Intuition is wrong, unless these conditions are met | Hacker News Reader

Intuition is wrong, unless these conditions are met

thinkadvisor.com

42 pointsprostoalex17 comments

Previous discussion: https://news.ycombinator.com/item?id=18570797

1. Regularity (predictability) 2. Practice 3. Immediate feedback

i.e. conditions for learning

Learning is more general than that, because you can learn by thinking carefully even when feedback is delayed. Likewise, you can learn without practice, for example when you learn why a theorem is true you don't need to repeat the epiphany many times.

The only thing it has in common is the regularity condition because no matter how much you learn, if the system is not possible to predict then there is nothing to learn in the first place.

So... This is just like, his own intuition, isn't it? I mean, maybe there are savants out there who know things with great accuracy without familiarity, practice, or feedback. As a general claim it seems weak.

I'll go further and say the vast majority of human knowledge is produced by intuition. For example, the scientific method relies on inductive leaps made based on drawing on repeated observations. This method cannot yield knowledge without intuition; Einstein sits in a room and ponders the Michaelson-Morley experiment, then comes up with his model of relativity. What brings him to that point? How does the previous knowledge yield a new supposition? It emerges ex nihilo from his stupendous brain.

Of course the nice thing about intuition is we can later validate it in other ways and shore it up. We dont just have to skate along believing our raw hunches. But without the ability to just dream it up we would be left with very little real knowledge to work with.

I agree with what you've said, but I dont think it contradicts the article. The article is about when to trust your intuition, e.g. when buying stocks. I don't think Einstein had to have the same level of faith in his own ideas to whiteboard them out a little to see if it was tennable.

I think he did. Buying stocks is ultimately a private decision (unless you work for a firm), and it takes no special knowledge to buy them. But Einstein had to submit a paper to some of the smartest people on the planet for review.

At any rate, yes, hypotheses are basically formed by intuition. They're not random guesses, that would be quite inefficient.

I think there is a big difference between having an intuitive stroke of genius and then spending lots of time crafting a paper using rigorous mathematics and physics based on that idea. The later is not a process driven by intuition. And the former doesn't have a real cost like buying stocks do. If the idea popped into his head, but after an hour he realized the math didn't work, it would have been no big deal.

I don't know if you know about his famous thought experiments imagining chasing light beams, but it was this intuitive insight that led him to discover relativity. He would have never gotten there through a formal process. Sure the math had to validate his hypothesis, but through hypothesis itself, the insight itself is the most important thing, because it was such an original idea.

Einstein was building on the work of Lorentz and Poincare, which is why we call some of the basic equations of relativity the Lorentz transformation, not the Einstein transformation. He didn't come up with it ex nihilo (on the other hand, general relativity was much closer to purely his).

There were a lot of people trying to make a relativistic theory of gravity. Nordstrom's theories were precursors to General relativity and there is also an argument to be made that Hilbert discovered general relativity before Einstein.

You see intuition at play all the time in how people decide which side of an argument is the baseline and which requires evidence. So some people will need extraordinary evidence to prove the existence of climate change. Others would need extraordinary evidence to prove the absence of human causes. Which side you fall depends a lot on intuition.

I have no data supporting this (ironically, just intuition), but examples you give might only be a small chunk of attempts that brought a successful outcome. In other words, when one tries to achieve something, they'll initially fail and only after some more trial and error they have enough data to succeed.

I'd agree that ideas start from human intuition, but above would suggest that intuition is often wrong, so people need to apply it at least a few times in order to get it right. I think that's the point of the article.

Uhm, no, but funny. Your intuition that Dr. Kahneman's general claim about intuition is weak, is way off, but humorously apropos.

This is not just his own intuition. Daniel Kahneman is a Nobel prize winning psychologist principally responsible for inventing an entirely new field of social psychology called Behavioral Economics, which is based largely on how humans misapply intuition.

Working with Amos Tversky, with whom he shares the Nobel prize, and others, Daniel Kahneman did extensive research demonstrating just how our intuition is wrong. You can read more about it in his book "Thinking, Fast and Slow".

Experiment after experiment show that most people's intuition is generally wrong, outside of very specific contexts as the article discusses. They also show that people are resistant to intuitive contradictions and use their intellect to create arguments to justify their flawed intuition, rather than accept their intuition was wrong when confronted with new information. In other words not only are intuitions often wrong, but wrong intuitions are sticky and very difficult to correct.

Now my question to you is, do you feel any resistance to changing your initial intuition despite this new information?

No. I know who Kahneman is and realized after a moment's thought that my initial response was wrong; probably based on an impulse to reject any authority. I tossed off the comment just before going to bed, immediately reconsidered and tried to delete it but could not. I left it as is figuring the responses would be interesting. That intuition, at least, was correct.

Those are basically the conditions for training a neural network: regularity (there is a pattern that can be learned), repetition, and feedback.

Whatever theory you come up with for when intuition is right or wrong, it seems to me that this theory is itself validated (rightly or wrongly) by intuition. That we adopt the ideas that we intuitively prefer is inevitable and not inconsistent with rationality and empirical evidence. The latter are themselves clusters of philosophical ideas which we prefer.

I have a trust-but-verify relationship with my feelings.

Reply on news.ycombinator.com