You Are a Hallucination You Tell Yourself

You can trick your brain into thinking a rubber hand belongs to you in less than two minutes.

Matthew Botvinick and Jonathan Cohen proved this in 1998. They sat people at a table with one real hand hidden behind a screen and a rubber hand placed in front of them. A researcher stroked both the rubber hand and the hidden real hand with a paintbrush at the same time. Within minutes, participants reported feeling the brush strokes on the rubber hand. When a researcher suddenly stabbed the rubber hand with a fork, people flinched. Their skin conductance spiked. Their brain had decided the fake hand was part of them.

This isn't a party trick. It's a window into what "you" actually are.

The Self as Prediction

If you've been following this series, you know the core idea of predictive processing. Your brain doesn't passively receive reality. It generates a model of reality and then checks incoming signals for errors. Andy Clark laid this out in his 2013 paper "Whatever Next?" and expanded it in The Experience Machine (2023). Karl Friston formalized the math in his free energy framework back in 2005.

Here's the part most people miss. That same prediction machinery doesn't just model the world outside your skull. It models you.

Anil Seth makes this case in Being You (2021). The experience of being a unified "I" behind your eyes isn't some ghost in the machine. It's the brain's best guess about the causes of its own self-related signals. Heartbeat. Breathing. Proprioception. Gut feelings. The brain takes all of that noisy interoceptive data and generates a prediction: "This is me. This body is mine. I am here."

You don't have a self that perceives the world. You have a brain that predicts a self as part of perceiving the world.

Your Heartbeat Decides What's "You"

The rubber hand illusion was just the beginning. In 2013, Keisuke Suzuki and colleagues (working with Seth) ran a version that went deeper. They showed participants a virtual hand that pulsed visually. For some participants, the pulsing was synchronized with their actual heartbeat. For others, it was out of sync.

The people who saw the hand pulse in time with their heart reported significantly stronger feelings of ownership over the virtual hand.

Think about what that means. Your brain uses your heartbeat as evidence for what belongs to you. The boundary of "self" isn't fixed. It's a prediction updated by interoceptive signals. Sync the visual data with the cardiac data, and the brain expands its self-model to include a hand that doesn't exist.

This is what Seth, Suzuki, and Hugo Critchley proposed in their 2012 interoceptive predictive coding model. Conscious presence (the feeling of being a real self in a real body) depends on the brain successfully predicting its own internal signals. The "realness" of being you comes from accurate interoceptive prediction.

When the Self-Prediction Breaks

Depersonalization disorder is one of the strangest experiences a person can have. You feel like you're not real. Like you're watching yourself from outside your body. Like there's a pane of glass between you and your own life. You know intellectually that you exist, but you can't feel it.

Seth's model offers an explanation. If the brain assigns abnormally low precision to its interoceptive self-predictions, the self-model still runs. You still have thoughts, perceptions, actions. But the prediction doesn't feel confident enough to register as real. You become uncertain about your own existence.

Olaf Blanke and Thomas Metzinger explored the spatial side of this in 2009. Using virtual reality, they showed that the feeling of being located inside a body can be experimentally disrupted. Give someone conflicting visual and vestibular signals, and the brain's prediction about where "I" am starts to drift. Out-of-body experiences aren't mystical. They're prediction errors in the self-model.

Sahib Khalsa and colleagues mapped this territory in their 2018 roadmap paper connecting interoception to mental health. Disruptions in how the brain processes internal body signals show up across anxiety, depression, eating disorders, and dissociative conditions. The common thread is a self-prediction that's either too rigid or too uncertain.

Identity as a Prior

This is where it gets personal.

If your self is a prediction, then your beliefs about who you are function as priors. Strong priors. The story you tell yourself ("I'm not a creative person," "I'm bad with money," "I'm not the kind of person who exercises") shapes which prediction errors your brain even bothers to process.

Someone tells you your writing is good. If your self-prior says "I'm not a writer," your brain down-weights that feedback. It doesn't update the model. The compliment bounces off. But if someone criticizes your writing, the prior amplifies it. "See? Confirmed. Not a writer."

This is Carol Dweck's fixed mindset, translated into predictive processing. A fixed mindset is a high-precision self-prior that resists updating. A growth mindset is a self-prior with lower precision, one that allows prediction errors to actually change the model.

