Brain Interfaces, Brain Copying, and 'Sleeves'

The Last Boundary

Every technology in this book extends human capability: AI extends cognition, robotics extends labor, medicine extends life. But all of them, in their current form, leave the fundamental architecture of humanity intact. You are still a biological being with a brain made of neurons, experiencing life through a body that was born, will age, and will die.

This chapter explores what happens when that architecture itself becomes modifiable.

Brain-computer interfaces let the mind reach beyond the body—controlling external devices, communicating without words, perceiving beyond biological senses. As interfaces become more sophisticated, the boundary between mind and machine begins to blur.

Brain copying—the idea that the patterns constituting your mind could be captured, stored, or run on different substrates—pushes further. If what makes you "you" is information, could that information be preserved, duplicated, or transferred?

"Sleeves"—a term borrowed from science fiction—refers to bodies other than your original biological one: synthetic bodies, cloned bodies, robotic bodies, or virtual environments. If identity becomes portable, bodies become interchangeable.

These possibilities range from already happening (basic BCIs exist) to speculative (full brain copying remains theoretical). But the trajectory points toward deeper integration of minds and machines, and eventually, perhaps, the decoupling of identity from any particular biological substrate.

This is the last boundary. Everything else accelerates human capability while leaving human nature fixed. This transforms human nature itself.

2026 Snapshot — Where Brain Interfaces Stand

Brain-computer interfaces are real, deployed, and improving rapidly.

Current BCI Technology

Non-invasive BCIs read brain activity through the skull:

EEG (electroencephalography) detects electrical activity via electrodes on the scalp. Resolution is low, but the approach is safe and accessible. Consumer EEG devices are used for gaming, meditation, and neurofeedback.
fNIRS (functional near-infrared spectroscopy) measures blood oxygenation changes in the cortex. Also non-invasive but limited in depth and resolution.
These systems can detect gross mental states (attention, relaxation) and, with training, enable slow control of external devices (spelling systems for people with severe disabilities).

Invasive BCIs place electrodes inside the skull for higher-quality signals:

Electrocorticography (ECoG) places electrodes on the brain's surface. Higher resolution than EEG, requires surgery but doesn't penetrate brain tissue.
Intracortical arrays insert electrodes into brain tissue itself. The Utah Array (Blackrock Neurotech) has been used in research for decades; newer approaches from Neuralink and others aim for higher electrode counts.
These systems can decode individual neuron firing, enabling control of robotic arms, computer cursors, and recently, speech reconstruction from intended words.¹

Current clinical applications:

Paralyzed patients controlling computer cursors and robotic arms through thought
Patients with locked-in syndrome spelling out messages letter by letter
Early experiments in speech decoding from neural signals
Deep brain stimulation for Parkinson's disease, epilepsy, and depression (not strictly BCI but involves brain implants)

State of the art: Neuralink's N1 device has over 1,000 electrodes and has been implanted in at least one human patient.² Synchron's Stentrode is deployed via blood vessels without open brain surgery. Both are in early human trials.

The Gap Between Science Fiction and Reality

Popular culture imagines BCIs as enabling instant telepathy, direct memory download, and seamless merger of human and artificial intelligence. Reality is more modest:

Bandwidth is limited: Even the best current BCIs read from thousands of neurons in a brain with 86 billion. This is like trying to understand a city by listening to a few thousand phone calls.
Output is easier than input: Decoding intended movements or speech from neural activity works reasonably well. Sending information into the brain is harder—science does not understand the "code" well enough to write meaningful messages.
Stability is challenging: Brain tissue reacts to implanted devices; signal quality degrades over time. Long-term implants must maintain function for years or decades.
Individual variation: Every brain is wired differently. BCIs must be calibrated to individuals and adapt as brains change.

The honest assessment: BCIs are real and improving, but the field is far from the science fiction vision of seamless brain-computer merger.

Notable Players

BCI Hardware

Neuralink has attracted attention through Elon Musk's profile and ambitious goals. Their approach uses a surgical robot to implant thin, flexible electrodes with high channel counts. Early human trials are underway for paralysis patients.

Synchron offers a less invasive approach using a stent-like device inserted through blood vessels. Lower risk surgery trades off against lower resolution signals. Also in human trials.

