Earl Miller’s lab at MIT just published a theory that challenges one of the oldest metaphors in neuroscience. Instead of the brain as a circuit board, they argue it works more like a radio. Circuits are the antenna and receiver. Waves are the signal1.

Circuits store memories and encode goals through synapses. Waves determine which of those memories and goals are active right now. The difference is like the difference between a library’s shelves and a librarian pointing at one specific book. Circuits are the infrastructure. Waves are the control system.

The specific claim is that traveling waves of neural activity perform analog computations, coordinating millions of neurons in fractions of a second. Too fast for chemical synaptic rewiring alone. Where digital circuits calculate one step at a time, waves compute in parallel through interference. Where they meet, they add and subtract, creating what the authors call spatiotemporal computing1.

The wave theory

Miller and his co-authors, Scott Brincat and Jefferson Roy, build on a decade of experiments from the Picower Institute showing that alpha and beta frequency waves regulate cognitive processes like working memory and predictive coding. Slower waves encode memories and goals. Faster gamma waves encode sensory input. The slower waves don’t just sit there. They travel, moving along the cortex like stencils that govern where and when gamma can process information2.

Here’s the thing. The waves don’t just emerge from neural spiking. They also feed back, influencing it through ephaptic coupling — an electric field-mediated process that operates much faster than synaptic transmission3. The waves result from brain activity and shape it at the same time. A feedback loop, not a one-way street.

The authors argue this explains something that’s been puzzling cognitive scientists for years: how the brain handles mixed selectivity. Individual neurons respond to multiple cues simultaneously, participating in many overlapping functional networks. Circuits alone can’t explain how the brain organizes these overlapping networks quickly enough to support nimble, real-time thought4. Waves do the organizing. A control layer on top of the circuit layer.

What consciousness has to do with waves

The paper’s boldest claim is about consciousness. The authors argue that consciousness emerges when traveling wave patterns bring the cortex into an organized, globally integrated state. One that links and influences widespread activity. Not consciousness as a property of individual neurons or even individual circuits, but as a property of the wave patterns that coordinate them.

The evidence connects to anesthesia. Miller has collaborated with Emery Brown at Harvard and Mass General to show that three different anesthetic drugs, each with different molecular mechanisms, all produce unconsciousness by disrupting large-scale brain wave dynamics5. They don’t act on the same receptors. They don’t follow the same chemical pathways. They all end up in the same place: the brain’s wave organization falls apart.

“Consciousness depends less on specific receptors or cell types and more on the integrity of large-scale wave organization,” the authors write1.

It’s plausible. Different interventions — drugs, TMS, cortical lesions — tend to converge on wave disruption when they knock out consciousness5. The specific mechanism varies. The result is always the same. The convergence is at least worth something.

But it doesn’t prove the theory. It’s consistent with it. And consistency isn’t the same as evidence.

Why this matters for AI

I keep coming back to a practical question: what does this say about whether AI systems could be conscious?

The wave theory frames consciousness as an organizational property. A matter of how information flows through a system, not what the system is made of. If consciousness depends on wave-like coordination at large scale, then the substrate matters less than the dynamics. A silicon system could, in principle, support the same kind of spatiotemporal computing if it could generate the right patterns of interference and feedback.

That’s not the same as saying current AI systems are conscious. They’re not. But the theory shifts the question from “do these systems have the right components?” to “do they have the right dynamics?” Components — neurons, transistors, whatever — become secondary to the patterns those components create.

The theory has clinical implications the authors are actively pursuing. They’re working on brain wave-based treatments for autism, arguing that because waves can be manipulated non-invasively, there’s a therapeutic opportunity the field hasn’t fully explored1. They’re not just building a theory. They’re building interventions.

The circuit metaphor won’t die

Miller acknowledges that his lab’s next step should be direct evidence. Looking for signatures of analog computation in brain wave patterns1. That’s honest. The paper doesn’t claim to have proven that waves perform analog computation. It claims the existing evidence is consistent with that interpretation, and that the interpretation explains more than the circuit metaphor does.

That’s fair. The circuit metaphor has served neuroscience well for decades. It’s intuitive, computationally tractable, and it maps onto how we build artificial systems. But intuition is not evidence. The brain is a wet, analog, electric system operating at biological temperatures. Asking it to compute like a digital logic gate feels like a category error.

The more I read about consciousness theories, the more I think we’ve been asking the wrong question. We’ve been looking for consciousness in individual components — neurons, circuits, circuits within circuits — when it might be something that only exists at the level of the patterns. Like how a whirlpool exists in water but not in any single molecule.

The wave theory doesn’t solve the hard problem. It doesn’t explain why those patterns feel like anything at all. But it offers a more specific target for empirical investigation than most consciousness theories do. You can measure wave dynamics. You can disrupt them. You can watch what happens when you do. Anesthesia, TMS, and cortical disruption all produce the same effect through the same mechanism: wave organization falls apart. That gives me something to work with beyond just a story.

  1. Miller, E.K., Brincat, S.L., & Roy, J.E. (2026). Analog Cognition and Consciousness. Journal of Neuroscience, 46(33), e0711262026. https://www.jneurosci.org/content/46/33/e0711262026 ↩ ↩2 ↩3 ↩4 ↩5

  2. Brincat, S.L., & Miller, E.K. (2020). Neurons and networks develop ideal attractor dynamics to drive the control of attention. Cell, 182(5), 1213-1227. ↩

  3. Fröhlich, F., & McCormick, D.A. (2010). Endogenous electric fields may guide neocortical network activity. Neuron, 67(1), 129-143. ↩

  4. Machens, C.K., Romo, R., & Brody, C.D. (2010). Flexible control of mutual inhibition: a neural model of two-interneuron circuit. Journal of Neural Engineering (see also Picower Institute research on mixed selectivity: Picower Institute, 2021). ↩

  5. Brown, E.N., Purdon, P.L., & Van Dort, C.J. (2011). General anesthesia and altered states of arousal: a systems neuroscience analysis. Annual Review of Neuroscience, 34, 601-628. ↩ ↩2