In Alan Turing’s “Computing Machinery and Intelligence” he writes in 6(9) The Argument from Extrasensory Perception that

I assume that the reader is familiar with the idea of extrasensory perception, and the meaning of the four items of it, viz., telepathy, clairvoyance, precognition and psychokinesis. These disturbing phenomena seem to deny all our usual scientific ideas. How we should like to discredit them! Unfortunately the statistical evidence, at least for telepathy, is overwhelming. It is very difficult to rearrange one's ideas so as to fit these new facts in. Once one has accepted them it does not seem a very big step to believe in ghosts and bogies. The idea that our bodies move simply according to the known laws of physics, together with some others not yet discovered but somewhat similar, would be one of the first to go.

Based on Turing’s remarks, I suspect the task of telepathy is not a computable process, that is, a task one could expect a Turing machine to perform.

Although it seems intuitively obvious that they are not computable tasks, how would one argue that telepathy, or other ESP tasks, are not computable?

Edit: There is a psi phenomenon I am particularly interested in. In Dean Radin's "Selected Psi Research Publications" there is one he co-authored called "Consciousness and the double-slit interference pattern: Six experiments". Human subjects, but not computers, were able through their intentionality to affect a double slit experiment.

How would one formulate that task as an algorithm to run on an AI machine? If that can be formulated it looks like it would serve as a modified Turing test without involving human judges to determine if the simulation convinced them. If that can't be formulated, then strong AI does not explain human mental behavior.

A video of a talk by Radin, "New Experiments Show Consciousness Affects Matter", summaries the above and similar results.

  • 3
    I think there is terminological confusion. These are not computable tasks because they are not tasks at all in the sense of the theory of computation. Neither is ordinary perception or doing back flips, Turing machines can not do those either, computation is a procedure for transforming specified inputs into specified outputs. Not everything is a computation, and not everything that violates the laws of physics (as we know them) is uncomputable.
    – Conifold
    Jan 2 '18 at 22:54
  • @Conifold I couldn't find anything in the article you cited that would claim that doing a backflip is not computable. There is input: a command to move in some way. There is a state change: configure to perform the backflip. There is output: perform the backflip. It seems like something a finite state machine could perform let alone a Turing machine. However, I will look for better definitions of this. It is precisely the kind of objections you are raising that interests me. Thanks! Jan 3 '18 at 0:58
  • 2
    Computation consists of transcribing symbols into symbols, any physical action or sensory perception are already beyond the word's usefulness. What you describe concerns simulating a backflip, not the action itself, and there is of course no problem with simulating ESP either, as movies do.
    – Conifold
    Jan 3 '18 at 1:39
  • 1
    @Conifold The problem with the traditional and alternative Turing tests is that they are like magic shows. If the magician convinces you that he pulled a rabbit out of a hat then the claim is that he really pulled the rabbit out of the hat. Such results are hard to take seriously. I've edited my question to focus on a particular experiment using a double-slit experiment. Jan 3 '18 at 3:37

Although it seems intuitively obvious that they are not computable tasks, how would one argue that telepathy, or other ESP tasks, are not computable?

One would have to argue that there exists no test that can measure the phenomenon of telepathy and, further, no such test could possibly exist. In other words, not only has such a measurement never been made (it just might not have been found yet), but it is logically impossible that it could ever be made. This response assumes the laws of physics as currently known. However, if ever it were ever discovered that telepathy follows a discrete set of laws, then once the process became clear, a telepathic Turing machine would become theoretically possible.

  • Measuring psi phenomena in humans has been documented. See Dean Radin's "Selected Psi Research Publications": deanradin.com/evidence/evidence.htm So Turing's concern is justified. As you mention there could be a "discrete set of laws" explaining it that we currently are unaware of. I will edit my question to include a specific psi phenomenon I am particularly interested in examining. Jan 3 '18 at 2:52

Communications usually happens via energy, or vibration. Somewhere along the EM spectrum. There's RF, but X-rays are also in that spectrum. Telepathy would be a form of communication.

If telepathy exists, might close proximity telepathy differ from telepathy between two minds separated by a great distance?

If we're considering the former, some EM energy put out by the brain, could explain it. If a brain's thoughts are associated with that specifically-patterned EM, then receiving that signal and interpreting it would be within the range of known physics.

Kind of like the physics of how two radios work, as a transmitter and receiver. (Also well within known laws of physics!)

Maybe it's spread-spectrum "signal?" The brain puts out a LOT of signal!

As for telepathy over distance, that's got to be a different phenomenon.

In any case, if it's the case that telepathy is linked with EM emissions, that is certainly computable, as is EM and energy science, generally.

  • Neutrinos, entanglement & Einstein-Rosen bridges, dark matter, higher dimensions, & many other candidates exist. Whereas, electromagnetic radiation could be easily detected, if it were involved.
    – CriglCragl
    Sep 24 at 15:57
  • “Physical” and “computable” are not equivalent. Some physics is not computable, particularly underdetermined physics such as quantum phenomena. While ESP can be characterized functionally and is therefore as compatible with being computed as any function.
    – Dcleve
    Sep 25 at 16:53

An excellent reference would be Peter Hasker’s The Shadow of Consciousness. He discusses the efforts in recent decades to understand both the easy and hard problems of consciousness both algorithmically and neurologically. Not everything can be computed. And Hasker details how this issue has bedeviled even the “easy” problem.

Not the lack of computability includes physics not just consciousness. Basically anything that is underdetermined is not computable. Quantum phenomena are underdetermined. So is any macro scale object which is chaotic in behavior, which has been found to be the case for a surprisingly large range of macro objects. For underdetermined physics, any effort to model it will have to be statistical.

You also presume that physics itself is closed. It cannot be, if science is non reductive, and the majority view of scientists and philosophers of science is that science is pluralistic. IR other sciences achieve useful insights about our world that physics cannot ever explain. Also science as a whole cannot be closed unless scientism is true (that all that matters in the world is subject to science), and the vast majority of current thinkers hold that it is untrue, and necessarily so.

So, to your experiment:
No, it is likely not computable. From a purely physicalist POV this for multiple reasons. First because the physics being used in the experiment was underdetermined quantum phenomena. Second the experiment treated mind as determining the undetermined physics, which violates physicalist assumptions. Third the physicalist models of mind, either emergent, or identity theory to neural processing, presume phenomena that are themselves likely underdetermined (emergence is likely a spontaneous statistical phenomena, and our recursive neural nets look likely to be chaotic). Finally, the details of consciousness, and emergence, are not understood in physicalist terms and this lack of explicit specificity makes computation impossible.

If one takes a non material perspective then the logic problems in the experiment can go away, but the lack of clarity of terms like intentionality, consciousness and emergence would remain.


You assume that “a Turing machine” is a Turing machine and nothing else. But when I was a little kid, I had a toy that could do backflips, and today someone could easily integrate a powerful computer into it, making it a Turing machine that can do backflips.

I assume when you talked about “turing machine” you really meant an artificial intelligence. If we could build what would be a genuine artificial intelligence, and assuming that telepathy, if it exists, must follow the laws of physics (maybe laws we don’t know yet), there’s no reason why a telepathic AI should be impossible. And doing backflips - no problem.


A Turing machine (in the strict sense) only takes input from its tape and so cannot interface to the real world in any other way. Turing-complete machines, or in the real world almost-Turing-complete machines such as conventional computers, can take input from other sources. If a computer had an interface that could read thoughts at a distance then it could be telepathic certainly, but of course once we understand how that process works, the magic is gone.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.