Blackrock Neurotech makes the Utah Array, the most widely used research BCI. They're developing next-generation systems for clinical applications.

Precision Neuroscience (founded by a Neuralink co-founder) is developing a cortical surface array that doesn't penetrate brain tissue.

Paradromics is building high-bandwidth BCIs focused on speech restoration.

CTRL-labs (acquired by Meta) worked on wristband-based neural interfaces reading signals from arm muscles—not directly brain but part of the broader interface ecosystem.

Research Institutions

University of Pittsburgh, Stanford, Caltech, Brown University, and others have pioneered BCI research for decades, developing the foundational science that companies are now commercializing.

DARPA has funded substantial BCI research through programs like Neural Engineering System Design (NESD), pushing toward higher-resolution interfaces.

The Wyss Center in Switzerland focuses on brain-computer interfaces for people with severe paralysis.

AI and Neural Decoding

BCI performance increasingly depends on AI. Decoding neural signals into intended movements or words is a machine learning problem. The major AI labs (DeepMind, Meta AI, OpenAI) have published research on neural decoding, though not all are commercializing hardware.

BCI Progression: The Next Decade

The trajectory of BCI technology follows predictable paths of improvement in resolution, coverage, safety, and usability.

Higher Bandwidth

Current BCIs read from thousands or tens of thousands of electrodes. The brain has billions of neurons. Closing this gap requires:

More electrodes in smaller form factors
Better signal processing to extract more information from fewer electrodes
Novel recording modalities (optical, chemical, magnetic) that might scale differently

Neuralink's roadmap envisions devices with millions of electrodes—still far short of comprehensive coverage, but orders of magnitude beyond current capability.

AI enhancement: Machine learning can infer more from limited signals. Better decoders extract more information from the same electrode count. This is a software problem that improves on AI timelines, not just BCI hardware timelines.

Trajectory: Near-term likely for continued increases in electrode count and decoding quality. Plausible for devices that can decode complex thoughts, not just motor intentions, within a decade. Full high-bandwidth coverage of the entire brain remains distant.

Bidirectional Interfaces

Current BCIs mostly read from the brain. Writing to the brain is harder.

Existing input approaches:

Deep brain stimulation delivers electrical pulses that modulate activity in specific regions—crude but effective for some conditions
Cochlear implants deliver sound-coded electrical stimulation to the auditory nerve
Retinal implants (discussed in Chapter 5) stimulate visual pathways

Challenges for sophisticated input:

Science does not understand the neural code well enough to write meaningful messages
The brain is not designed to receive external input in its internal language
Risk of unintended effects—stimulation in the wrong pattern could be confusing, distressing, or dangerous

Research directions:

Optogenetics (using light to control neurons) enables more precise stimulation but requires genetic modification
Ultrasound can focus on specific brain regions without surgery
Closed-loop systems that read neural states and adjust stimulation accordingly

Trajectory: Plausible for improved therapeutic stimulation and crude sensory input. Direct communication of complex information to the brain remains challenging and likely longer-term.

Wireless and Minimally Invasive

Current invasive BCIs require surgery and often wired connections. Wireless systems are essential for practical use.

Current progress:

Neuralink's N1 is fully wireless
Synchron's device is inserted without open brain surgery
Research systems are increasingly untethered

The goal: Devices that can be implanted safely in outpatient procedures, transmit data wirelessly, and last for years without maintenance.

Trajectory: Near-term likely for wireless functionality. Minimally invasive procedures are progressing but depend on specific technology approaches.

Consumer and Enhancement Applications

BCIs today are medical devices for severe disability. But the technology could eventually extend to enhancement:

Potential consumer applications:

Direct control of devices without hands (useful for AR/VR, gaming, or simply convenience)
Enhanced memory through external aids tightly integrated with natural memory
Augmented cognition—faster information access, computational assistance
Communication without speech or typing

Barriers to consumer BCI:

Safety: any brain surgery carries risk; the benefit must justify it
Long-term effects: it is unknown what happens when healthy brains have implants for decades
Regulation: medical devices are regulated; consumer enhancement products face different (and uncertain) oversight
Social acceptance: how many people would undergo brain surgery for convenience?

