October 2023
Climate change and other difficult challenges facing global society are prompting a search for new ways of approaching social innovation. At a time when increasingly complex circumstances are rendering accepted wisdom and existing systems of knowledge less relevant, the concept of “natural-born intelligence” put forward by the theoretical biologist Professor Yukio-Pegio Gunji represents a potential key to overcoming a growing rigidity in societal systems and technology. How can this natural-born intelligence be put to work in a society that is in thrall to artificial intelligence? A proponent of the idea that the natural-born intelligence of human beings is the wellspring of innovation, Professor Gunji here discusses this proposition with his former student, Youichi Horry, who has been involved in a wide variety of work at Hitachi, Ltd. In this section, Professor Gunji discusses his concept of “natural-born intelligence” and how to implemeit it in a society.
Yukio-Pegio Gunji,Ph.D.
Professor, Department of Intermedia Art and Science, Faculty of Science and Engineering, School of Fundamental Science and Engineering, Waseda University
He obtained a Ph.D. in science from the Graduate School of Science at Tohoku University in 1987. He was appointed a Professor in the Department of Earth and Planetary Science, Faculty of Science at Kobe University in 1999 and a Professor in the Faculty of Science and Engineering, School of Fundamental Science and Engineering at Waseda University and a Professor Emeritus, Faculty of Science at Kobe University in 2014.
He has published numerous works, including “Groups are Consciousness” (PHP Institute, 2013), “Life, Indomitable” (Seidosha, 2018), “Natural-born Intelligence” (Kodansha, 2019), and “Yattekuru” (Igaku-Shoin, 2020). His most recent publication is “From Whence Comes the Memory of Having Once Lived in that Game World?” (Seidosha, 2022).
You spoke about how sensing and responding to external factors equates to natural-born intelligence. Could you please go into this in a little more detail?
Gunji: While people often talk about systems being “open,” I believe that this is in fact a very difficult concept. We are frequently aware that the scope of our own knowledge is circumscribed and that more exists outside. This raises the question of how we link together those things that are inside and outside this scope to make a system open. The problem is that there are also “unknown unknowns” that are even farther outside of our scope of knowledge, what I distinguish as “external factors,” factors that we have never even thought to consider. What is important is how we handle these.
In terms of this division into what is inside and outside of our scope of knowledge together with unknown unknowns, what we call artificial intelligence works by applying its inference capabilities to its known world. As highlighted by the frame problem*2, because addressing what is outside our scope of knowledge will get us nowhere, we bring those factors we are able to consider into the scope of what we do know and provide a solution to the problem within this context. This is first-person intelligence whereby we make logical and rational decisions about things after having first established our own scope of knowledge. This goes beyond the realm of computers as we humans also frequently deal with things in the same manner as artificial intelligence.
“Natural intelligence” is the term I use for the transformation of this first-person perspective into a third-person one, meaning a way of thinking that comes from natural science. While it might make more sense in the context of natural history, we utilize the knowledge of the outside world that we have acquired over time as we set about finding out more about our world. This is intelligence that seeks to build an objective third-person scope of knowledge.
In contrast, natural-born intelligence does not seek to build a scope of knowledge. You could call it a “one-point-fifth-person” intelligence, one that simply responds to those external factors that are so far outside our scope of knowledge that we cannot even know they exist. While putting it this way might have you thinking it is some special intellectual capability, in fact natural-born intelligence is something that humans and other animals are equipped with innately. Unfortunately, many modern-day humans tend to think like artificial intelligences and in a convoluted sort of way there is a need to get back to natural-born intelligence.
Of course, the reason I choose to study natural-born intelligence is not because I want to discredit mechanistic artificial intelligence and put my faith in the sort of intelligence that is innate in animals. Rather, it is because I am fascinated by the question of how we can incorporate the natural-born intelligence approach into mathematical models and whether we can get it to work.
Youichi Horry, Ph.D.
Chief Engineer, Corporate Strategy Division, Water & Environment Business Unit, Hitachi, Ltd.
After graduating with a degree in earth sciences from the Graduate School of Science at Kobe University, he joined Hitachi, Ltd. in 1990 at the Central Research Laboratory where he worked on research into computer music and graphics. In 1997, he took up a position as a Visiting Researcher at the Institut National Recherche en Informatique et en Automatique (INRIA) in France, began his research into human interaction in 2000, and established the Hitachi Human Interaction Laboratory (HHIL) in 2003. Since 2010, he has been working on management science for social infrastructure at a number of institutions, including Hitachi’s Advanced Research Laboratory, Central Research Laboratory, and the Matsudo Research Laboratory of Hitachi Plant Technologies, Ltd. He was appointed to his current position in 2022.
