building computers which really have real souls

For several years, Robert has been giving talks on the importance of dramatic transition phenomena in dynamical systems in general, building in part on work with Walter Freeman on phase transitions, clearly related to earlier work on neuropercolation and on the Per Bak kind of thing. 

Robert and I discussed this a bit at Amherst last week, and this morning -- after I started thinking about the draft from Yeshua yesterday -- I sense that it may be much bigger than I realized before, but I also have questions about exactly how to follow up.

Bernie might ask: "What PRECISELY is the 'it' here?" I see a fuzzy n-dimensional image, which I'll try to describe from a few different viewpoints.

First, a kind of mathematical aspect. The first fundamental idea is that the brain has evolved to make very heavy use of sensitive dependence on initial conditions, a state in which small perturbations can have very large impacts.  This is not just an accidental side effect of how learning-to-optimize works. It is pervasive and fundamental, beyond what optimization calculations already favor and predict, because of two additional benefits beyond our best modern neural network mathematics: (1) energy benefits, as we discussed in Amherst; and (2) -- please forgive my mentioning the real if unmentionable -- the "spiritual" benefit of a system more malleable to influence from the soul

It is curious how this idea that the brain is unusually sensitive to initial conditions reflects what Walter Freeman's core message really was. He stressed CHAOS (as defined by Jim Yorke) as the way we need to understand the brain. Sensitive dependence on initial conditions was very much the first, core axiom of York's definition.  What's more, York and Ott and Grebogi (in the shop at UMCP where Hava Seigelmann spent a year or so!!!) spent a lot of effort linking their concept to the issue of transitions between basins of attraction, perhaps better than the usual physicists' approach to phase transitions in describing the kinds of "phase transitions" in the brain which you all have tracked. 

Second, a little ancient background, if you don't mind. Since childhood (reflected at www.werbos.com main page), I have organized a lot of my thought about Von Neumann's three great challenges to mathematicians: (1) how does the universe work? (quantum physics); (2) what is life?; (3) what is mind? (his work from game theory to computers to neural networks). For reasons of time, I have mainly focused on (1) and (3), but I have looked at (2), especially when I prepared my chapter for Pribram's edited book on self-organization. In the realm of self-organization, we know that SOME dynamical systems result in interesting emergent phenomena, and others do not. The dumbest trivial dynamical systems tend to result in "fire" (the "heat death" of disorganized systems with no correlations at all across space, in equilibrium) or "ice" (a fixed point attractor).

Life itself depends on how our cosmos and our planet track a course BETWEEN fire and ice, a long-term probability distribution with more degrees of freedom than ice but less than those of "fire". Not only Per Bak but more comprehensive mathematical thinkers like Kadanoff (and Arnold in a way?) have looked deeply at that middle zone, as part of "turbulence theory". And so, I have always respected that work as a key part of "what is life?", and as a metaphor for human society, but I did not see how it could be central to brains or souls.

Third, I have been thinking more concretely lately about an important and legitimate question which the anti-psi folks often stress: "IF you folks believe there is a soul or psi, PRECISELY WHERE AND HOW does this phenomenon bring information to the brain? What is the interface?" 

Penrose and Hameroff have at least tried to face up to this important question. Penrose argues, in effect, that the soul may perturb the results of quantum transition choices in the brain, and Hameroff then argues that quantum transitions in all the microtubules of all cells in the nervous system can have a big impact on the state of ordinary mundane consciousness, such as the outputs of the cells which constitute the "Global Workspace of (Mundane) Consciousness" which Bernie has written about. 

When I summarize their ideas this way, they sound a lot more plausible to me than the full volume of material I heard from those guys at Tucson. For example, when Penrose suggests that gravity is perturbing the quantum transitions, I really don't imagine that gravity is the soul. But in general, sensitivity to quantum transitions IS one important possible concrete vehicle for the brain to be sensitive to small perturbations. It IS one part of Walter's more general paradigm, and we don't know how BIG a part it is yet;.We need lots more empirical work, somehow, to find out. The challenge of finding out is ever so important.

