Anil Seth's argument: perception is not a window onto reality. It is a controlled hallucination — the brain's best guess, continuously updated. What that means for suffering, for stuck patterns, and for what contact actually is.
Anil Seth's phrase "controlled hallucination" is deliberately provocative, and it earns that provocation. His argument, developed across years of consciousness research at the University of Sussex, is that perception is not passive reception. The brain does not open a window onto the world and see what's there. It generates a model — a prediction — and then uses incoming sensory signals to correct and refine that model. What you experience as reality is the model, not the world. The world only enters as error signals, as the corrections the prediction needs. Ordinary perception and hallucination differ not in kind but in degree of control — in how tightly the model is constrained by incoming evidence. In both cases, you are experiencing the brain's construction. You are never experiencing the world directly.
Seth extends this to the experience of self. The sense that there is a "you" experiencing all of this — a continuous, bounded, coherent self — is itself a construction. The self is a prediction the brain makes about the organism it inhabits. A useful, stable, highly constrained prediction — but a prediction nonetheless. Not found in the world, not found in the brain, but generated by the brain's ongoing modeling activity. This has an unsettling implication: the self you experience as primary and given is as constructed as the chair you're sitting on. Both are the brain's best guess. Both are updated when prediction error arrives. And both can be wrong in ways that feel completely real.
A stuck pattern is not a failure of will. It is a highly confident prediction — one the brain has learned to hold with high precision because it has been accurate enough, often enough, that the system stopped updating it. Contact is what creates the conditions for that precision to drop.
Together, Seth and Friston complete the picture of why insight alone is insufficient for change. Insight updates the narrative layer — the DMN's story about the pattern. But the pattern itself is a prediction held at a level below the narrative, closer to the interoceptive signal, shaped by prior experience that predates language. Changing it requires something the brain registers as genuine prediction error at that level — an experience that is actually incompatible with what the simulation has been confidently generating. That is not an argument or a reframe. It is a felt experience, in the body, that something different is happening than what was predicted. That is what contact opens. That is where the simulation can revise.
Seth, A. (2021). Being You: A New Science of Consciousness. Dutton. — The full argument, highly readable.
Seth, A.K. & Friston, K.J. (2016). Active interoceptive inference and the emotional brain. Philosophical Transactions of the Royal Society B, 371(1708).