In 1990, having spent much of the 80s working on artificial intelligence, and with the idea of writing Brindabella 2200 starting to take shape in my mind, I decided to look closely at the brain to see whether it was plausible for AI systems (i.e. a PA) to closely mimic its function. Before long, looking from the perspective of systems engineering, I realised that the purely neural view I was being presented with couldn't be stable – not even over seconds, let alone decades.
Neuroscientists got around this by pointing to the very large numbers of neurons involved, but I'd seen that explanation used elsewhere. There's a limit to how often you can do that. I learned that glial cells not only feed neurons, thought to be their main role then, but during development of the brain they first establish themselves then direct neurons into position and direct how they connect. This seemed like clear evidence that they weren't just more important than assumed but were the dominant cell type when it came to information processing. At the very least they held the information that structured the brain down to the details of instinct.
Neuroscience had a problem it called the binding problem – what coordinated the firing of neurons that form a particular global response to sensory inputs, or an independent thought. To me, at least, a clear indication could be seen in measurements such as those depicted by Crick and Kocha in 1990. As described in the original caption to this diagram, the neural firing comes just before the negative peak of the oscillation in the local electrical potential of the region. The smooth oscillatory field is triggering and synchronising the neural activity.
Bursts of neural activity (lower) and the local field potentiala
The critical question, it seemed, was what was causing these oscillations – a smoothed out sum of local neural activity as was widely assumed, wide scale electrochemical glial oscillations, or most likely, a combination of both.
Digging through the academic literature on another topic I discovered that in 1985 Renato Nobilib had reported a new view of brain function that incorporated glial cells in addition to the conventional role of neurons. He said, in essence:
An explanation for electroencephalogram (EEG) activity is proposed. Under suitable assumptions ... a Schrödinger-like equation for ion displacement waves is easily obtained. ... Theoretical wave-propagator diagrams are in perfect agreement with experimental stimulus-response patterns directly recorded on brain cortexes. ... in good agreement with diagrams reported by neurophysiologists.”
In a follow-up articlec he reported:
A proof is given that self-sustaining ionic-wave propagations—heuristically inferred by the author in a previous paper concerning a new holographic theory of animal memory—are possible in animal tissues at normal physiological conditions. ...
Theoretical wave patterns and their general features are in excellent agreement with EEG (electroencephalogram) patterns detected on brain cortices and on scalps. Epileptic foci artificially generated by injection of Na+ ions into glial tissue and inhibition of EEG by K+ superfusion of brain cortex, are correctly accounted for by the theory.”
He found that the glial oscillations could be described by an equation that was equivalent to the Schrödinger equation used in quantum mechanics. His use of the word “holography” was his way of saying that the wave patterns were capable of storing multi-dimensional information in much the same way that an optical holograph does. This is a profound observation.
His views have largely been ignored by neuroscientists, possibly because he published in a physics journal. But over the years evidence has accumulated that supports a significant, if not dominant, role for glial cells.
Skipping forward three decades we have Fields et.al.d:
If “the connectome” represents a complete map of anatomical and functional connectivity in the brain, it should also include glia. Glia define and regulate both the brain’s anatomical and functional connectivity over a broad range of length scales, spanning the whole brain to subcellular domains of synaptic interactions.
Moreover, this broad spatial integration across distant brain regions is achieved across exceedingly wide temporal scales ranging well beyond the millisecond to seconds of electrical signalling typically recorded in neurons, to encompass instead hours, days, and months. These longer time frames are well-suited to the temporal dynamics of glial communication ... .
Does this whole brain perspective answer the question that originally puzzled me about the stability of the brain? Very well, I think. We now have two distinct systems operating in tight synchrony, neural networks and glial oscillations, that each has a unique mechanism for communication.
Neural connections (over short and long distances) are mediated by synapses that vary greatly in strength over short and long time scales. Inter-glial communication is via ion flows between adjacent cells through gap junctions – small tunnels through the cell walls – which can change in number to modify oscillatory patterns over long time scales, but are stable over short time scales.
In addition to the relative stability of the glial connections, the two systems are likely to respond to changing metabolic conditions in the brain in different ways, so each will tend to correct changes in the other that do not fit their combined function.
Going well beyond the stability question, if Nobili is right about the holography analogy we not only have a stable information storage system but one that is interestingly global. If you have a holographic picture of an object and just use any small portion of it, you still see the whole of the original object, just in less detail. This model ties in well with the glial mass having distributed information that can position and connect the neurons during development and continue to moderate their connection during normal brain function.
In Brindabella 2200, the main characters pursue a quest that leads them to discuss these issues in some depth. As far as I am aware, it's the first time this slowly emerging view of the brain has been discussed in terms of a complete information processing system. They also tentatively extend it towards a theory of mind.
Something else they discuss that doesn't seem to have been addressed in our contemporary literature is the evolution of the neuron.
a: Francis Crick, Christof Koch, Towards a neurobiological theory of consciousness, Seminars in the Neurosciences, V2, 1990, pp 263-275.
b: Renato Nobili, Schrödinger wave holography in brain cortex, Phys. Rev. A 32, 3618 - 3626 (1985).
c: Renato Nobili, Ionic waves in animal tissues, Phys. Rev. A 35, 1901 - 1922 (1987).
d: R. Douglas Fields et.al., Glial Regulation of the Neuronal Connectome through Local and Long-Distant Communication, Neuron 86, April 22, 2015.
Quotes From Brindabella Trust
‘... the electrochemical glial waves that resonate both locally and over the whole cortex have a kind of momentum like the physical momentum of the bike or water waves. They're activated by neural pulses that propagate around loops of connected neurons – synfire chains. Other chains with similar timing and connections are drawn into the resonance.’
‘Like the planetary resonances but fleeting.’
‘And like the distinct planets, they can form distinct multiple resonances when they're non-commensurable.’
‘Differing by the golden mean. Their resonant frequencies not related by a simple ratio of integers. If they are related they'll join, or if the phase is wrong they'll interfere and destroy each other in chaos.’
‘And if they don't interfere over many cycles we can keep multiple resonances or thoughts active simultaneously, even if only one is strong enough to be conscious. Eventually, synfire loops that the different resonant systems have in common may emerge as dominant because they're being stimulated by multiple resonances, and we become conscious of the commonality between them.
‘That's temporal synchronisation. It also acts out spatially through the brain. Particular patterns of neural excitation evoke particular glial resonances across the cortex. These can trigger new patterns of neural activity, or memories, in different cortical regions.’
‘Linking sights with sounds and so on.’
‘And more complex clusters of associated memories. The golden ratio gains aesthetic value by extending the breadth of associations. We seek and enjoy resonance.’
‘The arrangement of a flower's petals, the proportions of a beautiful face, and the music of the spheres in the harmony of a musical chord. If the components are separated by the golden mean we identify the whole from its parts more readily. We say it's “easy on the eye” because it literally is.’