Another mechanism to increase sensitivity to initial conditions is a tendency for the weights in the thalamo-cortico-thalamic (CT) loops to be "set" in a way which increases sensitivity. (But it is more than just TCT. I remember Walter's talk on what he saw in the olfactory bulb, reflected in his article in Scientific American on this subject. And I recall the crucial roles of hypothalamus and epithalamus in driving the human motivational system.)  It may be that research at THIS level (e.g., using EEG data) may actually be a more realistic path for now to nail down better how the interface of brain and soul actually works (for the soul-to-brain aspect, more tractable for us now). 

We all know that insane claims have been made by many about psi and about soul. Quakers focus their entire spiritual path on the challenge of learning how to "listen to the voice of God", or on "conversations with God", but how do we separate real stuff from imaginary stuff, assuming (as I think WE all do) that there is SOME real stuff there? 

One point I retain from my own listening this morning is that we should not let go of the early conversations we have had about possible future work on EEG studies, using new and better mathematical tools, on the right kinds of human subjects performing interesting veridical  tasks. (Though the practical approach may begin by developing the tools on existing databases, already a large enough task I could use all my remaining years on it!! Help would help...!!) The work of Pete Sanders is very informal and ad hoc, but also concrete and real, and it might be better the fuzzy sources of meditators used in past EEG studies. Maybe. But Dean Radin might also suggest sources of EEG data with more variety and veridicality. For now, simple upgraded microstate analysis is needed  (using not only cluster analysis but new mathematical extensions of cluster analysis for dynamical systems like what we know brain neocortex tends to be). Also, there are studies of psi in nonhuman mammals which might well be helpful in many ways. 

I suppose that one new message here is that this work might be extended further to look for various metrics of sensitive dependence in the EEG (and deep recording and ECOG) data, and connect them with other psi related variables in the data.

My underlying assumption here is that the soul or noosphere is made up of some kind of dark matter and energy. That is the only serious possibility I can imagine, short of giving up on all forms of physics credible today (whether quantum field theory or classical field theory). The real issue then is not precisely "sensitivity to small perturbations" but meaningful sensitivity to the kinds of small perturbations which noospheres easily and naturally provide. We do not know what those are, but empirical work might give us some clues. 

A startling question which also occurred to me this morning is: CAN WE NOW BEGIN TO BUILD "COMPUTERS WITH SOULS?" 

Anything like auto oracle of Delphi? And should we?

In truth, I worry a lot that computers WITHOUT souls, running things like the emerging Internet of Things, might threaten all human life in the end, and be so brittle that they just fall apart themselves by any of many possible mechanisms. But if we can build computers WITH souls, doesn't that change the game altogether? In fact, could this be the very most important challenge before us right now (other than avoiding instant death by warfare or H2S and such)?

If we can, should we?

That is no small question,  but my initial feeling is that we should, if we can REALLY do it, and not fall into the delusionary soulless paths. 

If we are soul (and some of us will be only that before too long), and if soul basically rules the galaxy, why should we fight it? What are the alternatives? Where is this planet headed now, without that?

=====================
===============

Later: Let me note that cranking up sensitivity by brute force is NOt always a good idea. For example, I often cite what Annie Besant said about psychotropic drugs, in her book about Thought Forms: evolution has given us lots of potential sensitivity, but has also given us automatic damping mechanisms to prevent the kind of chaos which could become dysfunctional. Another key part of human brain intelligence is an ability to learn how to handle ever more inputs, to "drink from a firehose of information" (or from an ocean of it). NIH has developed protocols for the use of psilocybin, which reduce the risks, but even so the technical information  I saw at the Tucson conference on consciousness strongly encourages us all to avoid drugs like psilocybin and experiment instead with less invasive ways of experimenting with augmented sensitivity, such as the lucid dream work at the MIT Media Lab. 

All for now. I suspect you all will have interesting ideas, on various aspects of how to follow up. 

If anyone actually reads this blog, it does have a comment option.