School of Computer Science THE UNIVERSITY OF BIRMINGHAM

REALISATION AND VIRTUAL MACHINES
Aaron Sloman


Installed: 15 Dec 2008
Last updated: 30 Dec 2008; 6 Feb 2012
This paper is
http://www.cs.bham.ac.uk/research/projects/cogaff/misc/clark-dennett-realisation.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


Background

What follows is a slightly edited version of a letter sent to the New Scientist on 15 Dec 2008. (It was not published.)

The letter was provoked by my reading a letter to the New Scientist by Andy Clark
and Daniel Dennett, published on 13th December, and available online here.

That letter was a response to this letter by Mario Beauregard and Jeffrey M. Schwartz,
which, in turn was a reaction to an article by Amanda Gefter, published on
22 October 2008, entitled 'Creationists declare war over the brain'.


The Letter

From: Aaron Sloman
To:   New Scientist
Date: 15 Dec 2008

Dear New Scientist,

In the letters page of New Scientist 13th Dec., Andy Clark and
Daniel Dennett criticise the leap from the existence of non-material
causes (e.g. in minds) to religious or anti-scientific dualist
conclusions. They rightly challenge the apparent assumption by Mario
Beauregard and Jeffrey M. Schwartz that thoughts, attending and
mental activity are not realised in material substance.

Unfortunately, for most people the notion of mental phenomena being
"realised" in material substance is incomprehensible -- as much
mumbo-jumbo as talk of spiritual substances is to materialists.

In my experience it is very hard to get people who lack a deep
appreciation of important developments in computer science and
engineering in the last half century to understand how such
realisation is possible.

A first step is to make them pay attention to the non-material
processes that they interact with in their computers every day, e.g.
playing a computer game, or using a word-processor running in the
computer with the mysterious ability to manipulate fonts,
characters, words, line-lengths and paragraph formats, and even
correct spelling mistakes, despite the fact that opening a computer
and inspecting it with the most powerful microscopes, physical
measuring devices or chemical tests, reveals only physical
mechanisms, and no fonts, characters, words, paragraphs, etc., and
not even numbers, calculations, or programs.

What is not generally understood is that as a result of very complex
collaborations between scientists, engineers, mathematicians and
logicians it has been possible to create physical machines of a very
special kind, namely with the ability to support the operation of
what have misleadingly been labelled 'virtual machines' running in
them and performing a very wide variety of non-physical tasks.

The label 'virtual' is misleading because these are unlike 'virtual
reality' systems that merely represent other things that do not
exist (e.g. non-existent people walking around on non-existent
planets). In contrast entities, events and processes in virtual
machines not only exist but can cause events and processes both in
virtual machines and in physical machines -- for instance when the
actions of a spelling checker cause the physical patterns on your
computer screen to change because an error has been corrected.

All this is possible only because of a host of developments over
several decades, involving electronic devices of many kinds with
complex interfaces between them, along with software and hardware
components performing memory management tasks, interrupt generation
and handling hardware and software, device drivers, operating
systems, compilers, interpreters, filesystem managers, garbage
collectors, network protocol handlers, and many more. A key feature
of all this design work is to ensure the simultaneous truth of a
complex network of conditional statements relating what would happen
if so and so occurred in physical or virtual machines.

The support for that network of truths is equivalent to support for
a complex web of causal connections.

As a result of all this, multiple virtual machines of different
kinds can coexist and interact in a single physical computer, even a
computer with only one CPU.

Virtual machines were originally designed to make the processes of
design, development, modification, extension of complex systems
easier for human engineers, because of the intractable complexity of
performing those tasks by thinking about all the physical structures
and processes involved (millions of transistors changing their state
millions of times a second, and even more electrons flying around
the connectors).

What is now clear is that the use of virtual machines, can, for the
same reasons, make it much easier for machines to monitor and
control themselves. If they had to observe and modify the operations
of all the millions of rapidly changing physical components the task
would be impossible. Instead, the controlling virtual machines
monitor and modulate behaviours of virtual machines, or components
of virtual machines. This can also enable some computing systems to
discover and report faults, detect intruders, and also learn and
improve their performance, e.g. by altering their preferences.

It seems that biological evolution 'discovered' the importance of
virtual machines long before we did and produced far more
sophisticated self-monitoring and self-modifying virtual machines
than any we have created so far. The empirical work of Beauregard
and Schwartz (among others) helps to shed light on some of the
detailed ways the biological virtual machines operate. The
intractable puzzles of philosophers show some of the limitations of
those biological virtual machines.

Unfortunately most philosophers (and also psychologists and
neuroscientists) who discuss mind-brain relations, supervenience and
implementation (unlike Dennett and Clark) are largely ignorant of
the existence and capabilities of virtual machines in computers and
completely ignore them when discussing and writing about mind-brain
relations, even though they use several interacting virtual machines
on their computers, while writing their papers, checking and sending
email, looking up references using web browsers, etc. To help people
who want to understand, I have recently produced an online tutorial
on 'Virtual Machines in Philosophy, Engineering and Biology',
expanding some of these points, based on a recent conference
presentation, here: http://www.slideshare.net/asloman/slideshows

It is also available here:
http://www.cs.bham.ac.uk/research/projects/cogaff/talks/#wpe08


Yours

Aaron Sloman
Honorary Professor of Artificial Intelligence and Cognitive Science
University of Birmingham
http://www.cs.bham.ac.uk/~axs/
http://www.cs.bham.ac.uk/research/projects/cogaff/talks/

Related


Maintained by Aaron Sloman
School of Computer Science
The University of Birmingham