Your Brain Is Hallucinating Right Now

Right now, as you read these words, your brain is hallucinating.

Not metaphorically. Not in the way people say "reality is subjective" at dinner parties. Your brain is generating a model of what it expects to see, hear, and feel, and then projecting that model as your conscious experience. The actual sensory data coming in from your eyes and ears is secondary. It's a fact-checker, not the author.

Neuroscientist Anil Seth at the University of Sussex calls this a "controlled hallucination." Your entire perceptual experience is a construction. A best guess. It happens to be constrained by reality most of the time, which is why you don't walk into walls. But it is not reality itself.

This idea has a name: predictive processing. And it might be the closest thing we have to a unified theory of how the brain works.

The Wiring Tells the Story

The evidence starts with basic anatomy. In your visual cortex, there are roughly ten times more feedback connections running top-down (from higher brain areas to lower ones) than feedforward connections running bottom-up (from your retina to your cortex).

Think about what that means.

If your brain were simply receiving data from your eyes and building a picture, you'd expect the opposite ratio. Ten times more wires going up than coming down. Instead, the vast majority of connections are sending predictions downward. The smaller set of bottom-up connections is sending back only the errors. The mismatches. The surprises.

Murray Sherman and Ray Guillery at the University of Chicago demonstrated this in their 2006 work on thalamic relay circuits. The thalamus, long assumed to be a simple relay station passing sensory data to the cortex, turned out to be something more interesting. It's a comparison engine. It checks the brain's predictions against what's actually coming in from the senses and flags the discrepancies.

Your brain isn't processing reality. It's predicting reality and then updating the prediction when it's wrong.

The Face That Isn't There

There's a demonstration of this that's almost too clean.

Take a hollow face mask. Turn it around so you're looking at the concave inside. The nose recedes. The cheekbones curve inward. Every physical cue tells you this is not a normal face.

You see a normal face anyway.

Your brain's prediction that faces are convex is so powerful it overrides the raw visual information. Richard Gregory first described this hollow mask illusion in 1970, and it's been replicated countless times since. But the really interesting data came in 2009 when Danai Dima and colleagues used fMRI to watch what happens in the brain during the illusion. They found increased activity in face-processing regions. The brain was literally generating a face where none existed. Building the hallucination in real time.

Here's the twist. People with schizophrenia are significantly less susceptible to the illusion. They see the hollow mask for what it is. Dima's team published this finding in NeuroImage, and it aligns perfectly with the predictive processing framework. Schizophrenia may involve a disruption in the brain's prediction machinery. When the top-down predictions weaken, you see the raw data more clearly. But you also lose the ability to distinguish between what your brain generates and what's actually out there.

Seeing through the illusion isn't a superpower. It's a symptom.

The Framework

Andy Clark, a philosopher and cognitive scientist at the University of Edinburgh, laid this all out in his 2013 paper "Whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science" in Behavioral and Brain Sciences. It became one of the most cited papers in cognitive science of the decade.

Clark's argument: the brain is fundamentally an "organ of prediction." Its primary job is not to react to the world. Its primary job is to minimize surprise. It does this by constantly generating hypotheses about the causes of its sensory inputs, testing those hypotheses against incoming data, and updating the model when predictions fail.

Karl Friston at University College London formalized the math behind this in his 2005 paper on the free energy principle in Philosophical Transactions of the Royal Society B. The core idea is that the brain tries to minimize "prediction error" across every level of processing. From the raw sensory data hitting your retina all the way up to your abstract beliefs about how the world works.

Jakob Hohwy at Monash University (now at Copenhagen) called the brain a "prediction machine" in his 2013 book The Predictive Mind, arguing that this single principle could explain perception, action, attention, and even consciousness.

Three different researchers. Three different disciplines. Same conclusion. Your brain is not a camera. It's a prediction engine with a camera for error correction.

You've Experienced This

You already know what prediction errors feel like. You just didn't have the vocabulary.

That moment when you step off a curb that's higher than you expected and your stomach lurches. Your brain predicted a certain distance to the ground. Reality didn't match. The jolt you feel is a prediction error propagating through your system.

