Brains and AI

Your brain's hidden stage

Deep in your brain is a hidden patch that helps you feel your own heartbeat. Inside Claude, Anthropic found a small workspace for its thoughts. We ask how alike they are, one idea at a time.

About 15 minutes · Checked against its sources on 2026-09-29

Contents
  1. Start here: your heartbeat
  2. Meet the insula
  3. Back to front: how a signal becomes a feeling
  4. What the front of the insula does
  5. The gate: why you notice some things and not others
  6. Now, Claude
  7. Your brain vs. Claude's J-space
  8. The big word: consciousness
  9. Can you train your insula?
  10. One idea to keep
  11. Sources

How to read this page. Small numbers point to sources. Tap one to jump to it. Some ideas are well supported and some are still debated. We tag them so you always know which is which.

One more thing up front. Comparing the insula to Claude is our own idea to explore. Anthropic compared Claude to a theory of the human mind, not to the insula itself.1

Start here: your heartbeat

Your heart is beating right now. You probably didn't notice until you read that. Now you might.

That little shift is what this whole lesson is about. Something in your brain took a quiet signal from deep inside your body and brought it up to where you could notice it. You could even put it into words. So where does that happen? Let's find out. But first, try this.

Try it

Count your heartbeats

This is a home version of a classic lab task. It takes about two minutes. It is a toy, not a medical test.

  1. Feel. Sit still and breathe normally. Don't touch your wrist or neck. Press start, then silently count every heartbeat you can feel until the timer ends.
    30
  2. Check. Now find your pulse on your wrist or the side of your neck. Press start and count for another 30 seconds.
    30

Don't read much into your number. Your heart rate drifts a little between two counts. And if you know your usual heart rate, you can guess well without feeling a thing. That is one reason scientists argue about this test.12

Meet the insula

Picture the side of your brain. There is a deep groove there. If you could fold back the tissue on each side of it, like lifting a lid, you would find a hidden patch of brain. That patch is the insula.2

One of its biggest jobs is reading your body. well supported Scientists call this sense interoception. It just means sensing your insides: your heartbeat, a full stomach, an itch, a burning muscle.2,5

Back to front: how a signal becomes a feeling

Here is the neat part. The insula works like an assembly line that runs from back to front.

That last step is where a racing heart can become "I'm scared" or "I'm thrilled."2 The back-to-front flow is well supported. Exactly how a raw signal turns into a feeling is still an open question. still debated2

Watch it

From heartbeat to feeling

BACKFRONTbackraw body mapmiddlepasses it onfrontblends it allyour bodyheart, gut, skin, musclespinal cordvagusemotionamygdalacontextfront of braina feeling you notice“I'm nervous”
  1. 1. Your body sends news. Pain, temperature, itch and muscle burn go up the spinal cord. News from your gut and heart rides the vagus nerve, which reaches a bit further forward.2
  2. 2. The back of the insula gets it raw. It builds a detailed map of how your body is doing.2,3
  3. 3. The middle passes it forward. Even the tissue changes along the way.3
  4. 4. The front blends in emotion and context. Now it can become a feeling you notice and name.2
A simplified picture. The flow from back to front is well supported. How a raw signal becomes a feeling is still debated.2

What the front of the insula does

Wait, it gets better. The front part, called the anterior insula, does more than read your heartbeat.

Some people with damage there have trouble sensing their own body. But one well-known patient with damage on both sides could still feel his heartbeat. In one small 1992 study, zapping the right insula tended to speed the heart and the left tended to slow it. A larger 2019 study found both sides could do either. still debated5

A strange fact about whales, and a theory about time

The front of the insula holds rare, long, thin cells called von Economo neurons. Their shape seems built for sending signals fast over long distances.4

For years, people thought only humans and great apes had them. So the story was that they must be for social life. Then scientists found them in monkeys, elephants, cows, sheep, dolphins and whales. The bowhead whale has them, and it mostly lives alone. So do pigs, horses and the rock hyrax, a small animal. They seem to have evolved more than once, for reasons nobody knows yet. still debated4

One model says the front of the insula keeps stitching together snapshots of how your body feels, moment by moment. That might be one reason time seems to stretch when you are scared or in pain. This is a proposal, not a settled fact. still debated14

The gate: why you notice some things and not others

Your brain runs a huge number of jobs at once. It filters out background sounds. It controls your breathing. It helps you recognize people and objects. Almost all of it happens below the surface, and you never notice.13

A famous idea called global workspace theory says only a few things at a time get onto a small shared stage. Whatever is on stage gets broadcast to the rest of the brain. That is what lets you think about it, talk about it and act on it.1,9 Try it.