Trajectory: Wild in the near term for widespread consumer adoption of invasive BCIs. Plausible for non-invasive or minimally invasive enhancement interfaces with limited capability. The line between medical and consumer will blur as disability becomes more of a spectrum.

Toward Brain Copying: The Prerequisites

"Brain copying" or "mind uploading" refers to the hypothetical possibility of capturing the information that constitutes a mind and running it on a different substrate—a computer, a new biological brain, or something else.

This is far more speculative than BCI technology. No one has copied a mind. It is unknown whether this is even possible. But understanding why people take the idea seriously requires understanding what would be required.

The Information Hypothesis

The theoretical foundation for brain copying is the view that minds are fundamentally information-processing systems. On this view:

What makes you "you" is the pattern of connections and activity in your brain
This pattern is, in principle, describable as information
Information can be copied without loss
Therefore, if the pattern could be read and replicated, the mind could be replicated

This is a philosophical claim, not a scientific fact. It may be right, wrong, or incoherent. But it's the assumption underlying brain copying discussions.

What Would Be Required

Complete structural mapping: This would require mapping every neuron and every connection (synapse) in the brain. The human brain has approximately 86 billion neurons and 100 trillion synapses.³ Current connectomics technology (electron microscopy of preserved brain tissue) has mapped a cubic millimeter of mouse brain, revealing about 100,000 neurons and nearly a billion synapses.⁴ Scaling to the full human brain is a challenge of at least eight orders of magnitude.

Dynamic state capture: Neurons aren't just connected; they have internal states—electrical potentials, protein levels, gene expression. A static map of connections might not capture what's needed. The state of every neuron at the moment of capture might be required.

Sufficient resolution: It is unknown what level of detail matters. Connections and states? Molecular configurations? Quantum states? Different answers imply different scanning requirements.

Computational substrate: Simulating a brain in software would require enormous computational resources. Estimates range from 10^18 to 10^25 operations per second, depending on assumptions about what level of detail must be simulated.⁵ This is beyond current capability but perhaps achievable with continued hardware progress.

Validation: How would you know if a brain copy was accurate? It would need to think, remember, and behave like the original. But validating this is philosophically and practically fraught.

Current State

Nothing close to full brain copying is currently possible.

Related work:

Connectomics is mapping small brain regions at synaptic resolution
AI is creating increasingly sophisticated models of neural function
Brain organoids provide simplified systems for studying brain development
"Digital twins" are being developed for simpler organ systems

Trajectory: Wild for any near-term brain copying. The technology required is not on a visible timeline. This doesn't mean it's impossible—just that no one can estimate when or if it might happen.

The "Grain": Memory Capture and Lifelogging

Short of full brain copying, more modest technologies could capture and externally store aspects of mental life.

Memory Capture

The idea: record sensory experience continuously, creating an external archive that supplements biological memory.

Current technology:

Wearable cameras (GoPro, smartphone) can record video continuously
Voice recorders capture conversations
Location tracking logs where you've been
Fitness trackers record physiological states
Social media archives your posts, messages, and interactions

Advanced vision (as imagined in science fiction like Black Mirror's "The Entire History of You"):

Continuous recording from the perspective of the user
Instant indexing and retrieval—search your memories like a hard drive
Playback that recreates the experience with full sensory fidelity
Sharing of memories with others

What would be required:

High-quality continuous recording (camera, microphone, perhaps other sensors)
Massive storage for a lifetime of high-definition recording
AI to index, search, and summarize recorded material
Interfaces that make retrieval natural and fast
Eventually, sensory playback (requiring input BCIs or advanced AR/VR)

Current trajectory: Some elements exist (life-logging, searchable photo archives, voice assistants that remember past conversations). Integration into a seamless "memory prosthetic" is plausible within a decade, though limited by social and privacy constraints more than technology.

Cognitive Digital Twins

A step beyond memory: models that predict how you would think, respond, or decide.

The concept: An AI trained on your data—your writing, your decisions, your preferences—that can simulate your cognitive style. Not a copy of your brain, but a model of your behavior.

