New Concept of How Higher Intelligence Actually Works

Last week, as I thought about the neuroscience part of Zindell's book The Orca's Song, I asked myself the question: Is it really possible that some of those Orcas are making a transition from the highest level of fundamental brain organization we see in humans to something a step level higher than that? Most humans would respond with reactive thinking which can block their ability to think honestly about this kind of question. Lots of neurotic extreme group loyalty of all kinds, for all kinds of human groups. Zinxell did not claim that orcas are free from such neuroses; indeed, he even portrays neuroses in orcas in his novel!! But the neuroscience part was about something more fundamental. This morning, I now see how the POSSIBILITIES directly refer to universal principles, like those referred to in the attached slide. I really wish that colleges today would reliably teach the prerequisites needed to fully understand the basic stages of ascent we see in the past evolution of life on earth. That sets the context here. Long ago, I remember a great little prentice-hall text on animal physiology, which discusses the major qualitative shifts in body plan, like fish to amphibian, cold blooded to warm blood, crawling to walking and flying, and so on. I often cite the fundamental work by M.E. Bitterman on qualitative differences in learning and behavior, as one goes up the ladder, and work on comparative neuroanatomy fits. The slide below comes from two of the papers I published in Neural Networks, in 2009 and 2012, which explain the mathematical principles underlying these kinds of shifts, principles which also apply in designing machine consciousness and -- I have claimed more recently -- in understanding noobrains as well, the kind of realm which people are immersing themselves in when they fully engage in behaviors like "astral projection." Those papers won me the Hebb Award of the International Neural Network Society, and fed into NSF-wide support for the COPN funding initiative I often cite. (That is on one of the eight slides I will use to introduce a session at the Apportunity conference this coming week. https://apportunity.eco/conf2022/4-domains-2022/.) BUT: Those mathematical principles only go up to the "sapient" level of intelligence or consciousness, the level which we see humans evolving to, which uses LANGUAGE to achieve a higher level than what monkeys have. The fundamental advances in the level and type of intelligence from rat to monkey to human go... orca??... all are grounded in the mathematics of SYMMETRY. For human, and PERHAPS for Orca, we are seeing an INITIAL INCOMPLETE "quantum leap" in evolution. But different levels of symmetry. The deep neural networks so popular to say among chauvanistic coders manifest a very low level of intelligence on his scale. One of the two major breakthroughs in machine intelligence circa 2009 was the popularization of Convolutional Neural Networks, which exploit symmetry with respect to spatial translation, achieving a great breakthrough in power compared to what appeared before. But that was just Euclidean symmetry. See https://arxiv.org/abs/0710.4182 for another major step up, the more general nonEuclidean Object Net Design. Garbled, ad hoc practical reflections of this principle have led to performance improvements in some applications, where designer/coders understood intuitively that the concepts of OBJECTS play a fundamental role. Object Symmetry -- the symmetry of objects of various types in your visual field, in their dynamics, and in what you see. That is fundamental even in reptile brains for sure, not only mammals. I suppose that the hippocampus in the human brain plays a fundamental role in the encoding process, orchestrating the formation of memories which exploit the concept of Object as part of... handling more complex scenes and episodes. And yes, I view neocortex (and its ancestor in the reptile) as another key PART of a system which exploits object symmetry in predicting and reconstructing what it sees from sensory inputs relayed to it. BUT THEN: Beyond the Object level of symmetry is another level which might be called Actor Symmetry or Subject Symmetry. This principle underlies what I had in mind in my personal definition of the word "multimodular" which I referred to as "collective intelligence" in the mountain slide. It is also crucial to the power of "federated learning" a concept rightly now popular in serious "AI" technology. From objectNets to SocialNets? But no, not the termite mound social networks we see today. The slide suggests that the level of intelligence or consciousness possessed by our noosphere may indeed be LIKE sapient human, augmented by multimodular foundations, and also augmented by quantum foundations (other slides on tQuA to be discussed June 3). Could it be that the new emerging paralimbic and parahippocampal structures emerging in some cetacean brains are a significant if early stage in exploiting THAT level of symmetry? the kind which provides better, more systematic use of faculties like empathy and mirror neurons and all that they lead to? And ability to connect to the noosphere itself, where a similar architecture is used? Is it even possible that a few very early, very fragmentary bits of convergent evolution are appearing in the human genome? I wonder.