I shall try to show how reflection on familiar phenomena can lead to a deep multi-disciplinary research project with a collection of related philosophical and scientific questions about life, information, evolution, physics, consciousness and mathematics. In 2011, inspired by some of Alan Turing's work I called this "The Meta-Morphogenesis project" [M-M]. I have recently started calling it "The Self-Informing Universe" project, since its core concern is with how the universe managed to start with physical/chemical matter and produce increasingly sophisticated information processing mechanisms needed by increasingly demanding information-users.
In 2015, Daniel Dennett gave a brilliant lecture, followed by a Question/Answer session, both recorded and available on Youtube[Den]. It's not essential to watch either recording, but if you do, see if you can think of any important questions Dennett doesn't seem to be aware of. No audience member brought them up either. (Some of you have heard me talk about them!)
Alan Turing's concept of a computing machine (now called a Turing Machine) was based on his attempt to present a precise, abstract, mechanisable, characterisation of what humans had been doing when computing, i.e. performing calculations or solving logical problems. As is now widely acknowledged, this contributed to profound advances in mathematics, technology, engineering, scientific methods, and some deep philosophical problems about minds and machines[AMTBook].
Shortly before his tragic suicide in 1954, Turing had been thinking and writing about biologically plausible mechanisms, including neural nets and chemical self-organising systems (in his 1952 paper on Morphogenesis included in [AMTBook]).
Since then, human-produced computing machines have increased enormously in memory capacity and instruction speeds, and in connectivity to other computers and physical devices and even bodies or brains of humans or other animals.
Moreover, the variety of applications has exploded, and the impacts on almost all aspects of industry, science, medicine, commerce, education, politics and social life continue to expand rapidly in variety and strength.
Yet there are curious bottlenecks: a suitably programmed and trained computer can defeat a human Chess or GO champion, but no robot seems close to matching aspects of the intelligence of human toddlers, squirrels, and other animals. Moreover, computers are thought to be especially good at mathematics, yet we don't have AI systems able to make some of the deep discoveries made by human mathematicians thousands of years ago (e.g. discovering that a simple extension to Euclidean geometry makes it trivial to trisect an arbitrary angle, as explained in [Trisect], which is impossible in Euclidean geometry).
Even some of the (proto-)mathematical discoveries (unwittingly) made and used by young children ("Toddler theorems") seem to be beyond the scope of current computer-based theorem provers[Toddler]. Likewise, the creative ability of a typical human child to somehow recreate powerful linguistic mechanisms that correspond closely to the language (or languages) used in that environment, including sign languages in the case of deaf children.[Lang]
Enormous amounts of money are now being spent on projects funded nationally or internationally by governments and commercial or philanthropic organisations, attempting, in different ways, to extend machine intelligence, e.g. by giving them more powerful learning algorithms, or by trying to use electronic devices to simulate the (assumed) functions of neurons in brain simulation projects, or by setting behavioural challenges, e.g. to design useful household robots, or robots to dismantle nuclear power plants, or machines that can acquire and use aspects of human intelligence by doing data-mining in digitised products of human intelligence.
I suspect many of these projects will fail (a) because they don't make good use of already available information about things humans and other animals can do, some of which are hard to characterise accurately, and (b) because they make simplified assumptions about what brains and other biological mechanisms do and how they do it. (This seems to have been one of Turing's interests before he died. Also John von Neuman shortly before he died [JVN])
The talk will cover links to psychology, cognitive science, biology (evolution of intelligence) and linguistics. I'll try to show that common evolutionary explanations (including Dennett's) ignore an explanatory gap that can be filled by a theory of fundamental and derived construction kits and scaffolding mechanisms (partially identified by Erwin Schrödinger: (1944)). We are still only in the early stages of a full understanding of requirements for evolving information processing mechanisms found in humans and other intelligent animals, without which no theory of consciousness can be adequate. (More details on that are in another draft paper.)
The connections between AI and Philosophy used to be taught as part of our AI half-degree course but in recent years the focus seems to have become more engineering oriented, partly justified by the public interest in practical applications of AI, and partly by the volume of non-philosophical material to be covered!
The Turing-inspired Meta-Morphogenesis project -- aiming to account for evolution of sophisticated information processing mechanisms.
Dynamic metaphysical grounding of consciousness in evolution
PDF slides for related talks
Information, Evolution, and intelligent design
Royal Institution Lecture by Daniel Dennett (May, 2015)
Question/Answer session after Dennett's talk. Interesting questions and answers.
How to trisect an angle (Using P-Geometry)
The Birth of New Sign Language in Nicaragua
From the documentary "Evolution", the episode "The Mind's Big Bang". It shows how Nicaraguan deaf children developed a new sign language. Children don't learn languages, they create them, cooperatively. Normally the child is in a tiny minority cooperating with others who have already made a great deal of progress. But not always.
Erwin Schroedinger (1944)
What is life? CUP, Cambridge,
I have an annotated version of part of this book here
[JVN] John von~Neumann, The Computer and the Brain (Silliman Memorial Lectures), Yale University Press, 1958 (3rd Edition, with Foreword by Ray Kurzweill. 2012).
See also [Newport] Tuck Newport, Brains and Computers: Amino Acids versus Transistors, 2015, https://www.amazon.com/dp/B00OQFN6LA
Progress report on the Turing-inspired Meta-Morphogenesis project(8 pages)
Paper accepted for AISB Symposium on Computing and Philosophy
AISB Conference Bath University, April 2017
Part of Abstract
Evolution of information processing capabilities and mechanisms is much harder to study than evolution of physical forms and physical behaviours, e.g. because fossil records can provide only very indirect evidence regarding information processing in ancient organisms. Moreover it is very hard to study all the internal details of information processing in current organisms. Some of the reasons will be familiar to programmers who have struggled to develop debugging aids for very complex multi-component AI virtual machines. The paper presents challenges both for the theory of evolution and for AI researchers aiming to replicate natural intelligence, including mathematical intelligence. This is a partial progress report on attempts to meet the challenges by studying evolution of biological information processing, including evolved construction-kits.