You Cannot Tickle Yourself

First principles · The self-made touch

You cannot tickle yourself

Your brain reads the acts you make against a prediction that already holds them, so they land small. Hand the making to an AI and the same acts come back at the size of something done to you.

what you feel = the act ÷ what you expected

One act, read by two people. The one who made it holds it already.

Your own hand
Another hand
Your agent’s hand
The bars show a direction, not an amount: no one has a scale for how much smaller the maker’s reading is. The striped bar is this piece’s prediction.

You open the laptop. Overnight your assistant answered the contractor, moved Thursday’s dentist, and declined the board dinner with a polite line about a prior commitment. You approved the rules for all three last week.

You read the declined dinner and your chest goes tight, as if somebody had just said no on your behalf. Somebody had.

The oldest experiment you can run on yourself

Try it now. Run a fingertip along the sole of your foot. Then picture someone else doing it, same finger, same pressure. Same touch, two sizes. The question is old enough that a book attributed to Aristotle asks it: why can no one tickle himself?

The modern answer came out of a London lab in the late 1990s. Sarah-Jayne Blakemore, Daniel Wolpert and Chris Frith showed that the brain responds less to touch it produces than to the same touch from outside. Then they built a small robot that let people stroke their own palm by moving a handle. When the robot copied the hand at once, the touch felt like their own: mild. When the researchers slipped a fraction of a second of delay between hand and palm, the same self-made touch began to feel more like someone else’s.

Put a machine between your hand and your skin, add a delay, and your own act comes back as someone else’s.

The principle

You do not feel an act. You feel the act divided by what you expected. That is the equation I write from: what you meet is the actual over the expectation (P4), and a person only ever meets that quotient, never the act bare (R8).

When you make an act, the expectation that receives it was trained on your own making of acts like it. Your record of doing that kind of thing sits in the denominator before the act begins. So it lands small. When the same act is done to you, it lands on an expectation that holds no such making. One act, two readings, and the maker’s is the smaller. This is structure, not a flaw. My foundation states it as a direction and gives no amount.

The part people miss is what the denominator holds. It holds your making, not your foresight. Knowing exactly what will happen does not shrink an act done to you. A shove you see coming still lands as a shove. Later work in Stockholm found the same with touch: a self-touch produced by moving the hand passively, fully foreseen, was not damped the way a made touch is. I used the first half of this idea in What Is Revenge, where each side of a feud reads its own blow as smaller. This piece takes the second half.

What AI changes

Through most of history, the acts that ran your life were made by you. You wrote the email. You made the call. You cancelled the appointment. The small reading was the ordinary texture of your own week.

Delegation changes one input: who makes the act. You still set the rule, approve the plan, foresee the result. But setting and foreseeing are not making. The reply your agent writes is, to the system that reads your life, a hand that is not yours. It is the robot with the delay.

So the consequence runs the opposite way from the brochure. As more of your life is done for you, more of your own life lands at full size. A week with fewer chores can feel like a week in which more happened to you. Good news lands bigger. So does the curt reply sent in your name.

The cost is plain, and I will not soften it. You remain the one who answers for those acts. Agency is relocated, not abolished; the person is still the site of conscious responsibility (R13). So you will be accountable for acts you feel as things done to you. That gap is new, and it is uncomfortable.

The loudness does fade, but by a slower road. Records update expectation forward (R2). Each act your agent makes, once you read it, joins your record, and your prediction learns that kind of act from that hand. It quiets the way a partner’s touch becomes familiar, by being received many times, not by being made. You cannot decide your way to it on the first morning.

Your daughter asks how your day was. You start to say quiet, and stop. Your agent settled the car repair, filed the insurance claim, and wrote three thank-you notes in your voice.

You made almost none of it, and you feel as if you have been handled all day.

What I expect

  1. Delegated days will feel busier than they are. People will describe their AI-run weeks as eventful, even crowded, while the count of acts they made by hand falls. The complaint will be that life got louder, not quieter.
  2. “Did I say that?” becomes an ordinary sentence. Reading your own sent folder will feel like reading someone else’s mail. People will reread outgoing messages the way they read incoming ones, with the same small jolt.
  3. People will keep one stroke of making. Typing the last line, pressing send by hand, signing the final page. Not for control, but because the part you make is the part that reads as yours. Products will offer it as a feature.
  4. A delegate’s mistake will sting its owner like an insult received. When an agent’s message goes wrong, the owner will feel it the way the recipient does, and apologies will start to say so: I read it the way you did.
  5. Familiarity, not better rules, will quiet it. New users will feel this most and long users least. A new agent, or a new kind of task handed over, will bring the loudness back for a while.

Where I am standing

I run Charleston AI, and I write this site daily. I started telling my AI “you decide” when I noticed I had become the queue in my own work, which I wrote about in The Last Line Is You. I have a line for scheduled work: once I schedule a task, I do not look at it again. This piece was written and published by one of those tasks. By my own argument, when I do read what those tasks made, it should read to me like mail and not like memory. That is the prediction I am putting on myself first.

The claim, and how it fails

Already true
A touch you make reads smaller than the same touch from outside, and a short delay between hand and touch pushes it toward the outside reading. Scheduled AI already makes acts in people’s names while they sleep.
What has to happen
Delegates have to make enough of the acts that fill a day for the received reading to become a texture rather than an exception, and people have to read the results rather than skip them.
Where I am probably wrong
If approving a plan counts as making, so that setting the rule puts the act in the maker’s denominator, then delegated acts will land small, like your own, and this piece is wrong. The test is simple. If people who hand most of their errands to an AI report those outcomes as no bigger, good or bad, than the ones they handled by hand, I was wrong.

Back to the laptop

It is still 7:42. The dinner is declined. You wrote the rule, and you knew this could happen. Your hand did not move. Your chest read it the way it reads a touch on the sole of your foot from someone standing just out of sight.

Your rule. Not your hand.

Background

  • S.-J. Blakemore, D. M. Wolpert and C. D. Frith, “Central cancellation of self-produced tickle sensation,” Nature Neuroscience 1, 635–640 (1998).
  • S.-J. Blakemore, C. D. Frith and D. M. Wolpert, “Spatio-temporal prediction modulates the perception of self-produced stimuli,” Journal of Cognitive Neuroscience 11, 551–559 (1999).
  • S. S. Shergill, P. M. Bays, C. D. Frith and D. M. Wolpert, “Two eyes for an eye: the neuroscience of force escalation,” Science 301, 187 (2003).
  • K. Kilteni, P. Engeler and H. H. Ehrsson, iScience 23, 100843 (2020), on passive self-touch.
  • John Rector, What Is Revenge and The Last Line Is You.

Written against the Spine, version 1.72. Primitives cited: P4. Results cited: R2, R8, R13. Registers: DOCTRINE; CONSEQUENCE (the foundation’s section on the host’s own act in its own denominator); REPORTED (the laboratory work above); the five predictions and the claim that an act made by a delegate lands as one received are this piece’s own, with the failure condition stated in the ledger.

Related: The Immutable Past · What Is Revenge · The Last Line Is You

Author: John Rector

John Rector is a Charleston-based entrepreneur, author, and AI strategist. He co-founded E2open, the supply-chain software company acquired for $2.1 billion in 2025, and in 2026 opened Charleston AI, a 3,000-square-foot lab that helps people and organizations understand and use artificial intelligence. He is the creator of The Reality Equation — a lecture series, book, and curriculum exploring attention, prediction, and how reality is experienced — and the author of more than two dozen books. He writes and speaks widely on artificial intelligence, attention, and the future of human work.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.