School of Computer Science THE UNIVERSITY OF BIRMINGHAM

Work on this temporarily suspended
while I prepare for a lecture

DRAFT
INCOMPLETE
DRAFT

The Self-Informing Universe
A contribution to the University of Birmingham's "Big Discussion" about university education in 2026.
(DRAFT: Liable to change)

Aaron Sloman
School of Computer Science, University of Birmingham

Installed: 6 Feb 2017
Last updated: 7 Feb 2017
This document is
http://www.cs.bham.ac.uk/research/projects/cogaff/misc/big-conversation.html
A PDF version may be added later.

A partial index of discussion notes is in
http://www.cs.bham.ac.uk/research/projects/cogaff/misc/AREADME.html

BIRMINGHAM BIG DISCUSSION

A new proposal stewing:

The Self-Informing Universe

BACKGROUND

This proposal is related to but very different from the "Birmingham Digital" Proposal,
available here:
https://blog.bham.ac.uk/bigconversation/2017/01/12/birmingham-digital/

That proposal is based on the observation that

"The character of every academic discipline has been transformed by the advent of digital technology"
That proposal aims to support students in acquiring the kind of mastery of recently developed human technology that is increasingly required in every academic discipline.

The new proposal: The Self-informing universe
The new proposal is not primarily concerned with human-created information processing mechanisms. It focuses on naturally occurring forms of information and information processing created and continually extended by our universe, through biological evolution and its products. The processes started long before there were any humans, and were essential for the evolution of humans.

The common origins of the ever expanding variety of forms of biologically-rooted information processing were not noticed until Darwin and Wallace proposed the theory of evolution by natural selection, providing many kinds of evidence, to which other researchers have been adding ever since.

Evolutionary processes, reproductive and developmental processes, and many of the processes involved in survival of individual organisms involve use of information -- some of it provided by the mechanisms of reproduction, some by sensors detecting states and processes within organisms and in their immediate environment, and later processes of teaching, learning and social interaction.

The basic use of information is control: initially influencing selection between internal or external behavioural options, then later control of more complex and indirect uses of information, such as generating and testing explanatory hypotheses, or getting information from other individuals, either deliberately or as an automatic side effect of some behaviour. which has inspired a vast amount of research into information transmitted through biological mechanisms, including natural selection.

But there are still many gaps in our knowledge about that process, especially as regards the evolutionary changes in information processing mechanisms and types of information acquired and used.

Moreover the theory of natural selection assumes the existence of mechanisms generating the novelty and diversity from which selections can be made. Natural selection alone cannot explain where the novelty generators come from or how they work.

A few biologists have noticed this gap. E.g. it is confronted in M.W. Kirschner and J.C. Gerhart (2005).

See also this interview with Kirschner
http://bmcbiol.biomedcentral.com/articles/10.1186/1741-7007-11-110
"So I think that to explain these developments in terms of the properties of cell and developmental systems will unify biology into a set of common principles that can be applied to different systems rather than a number of special cases that have to be learned somehow by rote."

However, there is still much to be learnt about the complexity, diversity, and in some cases novelty, of the novelty generators. I suggest that we can make progress by continuing to develop a theory of initial, evolved, and still evolving "construction kits" (including scaffolding mechanisms).

Some ideas about this are being developed here:
http://www.cs.bham.ac.uk/research/projects/cogaff/misc/construction-kits.html
I suspect many researchers are now working on aspects of this in various disciplines and sub-disciplines, without communicating their problems, ideas about solutions, and results already achieved.
Putting the pieces together should accelerate the process of identifying and filling gaps in our knowledge, especially knowledge about how new varieties of information, along with increasingly complex required processing mechanisms, continually emerged in this planet's history.
Some ideas about evolution and uses of construction kits by evolution are explored in:
http://www.cs.bham.ac.uk/research/projects/cogaff/misc/construction-kits.html
As natural selection produced richer mechanisms, one side effect was expanding the mechanisms of evolution through processes like symbiosis, mate selection, and recent uses of new biological information to influence reproductive success.

Another side-effect, still in its early stages but accelerating, is expanding scientific knowledge about evolution, how it works, what it has achieved and what it can do.

This knowledge can, in turn, influence evolution by producing new side-effects through new knowledge of how to control the mechanisms, and increased understanding of requirements for replication of biological functionality in new physical/chemical media: e.g. in future intelligent machinery able to do more of what could previously be done only by direct products biological evolution. This proposal for study of a "self-informing" universe will help students and researchers in a wide variety of disciplines to understand the common origins in that "natural" process, and the consequences of that deep commonality.

It will produce new kinds of undergraduate and postgraduate degree revealing a previously mostly unnoticed common background to a wide range of existing disciplines helping to link separate disciplines, by providing new theoretical ideas and problems that can be developed in studies (and projects) relating to those disciplines after at least two years of study of the common background in evolved forms of information processing by which, in a sense, the universe has steadily become increasingly aware of itself.