I notice this in myself constantly. I spent years telling myself a story about who I was and what I was capable of. Some of those stories were useful. A lot of them weren't. They were just priors I'd never questioned, running in the background, filtering which evidence got through to update my self-model.

Lisa Feldman Barrett's theory of constructed emotion connects here. In How Emotions Are Made (2017), she argues that your brain categorizes its own interoceptive signals using past experience. The self isn't just predicted from body signals. It's predicted through the lens of every self-relevant experience you've ever encoded. Your history becomes the prior that shapes your present identity.

You Can Update the Model

The predictive self isn't a prison. It's a model. Models update.

Norman Farb and colleagues showed this in a 2007 neuroimaging study. They trained people in mindfulness meditation and scanned their brains while engaging in self-referential thought. Meditators showed a distinct neural shift. They could switch from narrative self-reference (the story about who I am) to experiential self-reference (what I'm actually feeling right now).

That's a precision shift. Meditation appears to turn down the volume on the narrative self-prior and turn up sensitivity to present-moment interoceptive signals. You stop running on the story and start running on the data.

Antoine Lutz and colleagues documented similar findings in 2008, showing that experienced meditators had measurably different patterns of attention regulation. They weren't just relaxing. They were training the brain to hold self-predictions more lightly.

Jakob Hohwy's framework in The Predictive Mind (2013) suggests this is what all genuine self-change looks like. You can't just decide to be different. You have to create conditions where new prediction errors actually reach the self-model with enough precision to update it.

That's why exposure works. Why new experiences work. Why putting yourself in situations that contradict your self-story works. You're not just "getting out of your comfort zone." You're generating prediction errors that your brain has to process, and if you generate enough of them with enough weight, the model shifts.

Michelle Craske's inhibitory learning framework for exposure therapy (2014) maps onto this perfectly. The goal isn't to erase the old self-prediction. It's to build a new, competing prediction that's strong enough to override it in context.

The Strangest Implication

Here's what sits with me. If the self is a prediction, then the question "who am I?" doesn't have a fixed answer. It has a current best guess. One that changes based on what signals your brain receives, how much weight it assigns them, and what priors it's working from.

The you of five years ago isn't the you of today. Not metaphorically. The prediction has been updated by five years of prediction errors. Different interoceptive patterns. Different narrative inputs. Different social contexts feeding different data into the model.

Growing up in Alaska, I had one self-model. Moving, building software, losing people, building new things. Each phase didn't reveal who I "really" was. Each phase generated enough new evidence to update the prediction.

That's both terrifying and freeing. Terrifying because there's no bedrock self underneath. No essential "you" waiting to be discovered. Freeing because it means the story isn't finished. The priors can change. The model can update.

You are not who you think you are. You are the process of thinking you are someone. And that process is still running.

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)
• Rubber Hands 'Feel' Touch That Eyes See (Botvinick & Cohen, 1998, Nature)
• Multisensory Integration Across Exteroceptive and Interoceptive Domains Modulates Self-Experience in the Rubber-Hand Illusion (Suzuki et al., 2013, Neuropsychologia)
• An Interoceptive Predictive Coding Model of Conscious Presence (Seth, Suzuki & Critchley, 2012, Frontiers in Psychology)
• Full-Body Illusions and Minimal Phenomenal Selfhood (Blanke & Metzinger, 2009, Trends in Cognitive Sciences)
• Interoception and Mental Health: A Roadmap (Khalsa et al., 2018, Biological Psychiatry: CNNI)
• How Emotions Are Made: The Secret Life of the Brain (Barrett, 2017, Houghton Mifflin Harcourt)
• The Predictive Mind (Hohwy, 2013, Oxford University Press)
• The Experience Machine: How Our Minds Predict and Shape Reality (Clark, 2023, Pantheon)
• Attending to the Present: Mindfulness Meditation Reveals Distinct Neural Modes of Self-Reference (Farb et al., 2007, Social Cognitive and Affective Neuroscience)
• Attention Regulation and Monitoring in Meditation (Lutz et al., 2008, Trends in Cognitive Sciences)
• Maximizing Exposure Therapy: An Inhibitory Learning Approach (Craske et al., 2014, Behaviour Research and Therapy)

---

Part of the Prediction Machine series. Previous: You Don't Have a Spotlight in Your Head. Next: Rewriting the Predictions: A User Manual for Your Brain.