Or the gorilla experiment. Daniel Simons and Christopher Chabris ran their famous study at Harvard in 1999. Participants watched a video of people passing a basketball and were asked to count the passes. Halfway through, a person in a gorilla suit walked through the scene, faced the camera, beat their chest, and walked off. About half the participants didn't see the gorilla at all.

Not because their eyes didn't register it. Because their brain was predicting "basketball passes" and the gorilla didn't match the prediction model. The raw sensory data was there. The prediction filter threw it out.

This is inattentional blindness. And it's not a flaw. It's the system working as designed. Your brain can't process every piece of incoming data. It predicts what matters, attends to what's relevant, and discards the rest. Usually this works beautifully. Sometimes a gorilla walks through.

The Rubber Hand and the Boundaries of You

The predictions don't stop at perception. They extend to your sense of self.

Matthew Botvinick and Jonathan Cohen published a deceptively simple experiment in Nature in 1998. Place a rubber hand on a table in front of someone. Hide their real hand behind a screen. Stroke both the rubber hand and the real hand simultaneously with a paintbrush.

Within minutes, people report feeling the touch on the rubber hand. They flinch when the rubber hand is threatened. Their brain has incorporated a piece of plastic into its body model.

Your sense of where your body ends and the world begins is not fixed. It's a prediction. And that prediction can be updated with surprisingly little evidence.

Olaf Blanke and Thomas Metzinger extended this to full-body illusions in 2009, published in Trends in Cognitive Sciences. Using virtual reality and synchronized touch, they could make people feel like they were inhabiting a different body entirely. Or hovering outside their own body. The experience felt completely real because the prediction model said it was real.

Your sense of having a body. Of being located inside that body. Of that body being yours. All predictions. All updatable. All hallucinated.

Why This Matters

I'm a developer. I spend my days building systems that take inputs, process them, and produce outputs. When I first encountered predictive processing, it hit different because the brain isn't doing that. It's not input-process-output. It's predict-compare-update. That's a fundamentally different architecture.

And it explains so much.

Why placebos work. Why anxiety feels so real even when nothing is wrong. Why meditation changes perception. Why two people can look at the same situation and experience completely different realities. Why you sometimes see or hear things that aren't there. Why your emotions feel like they're happening to you rather than being generated by you.

All of it traces back to one principle. Your brain is a prediction machine. It builds a model of reality and projects that model as your experience. Sensory data is just the error signal.

Anil Seth puts it clearly in his 2021 book Being You: we don't perceive the world as it is. We perceive the world as it's useful for us to perceive it. The hallucination is optimized for survival, not accuracy.

Andy Clark's 2023 book The Experience Machine goes further. If all experience is prediction, then changing your predictions changes your experience. Not in some woo-woo manifestation way. In a measurable, neurological, this-shows-up-on-brain-scans way.

That's what this series is about. Your brain is a prediction machine. Once you understand how it works, you can start working with it instead of against it.

Starting with the fact that right now, everything you see is a controlled hallucination that happens to be mostly right.

Mostly.

Sources

• Whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science (Clark, 2013, Behavioral and Brain Sciences)
• A Theory of Cortical Responses (Friston, 2005, Philosophical Transactions of the Royal Society B)
• Being You: A New Science of Consciousness (Seth, 2021, Dutton/Faber & Faber)
• The Experience Machine: How Our Minds Predict and Shape Reality (Clark, 2023, Pantheon)
• The Predictive Mind (Hohwy, 2013, Oxford University Press)
• Understanding Why Patients with Schizophrenia Do Not Perceive the Hollow-Mask Illusion Using Dynamic Causal Modelling (Dima et al., 2009, NeuroImage)
• Gorillas in Our Midst: Sustained Inattentional Blindness for Dynamic Events (Simons & Chabris, 1999, Perception)
• Rubber Hands 'Feel' Touch That Eyes See (Botvinick & Cohen, 1998, Nature)
• Full-Body Illusions and Minimal Phenomenal Selfhood (Blanke & Metzinger, 2009, Trends in Cognitive Sciences)
• An Interoceptive Predictive Coding Model of Conscious Presence (Seth, Suzuki & Critchley, 2012, Frontiers in Psychology)

---

Part of the Prediction Machine series. Next: Your Brain Learns by Being Wrong.