Current precursors:

Email auto-reply suggestions that mimic your tone
Recommendation systems that predict your preferences
AI assistants that learn your communication style

Advanced applications:

An AI that could draft emails in your voice while you sleep
A simulation that could answer questions as you would
A "backup" that could continue your projects if you were incapacitated
A persistent representation that outlives your biological self

This is not brain copying: The model doesn't experience anything; it doesn't have your memories in detail; it's not conscious (as far as anyone knows). It's a sophisticated imitation, not a continuation.

Trajectory: Near-term likely for increasingly sophisticated personal AI models. Plausible for models that convincingly imitate individuals in limited domains. Whether this constitutes any form of "cognitive continuity" is a philosophical question with no agreed answer.

"Sleeves": Bodies Beyond Biology

The term "sleeve" comes from Richard K. Morgan's science fiction, where consciousness can be transferred between bodies (biological or synthetic). While full consciousness transfer remains speculative, the concept of alternative bodies is becoming technologically relevant.

Robotic and Telepresence Bodies

Telepresence robots already exist: wheeled screens that let remote users move through and interact with physical spaces. These are crude "sleeves"—your consciousness remains in your biological body, but you perceive and act through the robot.

Advanced telepresence would include:

Humanoid robots with dexterous manipulation
Full sensory feedback (vision, hearing, touch, proprioception)
Low-latency control that feels natural
VR integration for immersive experience

BCI-controlled telepresence would be the next step: rather than using joysticks or gestures, you control the remote body through thought, as naturally as you control your biological body.

Applications:

Working in hazardous environments without physical risk
"Being present" at distant locations
Continuing to interact with the physical world if your biological body is impaired

Trajectory: Near-term likely for improved telepresence robots. Plausible for BCI-controlled telepresence with high dexterity within a decade. The experience of "embodiment" through a remote body is an active research area.

Synthetic Bodies

If BCIs become sophisticated enough to provide full sensory input and motor output, the interface between mind and body becomes more flexible.

Theoretical progression:

BCI controls external devices (current state)
BCI controls a remote robotic body with sensory feedback
Biological body becomes optional for interaction with the world
Eventually, mind could "inhabit" a fully synthetic body if biological body fails

Challenges:

Sensory input to the brain at the quality of biological senses
Motor output that provides the nuance and speed of biological movement
Proprioception and body sense—feeling the synthetic body as "yours"
Long-term stability and maintenance
Psychological adaptation to radically different embodiment

Trajectory: Wild in the near term, but the research directions are visible. Each step (better BCIs, better robots, better sensory feedback) is advancing.

Digital Existence

The most radical "sleeve" is no body at all—existence in a purely digital environment.

Requirements:

Full brain copying or emulation (highly speculative)
Computational substrate to run the emulated mind
Virtual environment that provides meaningful experience
Interface with the physical world if desired

Philosophical questions abound:

Would a digital copy be conscious?
Would it be you, or a copy that merely resembles you?
What rights would it have?
What would subjective experience be like without a body?

These are not currently answerable. They may become relevant within the timeframe of this book, or they may remain science fiction indefinitely.

Second-Order Impacts

The technologies in this chapter, if they mature, would transform human experience more fundamentally than anything else in this book.

Identity and Continuity

The current concept of personal identity assumes continuity of body and brain. A person is the same individual they were yesterday because body and brain are continuous with yesterday's body and brain.

BCIs begin to blur this:

If part of your cognition happens in external hardware, is that hardware part of "you"?
If your memories are stored externally, is your identity bound to that storage?
If you can copy your cognitive patterns, are the copies "you"?

There may be no single correct answer. Different cultures, legal systems, and individuals may resolve these questions differently.

Death and Mortality

If some form of cognitive continuity beyond biological death becomes possible—even imperfect continuity through detailed memory archives and cognitive models—the meaning of death changes.

This could be:

Comforting (your influence, your thoughts, your "voice" persist)
Disturbing (the line between living and dead becomes unclear)
Practically important (estate law, insurance, and social structures assume death is final)

Relationships and Communication

BCIs enabling direct brain-to-brain communication would transform relationships:

Intimacy could become literal—sharing experiences and feelings directly
Privacy would require new boundaries—what can you share, what can you keep private?
Deception becomes harder (or easier, depending on the technology)
Neurotypical/neurodivergent distinctions might shift as cognitive differences become more visible or bridgeable

Power and Control

Technologies that interface with minds are technologies that could potentially control minds:

Who has access to your neural data?
Could your BCI be hacked?
Could governments or corporations mandate BCIs for certain populations?
Could BCIs be used for surveillance, manipulation, or coercion?

