7
$\begingroup$

I come from a computer science background and am a big time AI enthusiast. I am trying to think about how can we map 'intuition' to a process in artificial intelligence. As I understand, 'intuition' means generating (correct) knowledge (for ourselves) about something without knowing 'how' we did it. This leads to the following questions:

  • Is 'intuition' related to 'extrasensory perception' (ESP)? As far as I know, ESP is not accepted by the scientific community but I could not conclude the same about 'intuition'.
  • If they are related, how does scientific community explain 'intuition'?

I seek pointers to relevant literature which would help me understand intuition from the perspective of cognitive sciences. My thought process has inclined me towards believing that there is nothing like 'intuition' and everything is deterministic which is why an AI can never be intuitive.

$\endgroup$
  • $\begingroup$ I think correct and useful Intuitions have a lot to do with experience. I don't think esp explains it. Anecdotal evidence often points to medical personal intuitively doing the right thing, most likely based on experience. $\endgroup$ – AliceD Jan 24 '15 at 10:02
  • $\begingroup$ I disagree: I think the OP is just confused about some terminology, and any reference to pseudo-science can be easily cleared up with a good answer. The original question implies that determinism allows ESP but not intuition. A good answer, providing clear definitions of these terms, should demonstrate that the opposite is true: Determinism allows intuition but not ESP. Perhaps this question could be answered equally well on philosophy.SE, but as the focus is on the meaning of "intuition", I think it belongs here. $\endgroup$ – Arnon Weinberg Mar 20 '15 at 18:30
  • $\begingroup$ @ArnonWeinberg Good points. Retracted close vote. $\endgroup$ – Christian Hummeluhr Mar 21 '15 at 13:42
8
$\begingroup$

No, intuition is not related to ESP in modern cognitive science.

A modern view on intuitive thinking

While ESP certainly retains its pseudoscience status (e.g., Rouder and Morey, 2011), intuition and intuitive thinking has been used in the psychological literature in evolving ways over the years. Outside the heydays of Skinnerian radical behaviorism, the notion of intuition has been consistently present in the cognitive sciences.

An influential modern model of intuitive thinking is called dual process theory (e.g., Evans and Stanovich, 2013). The dual process theory model, which entered the public consciousness with Daniel Kahneman's popular Thinking, Fast and Slow, proposes that reasoning happens by two discrete "paths": the deliberate and reasoned System 2, and the fast and intuitive System 1.

This account explains intuition and intuitive thinking in terms of different heuristics, which should be familiar. Heuristics can have multiple subtly different meanings depending on field and context, but broadly speaking, a heuristic is a decision process which is usually right. You can find a number of references about heuristics in this question.

While ESP is non-deterministic almost by definition, dual process theory does not take a position on determinism. We are simply not consciously aware of the underlying heuristic machinery's workings. (As Artem notes in his comment.)

References

$\endgroup$
  • $\begingroup$ Since the OP seems to have some concerns/confusions about intuition as a (potentially) non-deteministic process, I would suggest adding a bit to this answer on how the dual-process view of heuristic is not incompatible with determinism; you are simply not consciously aware of the machinery that arrives at the intuition, but that unconscious machinery can still be deterministic. I could write this as a separate answer myself, but I feel like that would be spending too many words on a passing comment by the OP. $\endgroup$ – Artem Kaznatcheev May 10 '15 at 1:55
2
$\begingroup$
  • Is 'intuition' related to 'extrasensory perception' (ESP)? As far as I know, ESP is not accepted by the scientific community but I could not conclude the same about 'intuition'.

You're correct, ESP is not accepted by the community, and presuming, for the sake of argument, it could be true, the scientific experiments should be trivial to be done and replicated.

  • If they are related, how does scientific community explain 'intuition'?

Intuition is deeply embedded knowledge you have acquired. As a simple example, when you hear a foreign speaker, you "just know" that some words "sound" wrong. And so would they, if you try to speak their language without having acquired it over many years (and most specially, the younger years).

There are at least three schools of thought here, not at all incompatible, in my opinion:

  1. System 1 versus System 2

In the first school/set of studies one calls intuition as "System 1", and research tends to focus on deviations from "rationality".

As an example: a bat and a ball cost 1.10, the bat costs 1.00 more than the ball. How much is the ball?

Subjects respond .10. But that is the wrong answer. Here is the wikipedia page on "Thinking, fast and slow", from Daniel Kahneman, and here is the one on its major school of thought.

  1. A second school, mostly headed by Gary Klein and colleagues, concerns recognition-primed decision.

Here we have the ongoing action of jet pilots or firefighters as they rush to solve rapidly changing situations. The books by Klein and the references on the wikipedia page will give you some pointers.

I have a presentation of his work here (never mind slide 10), and I have proposed that this model is reconcilable with the amazing results of Benjamin Libet, and the slides are here.

  1. Computer models of fluid concepts.

Another set of researchers, following Douglas Hofstadter, have built computer models that are, in a very simple way, "intuitive". You may want to check "Fluid Concepts and Creative Analogies" and the later thesis (Called Phaeaco, Musicat, Metacat, SeqSee, Letter Spirit, or my own project Capyblanca on an intuitive chess machine, published here and here, with source code available here).

These models have some deeply embedded knowledge that can reflect the properties of human intuition. A simple example is the project Numbo (on Hofstadter's book), that "knows" that 19x21 is "close" to 400, but still has to do some "thinking" to get to the correct answer. Here we get in-between System 1 and System 2: System 1 "points" to a number close to 400, and System 2 kicks in and does the final computing.

If your interest is on computers & intuition, you should definitely read Hofstadter's Fluid Concepts & Creative Analogies and the above thesis.

I also did a quick interview to "Psychology Today" some years back about intuition, but that's more about pop journalism than actual science. In my opinion, this is one of the most promising areas of research in cognitive science.

$\endgroup$

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.