17th
International Joint Conference on Artificial Intelligence
4th - 10th August, 2001. Seattle, Washington, USA
Tutorial Programme
TUTORIAL SP3 Sunday 5th August 2:00 -- 6:00 pm
Philosophical Foundations: Some Key Questions
PRESENTED BY
Aaron Sloman
and Matthias Scheutz
School of Computer Science
The University of Birmingham
Note: Matthias Scheutz has returned to the
University of Notre Dame. See:
http://www.cse.nd.edu/~mscheutz/
Matthias Scheutz
(http://www.cse.nd.edu/~mscheutz/)
This tutorial, presented by tutors with deep knowledge and experience both in philosophy and in AI, will introduce a range of philosophical questions relevant to the goals and methodology of AI as science and AI as engineering, including the contribution of AI to the study of mind, and some unresolved questions about the nature of computational systems and how they relate to physical systems.
The mode of presentation will be a mixture of lectures and interactive discussions.
The questions to be discussed include a selection from the following:
The tutorial can be viewed as a sequel to the Philosophical Encounter Session at IJCAI-95, organised by Aaron Sloman and presented jointly with Marvin Minsky and John McCarthy.
The slides were prepared in a hurry to meet the printing deadline, and were re-organised and improved for the final presentation. We prepared far more slides than there was time to go through at the tutorial. So a subset was chosen for presentation on the day. The slides may be downloaded from here (pointers below).
Some of the
themes are developed in more recent slide presentations in Aaron
Sloman's "talks" directory:
http://www.cs.bham.ac.uk/~axs/misc/talks/
especially the talks on
A related book now available online:
Aaron Sloman's 1978 book, The Computer Revolution in Philosophy
has been
out of print for many years, but is now available free of charge online,
at
http://www.cs.bham.ac.uk/research/cogaff/crp/.
The tutorial will introduce some old philosophical concepts along with the problems they generate, and attempt to show how these problems take on a new form in the light of developments in Computer Science, Software Engineering, Artificial Intelligence and Neuroscience.
The tutorial will begin with a brief overview of some of central problems in:
o Metaphysics (the study of what kinds of things exist (ontology), including questions about the nature of causation and the relations between mind and matter)
o Epistemology (the study of what kinds of things can be known)
o Philosophy of mind (a more detailed study of what minds are and how are they related to brains, or computers)
o Philosophy of computing
An introduction to ontology will include the idea of "comparative descriptive ontology": the study of different sorts of ontologies to be found in different sorts of organisms, different sorts of humans (e.g. children, adults, humans from different cultures) and ontologies that might be useful for various kinds of machines.
The idea of an architecture-based ontology will be presented, as an extension of the method of conceptual analysis in philosophy.
This will be illustrated by outlining some ways in which developments in computing have led to developments in our own ontology, in particular the recent extension of our scientific ontology to include information processing machines, including virtual information processing machines, on the basis of which our ontology has been extended with a host of new concepts referring to objects, events and processes that can occur in such machines (e.g. procedures, closures, memory management, garbage collection, circular lists, sub-symbolic computation, etc.)
There are a number of deep problems about the nature of such machines, what sorts of existence they have, whether they can have causal powers, and how they are related to the underlying physical machines.
E.g. when we talk about a lisp virtual machine running in a computer are we really talking in an obscure way about *physical* processes occurring in the electronic devices in the machine, or are there also non-physical entities, events and processes involving such things as numbers, strings, lists, trees, graphs, procedures, procedure activations, etc.?
Can events in these virtual machines have effects, including physical effects, and if so is that consistent with the assumption that physical reality is a causally closed system? This will have to be discussed in the light of hard philosophical problems about the nature of causation, and how causation is related to counterfactual conditionals.
The intuitive notions of "implementation", or "realisation", employed by software engineers and computer scientists will be analysed and compared with the notion of "supervenience" which is used by philosophers to discuss the relationship between mental and physical phenomena.
This in turn will lead to questions about whether minds might be implemented in computers, in the context of the more general question: What sorts of minds can be implemented in what sorts of physical machines?