The power dynamics of brain interface technology are profound. Governance frameworks that don't yet exist will be needed.

Risks and Guardrails

Safety

Brain surgery carries risk. Infections, bleeding, and device malfunction can cause permanent harm or death. As BCIs move from last-resort medical devices to broader applications, the risk-benefit calculus changes.

Minimum standards needed:

Long-term safety data before broad deployment
Clear informed consent about unknown risks
Mechanisms for device removal or deactivation if problems occur
Post-market surveillance for emerging issues

Security

Any device connected to your brain is a potential attack vector. Compromised BCIs could:

Leak private thoughts
Cause harm through malicious stimulation
Be used for surveillance
Be held for ransom

BCI security must be treated with at least the seriousness of medical device security—probably more, given the stakes.

Inequality

If BCIs provide cognitive enhancement, unequal access creates cognitive inequality:

Enhanced vs. unenhanced
Those who can afford and choose enhancement vs. those who cannot or do not
New forms of discrimination based on cognitive configuration

As BCIs become more valuable, pressure to adopt them could emerge:

Employers preferring enhanced workers
Insurance companies offering discounts for monitored brains
Governments mandating BCIs for certain populations (prisoners, soldiers, citizens)
Social pressure to participate in shared neural experiences

Protecting the right to remain unenhanced—and unmonitored—becomes important.

Existential Questions

If brain copying becomes possible, existential questions become practical:

Do copies have rights?
Can you "delete" a copy?
If the original and copy diverge, which is "you"?
What legal standing do purely digital persons have?

Legal and ethical frameworks for these questions don't exist. They'll need to be developed—ideally before the technology that creates the need.

The Path Forward

Near-term likely (5-7 years):

BCIs enabling natural speech from thought for paralysis patients
Expanded clinical applications for epilepsy, depression, and other neurological conditions
Improved wireless, minimally invasive devices entering trials
AI-enhanced neural decoding significantly improving performance
Early consumer EEG/neural interface devices with limited but real functionality

Plausible (7-15 years):

High-bandwidth BCIs with thousands of electrodes becoming routine for severe disability
Bidirectional BCIs providing basic sensory input alongside motor output
Brain-controlled telepresence with rich sensory feedback
Cognitive enhancement applications for healthy individuals in limited domains
Detailed cognitive models that convincingly simulate individual personalities

Wild (speculative):

Full brain copying preserving memories, personality, and (allegedly) consciousness
Direct brain-to-brain communication with meaningful content
Transfer of consciousness to synthetic bodies or digital environments
Merger of biological and artificial cognition into hybrid systems
Resolution of philosophical questions about consciousness and identity through empirical advances

The trajectory is clear in direction if not in endpoint. Minds and machines are becoming more closely integrated. The boundaries of the self are becoming negotiable. Whether this leads to transcendence or disaster—or, more likely, a complicated mix of both—depends on choices not yet made.

This is the last boundary. What lies beyond it is genuinely unknown.

Endnotes — Chapter 6

Speech decoding from neural signals has been demonstrated in research settings by groups at Stanford, UCSF, and elsewhere, with accuracy sufficient for communication. See Chang lab publications on speech neuroprosthetics.
Neuralink announced its first human implant in January 2024. The participant, Noland Arbaugh, demonstrated control of computer interfaces using the device.
Estimates of neuron and synapse counts vary but commonly cite approximately 86 billion neurons and 100 trillion synapses in the human brain. See Azevedo et al. (2009) "Equal numbers of neuronal and nonneuronal cells make the human brain an isometrically scaled-up primate brain."
The MICrONS project (Machine Intelligence from Cortical Networks) mapped approximately 1 cubic millimeter of mouse visual cortex at synaptic resolution, revealing about 75,000 neurons and 530 million synapses.
Estimates of computational requirements for brain simulation vary enormously depending on assumptions. Sandberg and Bostrom's "Whole Brain Emulation: A Roadmap" provides various estimates based on different levels of simulation fidelity.