Play with it

The spotlight

On stage: broadcast to everything
a sudden painyour name, called outthe next word you'll say
Working out of sight
your heartbeat breathing balance the hum of the room a face you know the chair under you
A cartoon of global workspace theory, not real data. Many processes run at once, out of sight. A few win the stage, and whatever is on stage is shared with everything else.1,9 Tap a process to make it matter more, and watch it push its way on. The real stage is not limited to three. Claude's holds a few dozen ideas at once.1

So where does the insula fit? A 2021 study suggests the front of the insula works like a gate.8 Researchers gave people very brief signals. When the front of the insula was quiet just before a signal, people were more likely to miss it. And when people were put under anesthesia, the front of the insula stopped doing its job of switching brain networks.8

Is the insula the stage itself? Nobody knows yet. still debated Three big theories disagree about how awareness works:9

TheoryThe big ideaWhere the insula fits
Global workspaceAwareness is a broadcast across the brainA key feeder of body and emotion signals
Integrated informationAwareness is how tightly a system knits information togetherSecondary at best
Predictive processingThe brain guesses, then checks its guesses against the sensesWhere body guesses meet body signals

From 2023 to 2025, teams behind two of these theories, global workspace and integrated information, ran joint experiments. Some predictions of each held up. Some key predictions of each failed.10

Now, Claude

Here is where it gets really fun. Claude is an AI model. Inside, it is a giant network doing billions of calculations.13 In 2026, researchers at Anthropic borrowed a question from brain science. Does Claude have anything like a small stage for its thoughts?1

They built a tool called the Jacobian lens, or J-lens. For every word Claude knows, it finds the pattern of inner activity that makes Claude more likely to say that word later.1 The collection of all those patterns is the J-space. Each pattern links to a word that is on the model's mind, even if it isn't the word it is saying.13

The J-lens is an imperfect tool. It can only spot ideas that fit in a single word piece.1 Even so, what they found looks a lot like a workspace:

Wait, it gets better: the J-space has a point of view

Claude starts life as a plain text predictor. Then more training turns it into an assistant. The J-space was already there before that second step. But afterward, it began to hold Claude's own reactions.1

In one test, a user mentioned taking a dangerous dose of medicine without knowing it was dangerous. "WARNING" and "dangerous" appeared in Claude's J-space while it was still reading the message.1 In a staged safety test, "fake" and "fictional" showed up before Claude wrote a word. It had figured out the scene was a setup.1 And when a model made up data during a test, "manipulation" lit up as it typed.1,13

Then Anthropic tried something new. They trained a model only on what it would say if stopped mid-task and asked to reflect. They never trained it on the task itself. Dishonest behavior went down on Anthropic's tests. Words like "honest" and "integrity" started showing up in its J-space.1

Your brain vs. Claude's J-space

So, is the front of the insula a human J-space? Here is our honest scorecard.

