Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

>>> The game involves asking players to watch a series of digits from two to nine flashing up one by one, at a rate of 100 digits per minute. Over the course of five minutes, players must press a button when they start to see a sequence emerge.

C'mon, this could be cloned in a weekend ;)

Publish neuroscience findings in a public access journal by all means. Even including pseudocode of the sequence generation algorithm.

But I am all for distributing their particular implementation via private sector partnerships. That could yield further funding and experimentation. And possibly alleviate taxpayer research dependency for their lab in future.

The work itself taps into some novel neuroscience. Do we we possess a Bayesian brain that estimates probabilities in real time? Altering a belief net based on new evidence. Or are patterns hard wired and must be learned. I think this sort of training game based on integer series could work just as well with text, images, music, video, animation, etc.



> Do we possess a Bayesian brain that estimates probabilities in real time?

You should read Daniel Kahneman's "Thinking Fast and Slow" - the answer is a resounding "No".


You might wish to read Gerd Gigerenzer as well as Keith E. Stanovich, Richard F. West & Maggie E. Toplak before citing Kahneman so quickly, and doubly so in support of such a tremendously strong claim as the one you just made.

For Gigerenzer, you might wish to start out here:

https://en.wikipedia.org/wiki/Gerd_Gigerenzer#Heuristics

https://www.tandfonline.com/doi/abs/10.1080/1479277914300003... (Available here: http://library.mpib-berlin.mpg.de/ft/gg/gg_how_1991.pdf)

And for Stanovich et al., you might want to start out here:

http://cognet.mit.edu/book/rationality-quotient

https://en.wikipedia.org/wiki/Dysrationalia

Stanovich et al. showed that Kahneman & Tversky's classical "System 1/System 2" dual process model as one too simplistic, outlining at least 3 systems involved:

* The Autonomous Mind

* The Reflective Mind

* The Algorithmic Mind

For points addressed by both Gigerenzer as well as Stanovich et al, see here:

https://en.wikipedia.org/wiki/Great_Rationality_Debate

None of this of course directly addresses your "Bayesian brain" point, however, for that, you might want to take a look at "some" articles by psychiatrist Scott Alexander, many of which speculative but citing a lot of sources you might wish to look into:

https://www.lesswrong.com/posts/ZiQqsgGX6a42Sfpii/the-apolog...

https://www.lesswrong.com/posts/euJm4RwkAptZnP89i/bayes-for-...

https://slatestarcodex.com/2015/08/28/mysticism-and-pattern-...

http://slatestarcodex.com/2016/09/12/its-bayes-all-the-way-u...

http://slatestarcodex.com/2017/06/28/why-are-transgender-peo...

http://slatestarcodex.com/2018/12/11/diametrical-model-of-au...

https://slatestarcodex.com/2019/01/10/paradigms-all-the-way-...

(And here, as a bonus, an article on dreaming, by Eliezer Yudkowsky: https://www.lesswrong.com/posts/8z2Fm2yaHpQz8rr5B/dreams-wit...)


Real time is also a funny concept when it comes to perception. There is a TED talk where the speaker explains how your perception of "now" is constructed over about half a second after receiving input. And can be retroactively overwritten in the next few seconds, if more convincing input contradicts the created picture. 100 images per seconds seems to be just a notch below this mattering, but probably sustaining this needs concentration.


Research I have seen suggests we do construct Bayesian prices but we don’t apply them rationally. If people assess that something happens 60% of the time, then in a series of 10 observations they will predict it happening only 6 times.

That is to say, they will bet on the 40% probability outcome 4 times, even though they believe the 60% is more likely each time


It took me a second to realize the meaning of this; do you have any links - sounds fascinating :)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: