NC unit secretary
2020-10-08 11:56:45

This is a discussion channel for " Consciousness and Intelligence" by Dr. Ryota Kanai (ARAYA Inc.) A link to the talk is the following. Please do not share the link to anyone outside of this slack workspace. Access Passcode can be found at the announcement channel.   URL: https://vimeo.com/471280015 (49 minutes)

👍 Taiki Miyagawa
Tomasz Rutkowski
2020-10-09 13:56:32

Thank you for an inspiring talk, Dr. @Ryota Kanai. Would you envision brain-computer/machine-interfaces (BCI/BMI) together with machine learning tools to be useful to probe or even validate your approaches to consciousness in practical applications?

Ryota Kanai
2020-10-09 15:35:27

I want to build brain to brain communication devices to establish new approaches to consciousness

✅ Tomasz Rutkowski, olewow
👍 Tomasz Rutkowski, Rm Banerjee, Taiki Miyagawa, Alexis Holgado
Tomasz Rutkowski
2020-10-09 15:55:40

*Thread Reply:* Wonderful idea! Thanks for the answer.

Ryota Kanai
2020-10-09 15:37:08

We recently got a huge grant to build BMI systems, and hope such experiments will become possible by joint effort by new neurotech industry and curiosity driven research

🙌 Tomasz Rutkowski, Sho Hoshino, olewow
👏 Tomasz Rutkowski, Rm Banerjee, Taiki Miyagawa
Tomasz Rutkowski
2020-10-09 15:57:58

*Thread Reply:* Congrats!!! Looking forward to the results of the project.

Ryota Kanai
2020-10-11 14:38:51

I just added a #recruitment post. If you are interested in the type of research we do and BMI, please contact me.

Emtiyaz Khan
2020-10-12 00:07:52

@Ryota Kanai Your idea of “metarepresetation” sounds like a very good idea. Some of the ideas from “function-space” view in wide-neural networks are trying to achieve this (of course this is early stage). But the idea makes a lot of sense, and could solve some of these problems! Also, thanks for your talk and the panel discussion.

Ryota Kanai
2020-10-12 00:16:29

*Thread Reply:* Thanks for the kind comment. Yes I’m also excited with this idea, and we are working on it both in terms of AI implementation and neuroimaging. If you are interested in this, let’s discuss further.

Emtiyaz Khan
2020-10-12 23:49:09

*Thread Reply:* Yes, very happy to be connected! I will send a DM.

Krzysztof Woś
2020-10-12 12:01:55

@Ryota Kanai You've said that "We do not seem to have a collective consciousness as a society." If we did have such collective consciousness, what would it be like? Whenever I think of an individual vs. society, I often wonder about the results presented in this little gem of a paper: https://cragl.cs.gmu.edu/diamonds/. It seems that just the way individual cortical columns in our brain sub-sample sensory inputs in order to create a distributed robust model of the world, which is also distributed across two practically separate but connected processing units (hemispheres), the same principle applies across individuals in a society. Whether the signal transduction happens over electrochemical or linguistic channels is just a matter of implementation. From the perspective of information it's always transmitter, receiver, and a channel. Our individual perceptions collectively, mediated through language, seem to allow us to converge on better and better approximation of the reality which in itself is not accessible to us.

cragl.cs.gmu.edu
👍 Sat
Ryota Kanai
2020-10-12 15:11:05

*Thread Reply:* Thanks for your comment. The truth is that we don’t know if a society has its own mind, and we can only say that consciousness of our scale doesn’t have access to it. Nevertheless, how a distributed system works as a coherent system is a relevant question for understanding the brain and the society.

Krzysztof Woś
2020-10-12 19:23:51

*Thread Reply:* The way I like to think about it is to apply Stephen Wolfram's definition of computation. What is computation? Any system that follows rules is performing a computation. As a great example in our field, Donald Michie's Matchbox AI, a reinforcement learning algorithm implemented with matchboxes and coloured beads comes to mind. So intelligence seems to be a property of computational systems and computational systems are systems that follow rules. Society is a system of rules, conventions, customs, traditions, etc. Thus society is a computational system and as such it's imbued with at least intelligence. If as you say intelligence and consciousness go hand in hand, then society is conscious. As a foreigner living in Japan I've always been struck by the level of social consciousness in this country. The very notion of individual as a part of society is embedded in the character denoting a person, or at least so the story goes.

Krzysztof Woś
2020-10-12 19:28:58

*Thread Reply:* Out of curiosity, have you ever read Erwin Schrodinger's "What is Life?" Interestingly he also writes about consciousness and I think you might find his conclusions interesting

Krzysztof Woś
2020-10-12 19:29:27

*Thread Reply:* Compared to What is Life, his work in quantum theory really doesn't seem that important

Krzysztof Woś
2020-10-12 19:35:46

*Thread Reply:* By the way, I'd like to clarify one thing. While my views on the subject of consciousness differ from yours in some places (largely due to the influence of Eastern contemplative traditions as well as Hermetic philosophy), the Information Closure Theory of Consciousness is thus far my favorite. It's very, very neat.

Sat
2020-10-13 01:45:24

*Thread Reply:* Great comment and thanks for posting the paper reference!

Motoaki Kawanabe
2020-10-12 12:15:39

@Ryota Kanai Thank you for your excellent talk. I also think that “function-space” idea is interesting. It sounds to me a central control system on top of your global workspace. I guess that such an architecture may not completely match the philosophy of IIT (I understood it prefers distributed systems more). I wonder whether your new NICT theory gives higher value of consciousness to the “function-space”+“global workspace” architectures.

Ryota Kanai
2020-10-12 15:15:13

*Thread Reply:* I also feel the embedding of functions is a promising direction. I haven’t thought about how this is linked with information closure, but we are also interested in updating IIT by considering shared goals, and as such there may be a way to connect embedding to information theoretic approaches to consciousness. I haven’t covered embedding much in this talk, but I’m trying to connect the idea to qualia and the so called higher order theory of consciousness.

olewow
2020-11-12 17:48:02

Dear @Ryota Kanai thank you a lot for your inspiring talk. I will be happy to study yours and your team's work further. I find it as a significant effort for unification and/or utilization of the known theories and models to create a powerful approach. It would be intresting to integrate IIT ideas and Φ measures, while the relation with Free Energy and Bayesian inference you presented seem very fruitful.

I would really appreciate some feedback to clarify some things. Maybe my questions are naive:

  1. When it comes to connectionist approaches, except scaling and boundaries should the nature of the connections and the physical substrate be taken into account, or is it irrelevant and the information processes themselves are enough for such phenomena to emerge? Shouldn't we care about how could it be integrated to specialize the model? How could that be approached?
  2. How does Information Closure relate to the philosophical notion of Intentionality? Does it deny it?
  3. Isn't there a problem of computational power, to really test such things? Because we maybe investigate a lower end of possible systems' sizes while the behaviour on a larger scale may be completely different, or not?