Detailed investigations of such questions will include analysis of a range of mental concepts including "experience", "feeling", "perceiving", "motivation", "emotion", "mood". We shall try to show how relating these old concepts to more refined new concepts derived from the capabilities of different sorts of virtual machine architectures, can provide a basis for answering questions such as: which sorts of emotions are possible in which sorts of machines? For example we can distinguish a range of increasingly sophisticated types of emotions (primary, secondary, tertiary, ...) and show how they presuppose increasingly complex information processing architectures.
More detailed discussions will lead on to analysis of familiar AI questions such as what representations are, how many varieties there are, which sorts of architectures can support them, etc.
If there is time, all this will be related to questions about the evolution of mind and whether minds of the sorts discussed in the tutorial could be produced by biological evolution.
The presentation will be highly interactive, with active encouragement of audience participation.
It is not expected that in the time available full coverage of all these topics will be achieved. However it is hoped that participants will acquire some understanding of problems like these and how to discuss them, e.g. with philosophers critical of AI.
The potential target audience:
The tutorial will relate to the following objectives listed in the call for tutorials
Research Fellow in AI
School of Computer Science,
The University of Birmingham,
Birmingham,
B15 2TT, UK
Phone: +44-121-414-2792 Fax: +44-121-414-4281
EMAIL M.Scheutz@cs.bham.ac.uk
CV is below.
Rhodes Scholar at Balliol College Oxford 1957-1960, Senior Scholar at St Antony's College 1960-62. GEC Professorial Fellow 1984-6. Elected Fellow of the American Association for AI in 1991. Elected honorary life Fellow of AISB in 1997. Elected Fellow of ECCAI in 1999. Regularly invited as keynote speaker at seminars, international conferences and workshops.
Taught philosophy at Hull then Sussex University for several years, but decided around 1970 that AI provided the best context for studying many old philosophical problems. Did battle with McCarthy and Hayes at IJCAI in 1971 (paper on non-logical representations reprinted in AI Journal 1971). After a year learning about AI in Edinburgh in 1972-3 as SRC senior visiting fellow returned to Sussex and later helped to found the School of Cognitive and Computing Sciences, and wrote an influential book ``The Computer Revolution in Philosophy'' (1978).
Continued working on philosophy, on the architecture of visual systems, and tools for teaching and research in AI. Managed the development of Poplog, a sophisticated multi-language AI development environment (sold by ISL for several years, but now available free of charge, though still used in commercial products, e.g. Clementine.)
After 27 years at Sussex wanted a change. In 1991 moved to the School of Computer Science in the University of Birmingham, where he also collaborates with members of the School of Psychology. Still working on AI tools (e.g. the SIM_AGENT toolkit), hybrid architectures for intelligent agents, the evolution of consciousness, architectural support for motivation and emotions, issues concerned with representation, mathematical thinking, vision, and related philosophical problems in AI and computing.
Most of my recent papers can be found in the Cognition and Affect project directory at Birmingham http://www.cs.bham.ac.uk/research/cogaff/
My software tools and much AI teaching material are accessible from
the Free Poplog site:
http://www.cs.bham.ac.uk/research/poplog/freepoplog.html
including the SimAgent toolkit
http://www.cs.bham.ac.uk/~axs/cogaff/simagent.html
Ph.D., Philosophy University of Vienna (Vienna), 1995
Leverhulme funded Research Fellow in AI, School of Computer Science, The University of Birmingham. Project title: Evolvable virtual information processing architectures for human-like minds}
Previously:
Visiting Assistant Professor in the Department of Computer
Science and Engineering, University of Notre Dame
(Returning to permanent post August 2001).
Lecturer in the Department of Philosophy of Science and Social Studies of Science, University of Vienna
Organizer of the conference and workshop NTCS'99 on "Computationalism - The Next Generation" (see http://www.univie.ac.at/cognition/conf/ntcs99/)
Co-founder of the Austrian Society for Cognitive Science (see http://www.univie.ac.at/cognition/)
Current main project at the University Birmingham: Evolvable architectures for human-like minds
An introductary overview of the Cognition and Affect project is here: http://www.cs.bham.ac.uk/~axs/cogaff.html
The Cogaff paper repository is here: http://www.cs.bham.ac.uk/research/cogaff/