The Psychology and Neuroscience of Curiosity [pdf] | Hacker News Reader

The Psychology and Neuroscience of Curiosity [pdf]

cell.com

32 pointsbrianclements11 comments

I've always been fascinated by curiosity. Like invention, it seems to integrate all the merits that make homo sapiens sapient. (Of special import to me after all my zoology classes seemed to dwell on that difference to excess).

I think too few past investigations of cognition, especially AI-motivated and abstract / computational models of cognition, have approached the topic from assessment of human higher cognitive functions like creativity, invention, and genius. It seems there'd be a lot of value toward understanding the mechanisms of cognition by recognizing patterns of extraordinary performance and then decomposing them to consistuents (as neurologists do clinically), rather than starting with the constituents and trying to compose them into higher models of cognition (as biologists or AI folks do, systematically or synthetically, respectively).

IMHO, AI approaches to cognitive synthesis have been largely a no-go. Perhaps more dissection of high-level cognition could help?

I do know some folks have pursued research agendas somewhat like this -- cognitive scientists like Doug Hofstadter, or cognitive psychologists like John Anderson (ACT-R) and AI engineers like John Laird (Soar) or cognitive modellers like Thad Polk.

But has anyone seen continuing progress toward more than building bottom-up synthetic models that demonstrate an occasional human-like cognitive pattern, or bottom-up observational psych/neuro research papers on phenomenology, using this focus? I'm not enough attuned to recent work to know.

Everyone is biased to some extent by their own fields of study, and as the saying goes, to a person with a hammer, every problem is a nail. Since AI fields have long been incubated by people in the software fields, it becomes a language/data/algorithm problem. And not that those things aren't important, they definitely are a piece of the puzzle. But ask them more philosophical, pedagogical, or creativity based questions, and they are easily stumped. You need to know what your trying to emulate! Or at least have a philosophical view on what consciousness or creativity is first.

I've long imagined what a software spec would look like for one trying to build an AI. I imagine it would be equal parts technical and philosophical.

I personally think more AI firms and researchers need to permanently employ professional artists, teachers, and creative types. But perhaps I too may be biased ;)

I assume everybody here is familiar with the feeling when you suddenly understand something; it's as if pieces of a puzzle suddenly click together and create the whole picture. So, my question, what is this feeling? How does the brain recognize that it should have this feeling? (Because we can feel it in so many abstract things...) Maybe curiosity then can be just explained as a craving for this feeling; but actual cognitive basis behind this feeling, at least to me, seems lot more interesting.

"So, my question, what is this feeling? How does the brain recognize that it should have this feeling?"

If we take the idea seriously that the universe is unified and all knowledge is unified, then it would be safe to treat the universe as a single entity. AKA everything is related to everything else in a whole host of mundane and non mundane ways that human beings are not aware of given the limits on the human brain/body and the time and energy required to process that information. AKA all the truth is there (the universes structure and its relationships). Rather our mind is busy highlighting things that are interesting, it's just the brain buildings models and finding interesting links unconsciously and when its done we get an update. AKA most thinking is done all the time unconsciously without our awareness of it. See the science:

https://www.youtube.com/watch?v=PYmi0DLzBdQ

I'm gonna only add a troll comment and downvote me if you will, but PDFs break the web. That most academic publishers use PDF drives me crazy!! Why did we allow this to happen?

I favor PDFs because they are in form of self contained concentrated knowledge.

I can save them, print them, read them on e-book and they play nice and they do not have unnecessary elements like menus or side panels or what ever.

The same can not be said about a web page. If nice printing option is not provided then it is normally near impossible to print correctly, if I save them, it is not sure that they would display correctly years later, if I do not save them, I could not be sure I could actually access the information years later.

Hmmm, This sounds like a standard whose time has come. Web pages that easily transform to a nice printable form, or perhaps multiple transforms. Occasionally I will be on a web page that has a "print format" option, but most of them suck. Now, How to monetize??? First thought is to be one of the new open science publishing platforms funded by donations and appropriate advertising.

I think the only way for everyone to get what they really want is to separate the "view" from the data. You can pack a lot of metadata, links, binary data (pictures) into a standardized and compressed format, then you can allow all the various mediums (web, desktop view, in-app text only, in-app pictures only, etc.) to create their own support/plugins for accessing that data package. Ship one package, view everywhere. And the plugins can be made for that medium safely from scratch. They can even simplify/allow/disallow certain categories of info as needed (allow text, disable links, don't view pictures) so that the data morphs to the needs of the medium/plugin.

addendum: I would add that mediums are usually just incidental and necessary means to get to content. Why force everyone to adopt a medium (adobe binaries) to get to the real focus? The content? Instead, just package the content, and people are great and finding their own paths to it via plugins if the interface and code is open and standardized.

I went on a google quest about this topic. Mapping my way through postscript, file encoding, pdf, xml, json. I eventually found my way onto a format that was right under my nose all along that kind of does all this. epub! Can extract to plain text, web, pdf, view natively, it's comprised of layers that you can use or not use, etc. Submitted it as it's own topic: https://news.ycombinator.com/item?id=10563472

Inertia. But academia is steadily moving towards more diverse presentation formats. Check out JOVE - http://www.jove.com/

(goes to jove.com)

(clicks one of the videos that sound interesting. start hearing about stem cell marker something-or-something)

(popup! "please register to continue watching")

No thanks. I'll take my pdf.

Reply on news.ycombinator.com