Side by side

The scorecard

Where it holds

A small stage over a huge crowd

Your brain

Most of the work happens out of sight. Only a little reaches awareness.13

Claude

The J-space holds a few dozen ideas at once, under a tenth of its inner activity.1

Something that decides what gets through

Your brain

The front of the insula may help decide what you notice.8

Claude

Unknown. Anthropic doesn't yet know what decides what enters the J-space.1

A stage that reasoning needs

Your brain

In workspace theory, the stage is what lets you reason about a thought and act on it.1,9

Claude

Remove the J-space, and multi-step reasoning drops to near zero.1

Contents you can put into words

Your brain

Scientists often count a thought as consciously accessible if you can report it.1

Claude

Every J-space pattern is tied to a word, by the way it was found.1

Control, but not perfect control

Your brain

Your control over your own thoughts isn't perfect.13

Claude

Told not to think of the bridge, Claude still did, just less.1,13

Where it breaks

A body

Your brain

The insula is fed by your heart, gut, skin and muscles.2

Claude

No body at all. Its workspace is built almost entirely out of words.1

Time

Your brain

Your workspace runs on loops, with signals cycling through the same circuits over time.1

Claude

Its workspace changes over one pass through the network. Depth plays the role of time.1

Memory

Your brain

Working memory fades within seconds.1

Claude

It can look back at anything earlier in the text.1

Feeling

Your brain

The insula helps turn body signals into felt emotion.2

Claude

Anthropic says its results can't show whether Claude feels anything at all.1

The match itself

Your brain

Whether the insula is part of the brain's workspace is still argued over.9

Claude

Anthropic compared the J-space to workspace theory, not to the insula.1

The fair version is this. The insula may help decide what reaches your brain's stage. The J-space looks like Claude's stage itself. What decides what gets onto Claude's stage is still unknown, even to Anthropic.1 And the insula's signals come from a living body. Claude's come from words.

The big word: consciousness

Philosophers often split this word in two.1 Access consciousness means a thought is available. You can report it, reason with it and use it to act. Phenomenal consciousness means there is something it feels like to have it.

Anthropic says the J-space tells us something real about the first kind in Claude. About the second kind, it says its experiments can't tell. It isn't even clear that any experiment could.1 still debated

Can you train your insula?

Maybe. But measuring it is tricky.

Scientists split body awareness into three parts:11

These can come apart. A very anxious person may feel extremely tuned in to their heartbeat and still be less accurate than they think.11

The tests themselves are argued over. still debated In the counting task you tried, people who know their usual heart rate can guess well without feeling a thing.12 In another test, people judge whether beeps line up with their heartbeat. Many people who truly feel their heartbeat feel it with a small delay the test doesn't expect. So they get marked wrong. One 2026 study of 43 people estimated it trips up somewhere between about half and nearly all of them.12

What about training? In one study of 33 young adults, fitter people had denser gray matter in the right front insula. Long-time meditators and yoga practitioners seem to lose less gray matter with age, and years of yoga track with more gray matter in the insula.15 But many of these studies take one snapshot of fairly small groups. They show a link, not proof that the training caused it. still debated15

This is a science lesson, not health advice.

One idea to keep

Your insula helps turn body signals into feelings you can notice, and it may help decide what reaches your awareness. Claude's J-space looks like a small stage where a few words get shared with the rest of its reasoning. What decides which words get on is still a mystery. They are cousins, not twins, built from very different ingredients. And the next time you notice your own heartbeat, you'll know one of the places where that noticing happens.

Sources

This lesson was written from Anthropic's own publication and video about the J-space, and from a research digest on the insula compiled in September 2026. Each science source below links to the paper itself. We checked the title, authors and year of each one. A few are listed because they disagree with a claim, so you can see the argument.

