[Reading Jeff Hawkins, A Thousand Brains, Basic Books, 2021]
Rene Descartes had an uneasy relationship to academics. He was very well educated, but he never held any academic positions and spent much of his life arguing with professors and theologians. He saw himself as a scientist, like Galileo, disclosing the secrets of the universe, and it just so happened that in his view those secrets were not best expressed in Aristotelian or scholastic terms. He seriously considered providing a detailed, point-by-point commentary and criticism on an academic treatise of the day (Francisco Suarez’s Metaphysical Disputations), but then apparently decided his time would be better spent just forging ahead and making his own path, trampling over carefully-articulated distinctions drawn by pointy-headed academics.
I get something of the same vibe from Jeff Hawkins’ book, A Thousand Brains. Hawkins was on his way to becoming an academic brain scientist, but at the time the academics were busy asking narrower questions he didn’t find as interesting, so he went his own way, made millions of dollars, and started his own research institute so he could study what he wanted. Unlike Descartes, I think his team is not fundamentally changing the paradigm for neuroscience, and their work is being published in respected journals; but like Descartes, he is marching ahead at a pace that academics might consider brazen. Not my field, so I don’t know. I can say that his knowledge of philosophy is shallow, but that doesn’t seem to hamper his research at all.
I will also add that while I am fascinated by the first half of his book, the second half is not so great. Hawkins offers his own view of humanity’s future, and he just doesn’t have much that’s interesting to say. So my advice is to skip that and just read through chapter 9.
Okay, on to the book. The crucial players in Hawkins’ story are the neocortex, cortical columns, and frames of reference. The neocortex is a thin sheet of networked neurons, about the size of a dinner napkin and about 3 mm thick. It wraps around the rest of your brain in crinkled fashion. It’s the part of the brain responsible for all of our fancy, distinctly human sorts of thought. If you zoom in on it, you will find about 150,000 cortical columns that span the thickness of the napkin. Those cortical columns have similar structures. So the neocortex seems to be made of 150,000 similar circuits, like just so many Raspberry Pis, though each one involves something like 100,000 neurons. They end up performing different functions because, basically, they are plugged into different parts of our bodies and different sets of other columns.
Cortical columns combine together to provide what Hawkins calls frames of reference. A frame of reference can be thought of as an algorithm that shapes and structures the information it receives, or works to sort out or combine information, or shapes and structures some activity or behavior on our part. Hawkins’ favorite example is our knowledge of a coffee cup. We know how to recognize one by sight; we know how it should feel in our hands, and what each finger should sense as we grip the handle or trace the edge of the rim; we know how much it should weigh in our hands, and so on. Our knowledge of coffee cups—and of the world at large—is built from these frames of reference, or our projections and expectations placed upon our experience. Again, the cortical columns all have pretty much the same structure; so the difference in the information processing between the smell of coffee and the feel of the coffee cup has more to do with cortical columns being plugged into your nose or your hands, rather than to do with the inner processing that takes place.
(Just an expression of wonder here that 150,000 cortical columns is enough to do the trick—not just for coffee cups, but all of what we think we know and experience! Can you exhaustively reduce your entire Weltanshauung into 150,000 components? It seems like there should be more, though I’ve been making a list and it ends at 12, so maybe so.)
All these processors plugged into one another provide us with a set of maps or models of what is in the world, where we are in relation to it, what we are doing, and what further actions are available to us. It is highly decentralized: there is no “headquarters” where “it all comes together” (in Daniel Dennett’s endearing terms), but instead there are tens of thousands of processors, all biting off bits of the problem and chewing their way through it and sending their results for other processors to use. In the end, we do and say stuff, and the eventual narration of what we do and say becomes the basis for beliefs about who “we” are and what “we” think “we” are up to.
Another crucial part of Hawkins’ account is the significance of motion. The neocortex does not simply store a bunch of pictures and sounds like a library. It tracks changes and movements and patterns in motion. Our model of the coffee cup is not just a bunch of slides, but clips portraying how it should change as we move in relation to it, or as we move it around through space. Motion and change are what we model in the world, and it is also how we think about ideas and concepts:
“If all knowledge is stored this way, then what we commonly call thinking is actually moving through a space, through a reference frame. Your current thought, the thing that is in your head at any moment, is determined by the current location in the reference frame. As the location changes, the items stored at each location are recalled one at a time. Our thoughts are continually changing, but they are not random. What we think next depends on which direction we mentally move through a reference frame, in the same way that what we see next in a town depends on which direction we move from our current location.”(Hawkins, Jeff. A Thousand Brains (p. 80). Basic Books. Kindle Edition.)
This is an exciting idea. We might know of memory palaces and the spatial hacks people use to remember long strings of information, but this appears to be evidence for the claim that our thinking is always movement through a landscape of ideas. Thinking is not a disembodied experience, but is something more like a virtual reality in which our projected motions take us from one concept to another, from one idea to a similar one, from one metaphor to its neighbors, and so on.
A commendable feature of Hawkins’ book is that he provides many clear analogies and illustrations of how his claims are borne out in both ordinary experiences and experimental ones. I have long been a fan of Dennett’s “pandemonium” model of the mind, and Hawkins’ book seems to be spelling out the details of how the congress of demons are actually embodied in the neocortex. The trick in any explanation of consciousness or intelligence is to show how a bunch of simple systems that aren’t so conscious or intelligent can sum up into a system that is conscious or intelligent (at least on our good days). There will always be philosophers who cry that in principle no such explanation can be achieved, but these are the folks who will be increasingly ignored as the results come in. (True in Descartes’s physics, true in Hawkins’ neuroscience.)
I do think that at some point our understanding of ourselves will have to shift from individuals to communities. That is to say, many of my cortical columns are plugged into not just each other and parts of my body, but to other people. Not directly of course (ew). But through language, culture, social dynamics, etc: all that squishy stuff gets reference framed in my brain, and exerts strong pulls over who and what I say I am and what the world is supposed to be. Neither Hawkins nor Dennett go far in this direction, but that’s okay, since we need both approaches, from individuals outward and from societies inward. But in the end, I think what we call “consciousness” and “intelligence” is going to be explained much more by communities and traditions than it is by cortical columns.