He obtained a Ph.D. in engineering from Waseda University in 2018. He was appointed a C4IRJ fellow of the World Economic Forum in 2020 and an expert at WG5 of ISO TC323 (Circular Economy) in 2022.
Horry: Given that, when you talk about what is inside and outside our scope of knowledge, you are only using this to demarcate the world that is already known, one way to sum up what you have been saying is that it is the world in which problems are able to be defined and in which there is a clear link between problem and solution. However, everyday life in the real world is not normally that clearly defined. Your mood might change unexpectedly and have you doing things you would not have anticipated doing, or you might suddenly forget things you knew or make a mistake in a simple calculation. That is what life is like for ordinary people. While people use the terms “logos” (logic) and “physis” (nature), the reason why these things happen is because the world is not made up of logic alone. There is no way to explain it other than to assume the influence of external factors of which we are unaware.
Gunji: What I talk about as “external” may be difficult to grasp, and when you think about what sort of things it might refer to, images immediately come to mind where you can draw arrows showing systematically how they relate to ourselves. However, we can only draw these arrows for things we know that we don’t know, not for external unknowns. When I talk about natural-born intelligence, what I mean rather is those situations where our unconscious or subconscious self reacts to something that happens due to external factors we are unaware of.
An example in the form of a problem and solution is how, when we intuitively imagine solutions to the problem we are considering, we instead come up with something completely unrelated. The intelligence that gives rise to such things is natural-born intelligence. You can also describe it as our capacity for routinely creating new things.
The question, then, is how best to model this type of intelligence? What I have been thinking about is how we should go about handling external factors in a systematic manner.
The book “Intelligence Simulation Using Cellular Automata—Implementing Natural-born Intelligence” that you co-authored came out in 2021. Have models based on natural-born intelligence already been developed?
Gunji: Yes. An indirect way of describing what sort of models these are would be to say, for example, that they involve two elements: a problem and a solution. When a problem is seen as a problem it has no solution. When a solution is found, it isn’t a problem any more. Which is to say, you do not normally have both at the same time. When you create a situation in which these incompatible things are simultaneously affirmed and denied, it is like a gap opens up between them and something comes in from outside to fill it.
You may think that both affirming and denying two things that cannot coexist is not logically tenable. In the real world, however, tiny changes in boundary conditions or a shift in context can cause a problem to not be a problem any more. Moreover, such boundary conditions or contexts that are exposed to external factors cannot be controlled or tinkered with. As we make our decisions in just such a world, one that is beyond our control, it is possible for such situations to arise.
Horry: Do you have a good example of where things are jointly affirmed and jointly denied?
Gunji: As I wrote about in my book “Yattekuru,” there was a handicapped high school student who communicated with people by showing them photographs of things he liked. His mother made up a ring binder for him filled with laminated photographs of his teachers and objects such as police cars or trucks. He would go about showing these to people he met, asking “What is this?” Once the person told him what the photograph showed, he would go on to the next person. When he came up to me to show me his photographs, I started by playing the game properly, but after a couple of rounds he showed me a photograph of a police car and I answered on a whim that it was a rhinoceros beetle. He broke out in a great big smile and hugged me saying “Gunji-san!” Since then, we have been on very good terms.
What happened between us was that, as he was someone who found it difficult to speak to other people, showing photographs to get a reaction was, for him, a way of achieving the communication of which he was only barely capable. Put another way, the situation was framed as one in which the photograph represented the problem and naming the object shown was the solution.
Then I came along and told him a police car was a beetle. He showed me a photograph and I gave him an answer, so in some sense that constituted a solution. On the other hand, it was also a problem in that he must have been wondering what on earth I was talking about, meaning that “rhinoceros beetle” became both a solution and a problem at the same time. You could also take the view that what I said was no more than a nonsense reply and so it was neither a problem nor a solution.
Something being both a solution and a problem is like a Zen koan: all you can do is ponder it. Were it neither a solution nor a problem he could just ignore me for not playing the game properly. In fact, he momentarily entered a state in which the two were in a delicate balance and so was able to experience a new phase of communication. I think it was an experience he enjoyed.
Horry: In terms of the technology, I understand you are using distributed computational models in the form of cellular automata.
Gunji: Generally, simulation by a distributed computational model involves synchronized calculation. Biological processes, on the other hand, arise from large numbers of cells working in an uncoordinated way. For this reason, asynchronous calculation is occasionally used in models that simulate biology. While a variety of methods have been developed to achieve this, they mostly involve uncoordinated actions happening randomly.
However, the workings of living systems are not entirely random, with the cells able to organize among themselves in some way to achieve coordination amid all the asynchronicity. That is, coordinated and uncoordinated behavior is negated while at the same time being achieved. This was the concept that I modeled.
