HN Explorer

Ask HN: Next steps for solo researcher arriving at Computational Theory of Mind?

6 points by aeim ↗ HN
There seems to be a simple way to think about and explain our minds that is consistent with experiential practices like meditation, observations of scientific method, and with mechanisms of evolution by natural selection.

A proposed Computational System design describes brain mechanisms and mind processes that explicitly and specifically impact the way we are - it is predictive and testable.

Initial reasoned observations appear to suggest consistency across phenomenological experience from volitional and circumstantial behaviours to often overlooked or dismissed idiosyncrasies (& points to explanation for a number of curious observations in scientific experiment).

Jolly good.

The problem is, I'm not in academia, and it's taking a while to write up. It has become much bigger than one person.

  - Any tips/ thoughts on moving forwards? 
    - Positive
    - Cautionary
  - Concept validation?
  - Connect with academia?
    - Or relevant specialists?
  - Independent release? 
    - Incremental?
    - What would you want to know first?
  - I'd prefer to remain anonymous initially as if it is shown to be formative it will upset some people
    - Any thoughts on this?
  - Would I be better asking somewhere else? 
    - If so where?
Many Thanks

5 comments

[ 4.2 ms ] story [ 17.0 ms ] thread
manx
If you want to share, I'm interested in the topic and willing to listen. I'm doing my own thought and code experiments in this area privately. Email is in profile.

I don't recommend going into academia. The system is unfortunately too broken right now.

You can reach many people by writing a good blog article, and posting it to interested communities like HN. Even anonymously, if you prefer.

aeim
Thanks for the connect, I will.
p1esk
You don't need to write anything up. No one would care to read it (no one you'd care about). I'm guessing at least a hundred "Computational Theory of Mind" papers get published every year. Build a demo of what your system is capable of - ideally it should be a task current deep learning models have trouble with. If it works and is in fact novel, people will notice, and maybe even want to play with it. Only then, if you ever get to that stage, it might become bigger than one person.
aeim
Thanks for responding, to clarify:

It's a conceptual model and the abstraction is a computational system - as opposed to the current trend of Integrated Information Theory (IIT). Writing up seems to be the means to demonstrate for the time being. If I've misinterpreted please do correct me.

  - IIT is exotic and complicated. Simply reducing the bounds of any problem IIT might be needed to explain may help. I'm absolutely suggesting a more conventional and therefore accessible logical abstraction can explain much of what we see.
  - Perhaps analogous to 'you don't need blockchain or machine learning for that problem!'
The phrase 'bigger than one person' is also ambiguous - I simply meant in the volume of work required to demonstrate the scope of significance. Perhaps more specifically:

  - How might an independent researcher find collaborators with related tangential special interests or experience?
I appreciate I may have said more than you're interested in, but thanks for the opportunity to respond.
p1esk
I am an AI researcher who's interested in how a brain works. Do I want to be your collaborator? No. Why not? For the reasons I stated above. Even if you had a working demo, it's not very likely it would capture my interest - unless it was doing something truly mind-blowing. I've seen a few demos over the years (OpenCOG, Spaun, Numenta's NUPIC, etc) - not impressed so far. And you haven't even bothered to build a demo.

Even if you describe something that's better than IIT, no one would really care, because no one really cares about IIT either. It does not provide any useful or particularly interesting insight into how a brain works. It's a philosophical theory, and I want to see something a lot more practical.