeks. They were surprised to find that the neurons that responded to the smell actually changed overtime—the neurons that responded on day 1 would not be the same ones to respond a few weeks later.

This is surprising given that the textbook definition of how the brain works involves our neurons “encoding” information about the world in their various connections with other neurons as they respond to different stimuli. If the neurons that respond to the world are constantly changing, how does the brain keep anything straight?

Well, if we think about what the Thousand Brain theory says, this could actually make sense. First, it’s not surprising that multiple areas of the brain are capable of understanding the same stimuli given that the Thousand Brains theory postulates a universal algorithm of intelligence that is used by every column in the neocortex.

Second, the Thousand Brains theory says that the brain can recognize stimuli (like smells) by combining the votes of individual columns. Jeff Hawkins describes this voting by having you imaging you are reaching into a black box for a coffee mug. Each of your fingers are touching a certain part of the coffee cup with incomplete information. One finger might be touching the rim and thinks this could be a mug or a cup. One finger could be touching the side and think that this could be any round object. One finger could be touching