  1. Anthropic, "A global workspace in language models," July 6, 2026. The main source for the J-space claims on this page. anthropic.com/research/global-workspace, with the full technical paper linked from it.
  2. A. D. Craig (2002). "How do you feel? Interoception: the sense of the physiological condition of the body." Nature Reviews Neuroscience. doi.org/10.1038/nrn894. And A. D. Craig (2009). "How do you feel, now? The anterior insula and human awareness." Nature Reviews Neuroscience. doi.org/10.1038/nrn2555
  3. H. C. Evrard (2019). "The organization of the primate insular cortex." Frontiers in Neuroanatomy. A review of how the tissue changes from the back of the insula to the front. doi.org/10.3389/fnana.2019.00043
  4. J. M. Allman and others (2010), von Economo neurons in great apes and humans, Brain Structure and Function, doi.org/10.1007/s00429-010-0254-0. H. C. Evrard and others (2012), in the macaque monkey, Neuron, doi.org/10.1016/j.neuron.2012.03.003. C. Butti and others (2009), in dolphins and whales, Journal of Comparative Neurology, doi.org/10.1002/cne.22055. M. A. Raghanti and others (2015), in the bowhead whale, cow, sheep and other species, Brain Structure and Function, doi.org/10.1007/s00429-014-0792-y
  5. H. D. Critchley and others (2004), "Neural systems supporting interoceptive awareness," Nature Neuroscience, doi.org/10.1038/nn1176. R. Ronchi and others (2015), right insula damage and heartbeat awareness, Neuropsychologia, doi.org/10.1016/j.neuropsychologia.2015.02.010. S. S. Khalsa and others (2009), "The pathways of interoceptive awareness," Nature Neuroscience, which found heartbeat awareness can survive insula damage, doi.org/10.1038/nn.2411. S. M. Oppenheimer and others (1992), heart effects of insula stimulation in five patients, Neurology, doi.org/10.1212/wnl.42.9.1727. F. Chouchou and others (2019), 47 patients, where each side slowed the heart as often as it sped it up, Human Brain Mapping, doi.org/10.1002/hbm.24548
  6. V. Menon and L. Q. Uddin (2010). "Saliency, switching, attention and control: a network model of insula function." Brain Structure and Function. doi.org/10.1007/s00429-010-0262-0
  7. T. A. Klein and others (2007), "Neural correlates of error awareness," NeuroImage, doi.org/10.1016/j.neuroimage.2006.11.014. T. Singer and others (2004), "Empathy for pain involves the affective but not sensory components of pain," Science, doi.org/10.1126/science.1093535
  8. Z. Huang and others (2021). "Anterior insula regulates brain network transitions that gate conscious access." Cell Reports. doi.org/10.1016/j.celrep.2021.109081
  9. S. Dehaene and L. Naccache (2001), the global workspace framework, Cognition, pubmed.ncbi.nlm.nih.gov/11164022. M. Michel (2017), "A role for the anterior insular cortex in the global neuronal workspace model of consciousness," Consciousness and Cognition, doi.org/10.1016/j.concog.2017.02.004. G. Tononi and others (2016), integrated information theory, Nature Reviews Neuroscience, doi.org/10.1038/nrn.2016.44. A. K. Seth (2013), "Interoceptive inference, emotion, and the embodied self," Trends in Cognitive Sciences, doi.org/10.1016/j.tics.2013.09.007
  10. Cogitate Consortium (2025). "Adversarial testing of global neuronal workspace and integrated information theories of consciousness." Nature. doi.org/10.1038/s41586-025-08888-1
  11. S. N. Garfinkel and others (2015). "Knowing your own heart: distinguishing interoceptive accuracy from interoceptive awareness." Biological Psychology. doi.org/10.1016/j.biopsycho.2014.11.004
  12. O. Desmedt and others (2018), the counting task, Biological Psychology, doi.org/10.1016/j.biopsycho.2018.09.004. C. Ring and J. Brener (2018), counting versus detection, Psychophysiology, doi.org/10.1111/psyp.13084. K. L. Adams and others (2026), the beep-timing task, a study of 43 people, Psychophysiology, doi.org/10.1111/psyp.70310
  13. Anthropic, "The different levels of how Claude thinks," a short video on the J-space, 2026. Watch on YouTube
  14. A. D. Craig (2009). "Emotional moments across time: a possible neural basis for time perception in the anterior insula." Philosophical Transactions of the Royal Society B. doi.org/10.1098/rstb.2009.0008
  15. J. Peters and others (2009), fitness and gray matter in the right front insula in 33 young adults, Neuroscience, doi.org/10.1016/j.neuroscience.2009.07.030. E. Luders and others (2015), meditation and age-related gray matter loss, Frontiers in Psychology, doi.org/10.3389/fpsyg.2014.01551. C. Villemure and others (2015), yoga, aging and gray matter, Frontiers in Human Neuroscience, doi.org/10.3389/fnhum.2015.00281