A mixed bag of papers on artificial
life, some previously published elsewhere, some written especially
for the book. Like all collections of this kind, it is difficult to
maintain a coherent story: some areas are ignored or assumed, others get
repeated in many chapters, and the level of treatment varies. But the
introduction is excellent, and some of the papers are classics.
Naturally, these papers stress the overriding importance of evolution,
genotype and phenotype for A-Life. One other recurring thread is that much
of the complexity of an organism's behaviour is due to the complexity of
its environment, and that evolution is open-ended and that
emergent properties occur only because the environment is open-ended.
Hence artificial computer-based environments must be open-ended; the big
question is, is that possible? If it is, computer-based A-Life could be
possible, if not, A-Life must be embedded in robot bodies in the physical
world.
The papers are gathered into five parts:
background and introductory material
specific examples of A-Life research
specific explanatory strategies
views of past philosophers on life in general
functionalism, and 'strong' A-Life
The part I found least satisfying was how the views of past philosophers
relate to A-Life. Personally, I find that I don't really care where Great
Names were close to current thinking, and where they were way off beam --
I just want to know what are the currently best available thoughts and
ideas. But this is only a small part of the total collection, and each
with their different preferences should enjoy some part of the whole.
Contents
- Introduction to The Philosophy of Artificial Life. 1996
-
The
central concept of A-Life, excepting life itself, is self-organization.
Self-organization involves the emergence (and maintenance) of
order, or complexity, out of an origin that is ordered to a lesser
degree. That is, it concerns not mere superficial change, but
fundamental structural development. This development is
'spontaneous', or 'autonomous', following from the intrinsic
character of the system itself (often, in interaction with the
environment) instead of being imposed on the system by some
external designer.
The
behaviour of flocks of birds, for instance, must be described on
its own level, although it results from the behaviour of
individual birds. [emergent properties
are] not expressible in terms
of --- though explicable by means of ---
[the lower-level properties]
Species
can evolve more quickly in the presence of predators because the
predator displaces prey from sub-optimal ... local maxima.
- Christopher G. Langton. Artificial Life. 1996
- [An updated version of his paper in Artificial Life I.]
Background, history and overview of the subject area, with a
whistle-stop tour of 18th century mechanical automata, von Neumann's
self-reproducing cellular automaton, Lindenmayer systems, flocking
'boids', genetic algorithms, Dawkins'
biomorphs, the 'tit for tat' strategy in Iterated Prisoner's Dilemma,
and Ray's Tierra.
- Autonomy and Artificiality. 1996
- The implications of A-Life for our understanding of human
autonomy, or freedom.
[the artificial
sciences] help us to see how
autonomous behaviour ... is possible, and to appreciate the
awesome complexity of much human choice.
- Thomas S. Ray. An Approach to the Synthesis of Life. 1992
- A description of his virtual Tierra
environment, for studying artificial open-ended evolution
- Richard M. Burian, Robert C. Richardson. Form and Order in Evolutionary Biology. 1996
- A critique of Stuart Kauffman's
The Origins of Order
- John Maynard Smith. Evolution --- Natural and Artificial. 1996
- A short overview of the current state of evolutionary biology
- David J. McFarland. Animals as Cost-based Robots. 1996
- An analysis of animal behaviour in terms of cost functions, with
the aim of making robots more animal-like:
(1) animal
behaviour is governed by mechanisms no different in principle from
those that could be put in a robot; (2) these mechanisms are
optimized (through evolution) with respect to the real costs ...
of the animal's ecological niche; (3) similar design principles
could be used in robotics
- Michael Wheeler. From Robots to Rothko: The Bringing Forth of Worlds. 1996
- An analysis of animal behaviour in terms of cost functions, with
the aim of making robots more animal-like:
(1) animal
behaviour is governed by mechanisms no different in principle from
those that could be put in a robot; (2) these mechanisms are
optimized (through evolution) with respect to the real costs ...
of the animal's ecological niche; (3) similar design principles
could be used in robotics
- David Kirsh. Today the Earwig, Tomorrow Man?. 1996
- A critique of 'moboticist' Rod Brooks'
position that "97 per cent of human activity is concept free,
driven by control mechanisms we share … with insects"
- Andy Clark. Happy Couplings: Emergence and Explanatory Interlock. 1996
- Using dynamical systems theory as a means of understanding
emergent properties.
As the
complexities of interaction between parts increases, so the
explanatory burden increasingly falls not on the parts but on
their organisation.
- Horst Hendricks-Jansen. In Praise of Interactive Emergence, Or Why Explanations Don't Have to Wait for Implementation. 1996
- Using situated robotics to help explain human intentional behaviour and thought
- Gareth B. Matthews. Aristotle on Life. 1996
- Peter Godfrey-Smith. Spencer and Dewey on Life and Mind. 1996
- Mark A. Bedau. The Nature of Life. 1996
- The offer of 'supple adaptation' as the essential principle of
life, and a statistical measure of the supple adaptation of
a population, which can be used to determine whether it is alive.
The
essential principle that explains the unified diversity of life
seems to be this suppleness of the adaptive processes
its unending capacity to produce novel solutions to unanticipated
changes in the problems of surviving, reproducing, or, more
generally, flourishing. ... supple adaptation involves responding
appropriately in an indefinite variety of ways to an unpredictable
variety of contingencies. ...
A changing
environment (or fitness function) is also a crucial part of supple
evolution:
When
selection is based on a fixed fitness function, the resulting
adaptive dynamics eventually stabilize rather than continually
produce adaptive novelty.
- Elliott Sober. Learning from Functionalism--Prospects for Strong Artificial Life. 1992
- Howard H. Pattee. Simulations, Realizations, and Theories of Life. 1989
-
Many of the
controversies in AI result from the multiple use of computation as
a conceptual theory, as an empirical tool, as simulation, and as
realization of thought. AL models will have to make these
distinctions.
... it was
also clear that the environment was too simple to produce
interesting emergent behaviour.