In other words:
The proposal is based around the deep relevance of information to all aspects of life on earth (long before the development of current information processing technologies). In particular:

Jane Austen's use of the word "information" can be contrasted with what Claude Shannon (unfortunately) labelled "information" a century later
http://www.cs.bham.ac.uk/research/projects/cogaff/misc/austen-info.html

He used the word to refer not to what we would call information content, or meaning, but to properties and relationships of various information bearers including switches, transistors, holes in punched card or tape, and DNA sequences. Shannon's usage confused many admirers into thinking he had provided a new way to measure what is communicated or represented, though he was not confused. He understood the difference between measuring properties of information vehicles and properties of information content.

The information contents relevant to or produced by living things is the primary focus of this proposal, though the many and varied vehicles are also important, including those used within organisms during, acquisition, storage, inference, learning, and use of information of many kinds, in organisms of many kinds at different stages of evolution and development.

This theme can make use of human engineered tools and techniques for processing information, as in the proposed "Digital" theme, but it's main focus will be on naturally occurring varieties of information and information-processing.

I have called the new proposal "The Self-Informing Universe", since it is about the evolution of information *content* and information processing *mechanisms* in the universe.

For example, our planet began life (in both senses) as a purely physical/chemical portion of a physical universe, but gradually added more and more sophisticated information processing systems producing and using increasingly complex varieties of information, all ultimately resulting from evolution by natural selection, but using different mechanisms and different types of information at different times.

More and more complex forms of life gradually emerged, using increasingly complex forms of information for many purposes, initially concerned with controlling relatively simple physical and chemical processes for self-maintenance, then increasingly using information for action control, control of use of internal resources, reproduction, growth, collaboration, etc., while concurrently developing increasingly complex information processing mechanisms, starting with chemical mechanisms, followed by a vast number of newly evolved mechanisms, including eventually nervous systems of steadily increasing complexity and diversity of function. Later still social/cultural information processing mechanisms emerged, including, initially chemical signalling systems and pheromone trails. (Stigmergy.)

(A huge collection of examples could be given.)

The Self-Informing Universe at Birmingham. This new educational project will need to *subsume* (i.e. not merely *serve*) a wide range of disciplines.

In order to develop this proposal, with a broader collection of examples and references, I'll try to consult a sample of known (or recommended) colleagues, at least in computer science/AI, physics, mathematics, linguistics, philosophy, biosciences, chemistry, biochemistry, chemical engineering, computational biology, cancer research, sportex, education, economics, psychology, neuroscience, and possibly other disciplines.

The key idea is that all such disciplines have a peep-hole view of a vast, constantly growing, self-constructing, self-modifying, collection of interacting control systems, and side effects and by-products of those systems, which, like systems studied in the physical sciences, depend on matter and energy, but *also* use and create ever-expanding mechanisms and resources for processing information -- initially for control, then later for a much wider range of purposes.

Looping in on itself that process also produces new mechanisms and methodologies, for investigating and using information processing systems.

A tentative conclusion is that attempts to understand such systems, or their products, from a single viewpoint can lead to over-simplified and possibly dangerously incorrect conclusions. Yet that is exactly what most disciplines do.

Education is a spectacular example since sophisticated learning and developing organisms are studied by people who could not design, build or debug a working learning system. But many other disciplines have the same sort of limitation, at present. Fortunately, evolution seems to have produced some great teachers who don't need to have a scientific understanding of what they are doing, and some of them can pass on their gifts to help other teachers. But that does not imply that they know what they are doing. Similarly adults who provide the environment in which children develop sophisticated linguistic competences don't need to know much about brain mechanisms required for language use.

But those rich intuitive competences do not provide a basis for a scientific understanding of what is going on when children develop, or fail to develop, their own linguistic competences.

Only a few disciplines have begun to realise that they are studying complex information processing systems that demand entirely new research techniques, concepts and theories. By 2026 it is likely that this realisation will have spread to more disciplines. But they will not all be able to teach separately what their students need to be able think about.

When the need for a multi-disciplinary education concerned with a self-informing universe is more widely accepted, the teaching and learning will need to make use of powerful new technologies (the topic of the Birmingham Digital Proposal) without primarily being *about* the new technology or its uses.

Some of this has already begun in interdisciplinary educational programmes elsewhere (e.g. MIT? others?).

There are already influential thinkers implicitly or explicitly contributing to this project, including Margaret Boden, our recent distinguished lecturer, Daniel Dennett, and David Deutsch.

I suspect that if Alan Turing had not died two years after publishing his work on Chemistry-based Morphogenesis (in 1952, now receiving attention from a biologists, physicists and mathematicians) he would have gone on to work on Meta-Morphogenesis: another possible label for a Self-informing universe.

===== Aaron Sloman
Web: http://www.cs.bham.ac.uk/~axs
Research: http://goo.gl/9eN8Ks
Honorary professor of Artificial Intelligence and Cognitive Science
School of Computer Science

REFERENCES AND LINKS TO BE ADDED