What If the Body Is Optional?

Some half-formed thoughts on senses, simulations, and why the five-bucket model might be holding us back

There’s this thing that happens once you’ve spent enough time with cognitive science literature where you start to feel a little embarrassed about the five senses. Not embarrassed for them, precisely. More like when someone confidently explains how we only use ten percent of our brains and you just sort of nod and sip your drink.

We’ve been running with sight, sound, smell, taste, and touch since Aristotle, and nobody seems to really push back. But spend five minutes with proprioception, interoception, thermoception, nociception, or vestibular processing and the whole framework starts to feel like someone tried to organize a library using five shelves.

I’ve been bumping back into predictive processing literature lately, and I think there’s something genuinely interesting hiding in the cracks between neuroscience, extended reality(XR), and the weirder corners of altered-states research. I’m not sure I’ve figured it out yet. But I want to try to lay out a question I have in my mind.

We’re not cameras

The basic move in predictive processing: the brain isn’t passively receiving sensory data and assembling it into a faithful picture of “outside”. It’s running a best-guess model of the world and checking that model against incoming signals. When the prediction matches, nothing interesting happens. When it doesn’t, we get a prediction error, and the model updates.

Karl Friston’s free energy principle formalizes this. A recent paper in Neuroscience (February 2025) frames Bayesian brain theory as a computational account of how beliefs form and revise through predictive processing. Perception isn’t reception. It’s inference (sound familiar?). We’re not seeing the world. We’re guessing at it, very quickly, and mostly getting it right (I see you GPT).

This is where I start to wonder whether the five senses framework was even asking the right question. If the brain is a prediction engine, then the relevant categories aren’t “which sense organ does the signal come from?” They’re more like: what kind of prediction error is this? How confident is the model? What’s the update?

Different hardware, different worlds

The biologist Jakob von Uexküll had this concept called Umwelt, roughly “surrounding world,” but what he really meant was: the perceptual world that a given organism actually inhabits. A tick’s Umwelt is built from temperature gradients and butyric acid. A bat’s from echolocation. Ed Yong’s An Immense World (2022) walks through how radically different sensory apparatus produces radically different realities.

All these organisms are navigating the same physical environment. The same energy interactions, the same causal structure. They’re just running very different translations of it. And each translation is internally coherent. The bat doesn’t see or feel what we do but it also isn’t getting the world wrong. It’s getting a world built for it. A functional world.

If perception is largely a translation, how many valid translations are there? And could we, in principle, build a new one from scratch?

The body as default setting

Here’s where it starts to sound a little out there, so bear with me.

If we take predictive processing seriously, then the body itself is showing up in phenomenal space as part of the model. We experience having a body not because it’s the objective case but because it’s the best functional model available. And we experience that body as stable, as ours, as the center of our perceptual world. But that experience is itself a prediction. The brain’s best guess about the organism it’s embedded in, and that organisms relationship to the matter around it.

We already know this prediction can be disrupted. The classic rubber hand illusion tricks the brain into incorporating a fake hand into its body model in under two minutes. VR embodiment research (there’s a solid 2026 review in Frontiers in Virtual Reality) shows that people regulate personal space around a virtual avatar and react to threats approaching it as if they were real. The body model is flexible. It’s not hardwired but instead a hypothesis that the brain keeps confirming because sensory evidence keeps supporting it.

So what happens when we systematically change the sensory evidence?

Two ways to mess with the translation

I think there are two research directions converging at the moment that many of us aren’t taking seriously enough yet.

Neurotech. Brain-computer interfaces are moving fast. Neuralink expanded its human trials into the UK in 2025, Paradromics is working on high-bandwidth implantable systems for speech restoration and motor control. Right now, BCI research mostly focuses on restoring lost function. But the underlying capability is about creating new input/output channels between brains and the world. Route novel signals into the brain’s predictive machinery and we’re potentially giving it new data to model. New prediction errors. New inferences.

Extended reality. Not VR in the “realistic video game” sense, but VR (or XR) as a tool for systematically altering the relationship between self, body, and world. The embodiment research shows the brain will adopt a virtual body if the sensory contingencies are right. What if we pushed further? Instead of simulating the normal world with a different body, what about a fundamentally different perceptual environment with internally consistent rules that a brain could learn to navigate?

Combine these and we’re not just changing what an organism sees. We’re changing the terms of the translation.

The psychedelic data point (briefly)

I don’t want to oversell this. The psychedelic research space has a tendency toward breathlessness and I’m trying to resist that. But there’s something worth noting.

A systematic review on DMN modulation by psychedelics (International Journal of Neuropsychopharmacology, 2023) documents how classic psychedelics consistently reduce connectivity within the default mode network while increasing coupling between brain regions that don’t normally talk to each other. The subjective correlate, often called ego dissolution, involves the boundary between self and world becoming porous or disappearing entirely.

What’s interesting isn’t the experience itself but what it suggests about our phenomenological architecture. If the sense of being a bounded self in a body can be pharmacologically disrupted while we remain fully conscious and (to some degree) functional, that tells us something about how contingent the default configuration is. The body-centered self might not be a necessary feature of consciousness. But instead a setting. And settings can be changed.

The question I’m actually asking

Could we design a reliable, reproducible framework for generating alternative phenomenological configurations?

Not a hallucination. Not a distortion. But Ann internally coherent alternative translation of the same underlying physical reality, where the body might not be the central reference point, where sensory categories might be organized differently, where the self-world boundary might be extended, or drawn somewhere else entirely?

I don’t know the answer. I’m not sure anyone does yet beyond theorycraft. But the pieces seem to be landing on the table: predictive processing gives us the theoretical framework, umwelt research gives us the biological precedent, neurotech gives us the input channels, extended reality gives us the environmental scaffolding, and psychedelic research gives us proof of concept that the default configuration can shift.

I wonder if someone is going to put these together in the next few years. And I wonder what we’d find on the other side.

If you’re working on anything adjacent to this, or if you think I’m wrong about something, I’d love to hear from you. I’m also at the Foresight Berlin AI Node, where we support researchers at the intersection of AI, science, and safety. Rolling grant applications are open for funding, compute, or node access (coworking). Check out upcoming events too.

References

Bayesian brain theory paper (Bottemanne et al., Neuroscience, Feb 2025)

https://pubmed.ncbi.nlm.nih.gov/39643232/

Ed Yong, An Immense World (2022)

https://www.penguinrandomhouse.com/books/616914/an-immense-world-by-ed-yong/

VR embodiment review (Vasudha & Choudhary, Frontiers in Virtual Reality, Feb 2026)

https://www.frontiersin.org/journals/virtual-reality/articles/10.3389/frvir.2026.1756137/full

DMN modulation by psychedelics systematic review (Gattuso et al., International Journal of Neuropsychopharmacology, March 2023)

https://academic.oup.com/ijnp/article/26/3/155/6770039

Neuralink UK expansion / BCI landscape (Forbes, Oct 2025)

https://www.forbes.com/sites/robtoews/2025/10/05/these-are-the-startups-merging-your-brain-with-ai/

Paradromics FDA IDE approval (2025 Neurotech Review)

https://www.techlifesci.com/p/2025-neurotech-review