Digital Creativity
2003, Vol. 14, No. 1, pp. 43–53
1462-6268/03/1401-0043$16.00
© Swets & Zeitlinger
Morphogenetics: generative
processes in the work of
Driessens and Verstappen
Mitchell Whitelaw
University of Canberra, Australia
mitchellw@comedu.canberra.edu.au
1 Introduction
Over more than a decade, artificial life has
provided a rich source of techniques,
architectures and metaphors for artists working
with generative computational processes. In fact
a-life art has come to be recognised as a distinct
sub-genre within the wider electronic or ‘new
media’ art world, as evidenced by, for example,
the annual Life x.0 competition (Fundación
Telefónica 2002). This category encompasses a
diverse range of practices, approaches and
agendas. Some artists have taken on a-life as
both a technique and a project, and set out to
realise its aims, in their strongest form, in their
own practice. Many more have adopted a-life
techniques while drawing and elaborating on
the metaphors built in to those techniques;
more rarely, artists apply those techniques
critically and reflexively, asking about their
wider meanings and implications. A-life art, as a
field, both adopts and adapts artificial life, in
parallel with Christopher Langton’s exhortation
to explore “life as it could be” (Langton 1992).
However one of the key problems for alife
art is the way in which any sense of “life as it
could be” is constrained through its inscription
into relatively fixed formal architectures and
templates. Forms such as artificial evolution
(genetic algorithms), agent-based ecosystem
simulations and cellular automata are often
taken up unaltered in a-life art practice. They
are recast, reclothed, couched in new agendas
and contexts, but the core formal mechanisms
remain. So too do the ideas of life embedded in
those techniques — particular notions of
heredity, evolution, morphogenesis, genetics,
agency, interaction, determinism and contin-
gency.
Abstract
Dutch artists Erwin Driessens and Maria Verstappen
work with generative techniques in a range of media,
from digital imaging and software to sculpture
and robotics. In their own words, they pursue “an
activity that explores the unseen, the unthought and
the unknown?”To this end they create rich, elegant,
self-constraining generative systems, which draw
on, and re-engineer, techniques from the field of
artificial life. This paper sets out a critical survey of
the artists’ generative work, and shows how their
application of a-life techniques destabilises, and
enriches, some of the problematic aspects of those
techniques. Specifically, where a-life frequently ignores
complex morphogenetic processes,
dematerialising them into a formal and instantaneous
moment of genetic expression, the artists demonstrate
the possibility and potential of richer, more
complex and more ‘materialised’ models.
Keywords: artificial life, cellular automata, generative
techniques, genetic algorithms, morphology.
DigitalCreativity,Vol.14,No.1
Whitelaw
44
Dutch artists Erwin Driessens and Maria
Verstappen take an unusual and instructive
approach to the generative techniques of
artificial life (Driessens and Verstappen 2002).
They apply central a-life techniques — artificial
evolution and cellular automata — though they
leave the biological analogies involved in those
techniques aside. Further, these generative forms
have been rewired and rethought; formal and
conceptual structures have been inverted,
bypassed or collapsed. These processes suggest
alternative morphogeneses, notions of genetic
expression and developmental time; above all,
though, they are clever, elegant and richly
generative.
2 Breed
Breed (1995–2000) is a project which spans
both process and product, comprising generative
morphological software and the virtual threedimensional
forms which that software generates,
as well as physical fabrications of those
forms. As the name suggests the project involves
a form of artificial evolution, though one quite
unlike that used in other ‘breeders’ of aesthetic
forms (such as those of Karl Sims (2002),
William Latham (Todd and Latham 1992) or
Steven Rooke (2002)).
Breed generates intricate three-dimensional
forms through a stepwise process of
spatial differentiation which the artists describe
as ‘cellular’. The process begins with a solid,
cubic volume — a single cell. When the cell
‘divides’, its volume is partitioned into eight
smaller cubic cells, units of space which may be
either solid or empty; so the space occupied by
the initial cube is coarsely differentiated — a
chunky, blocky assemblage is formed. The same
process continues for each of the eight new
cellular units, which divide in turn into eight,
and the resultant form is again more detailed
and differentiated. This process continues, being
applied to rapidly increasing numbers of smaller
and smaller cells. Viewed as an animated
process, the initial cube carves itself away in
ever-finer cellular chunks, finally resembling a
complex sculpture assembled from cubic Lego
blocks — a ‘pixelated’ mass perforated with
irregular hollows and voids.
This morphogenetic process is controlled
by a set of parameters which, as in other
breeders, are treated as the form’s genome. These
parameters are in fact simple rules governing the
process of cellular differentiation. At each step
in the process, a cell’s subsequent differentiation
is controlled by the presence (or absence) of
neighbouring cells; the morphogenetic rules
simply dictate how every possible combination
of present or absent neighbours influences the
next split. This is a highly elegant form of
artificial genetics, one in which a compact
‘genome’ generates a complex form through
recursive application at ever-finer scales, rather
than through a high-level or global specification.
Of course these parameters still entirely determine
the resultant form — a given genome will
give rise to the same form every time — but the
process of ‘expression’, the transition from code
to form, is tightly bound to the phenotypic
context of the virtual form. The formation of a
certain void or bump at a certain level of detail
depends not on an explicit representation
encoded in the genome, but on the geneticallyspecified
interaction of neighbouring units of
volume. In an architecture which follows a-life’s
‘bottom-up’ maxim, a set of simple, micro-scale
rules gives rise to a complex macro-scale result.
The evolutionary process in Breed occurs
as a form generated by a random genome is
evaluated according to a set of spatial criteria.
An automatic measurement is made of properties
such as volume, surface area and connectivity
— the degree to which different parts of the
form are joined to each other. These values are
stored, and the genome for that form is randomly
altered. A new form is generated, and its
fitness values measured and compared with
those of the initial form — if it’s fitter, the new
form is retained, if not, the initial form is
45
DigitalCreativity,Vol.14,No.1
Morphogenetics
Figure 1.
Erwin
Driessens and
Maria
Verstappen,
Breed 1.1
(1999),
detail. Image
courtesy of
the artists.
mutated again. Through repetition of this
automatic loop, a form will eventually be
generated which has ‘maximal fitness’, meeting
those predetermined formal criteria. While this
process gives the impression of a highly linear,
progressive drive towards an optimal ideal, this
is not the case: the relatively open nature of the
criteria (volume, surface area and connectivity)
mean that any single set can be met by a large
number of different forms. In another elegant
twist, the structure of the artificial evolutionary
process mirrors that of the morphogenetic
process: rather than searching for a single,
absolute goal, this simple stepwise evolution uses
only a local comparison; as it forms a sequence
of incrementally ‘better’ forms, the process
effectively paints itself into a corner: the final
‘optimal’ form is in fact only the most optimal
form that a specific sequence of random
alterations had produced.
This represents an interesting inversion
of an a-life convention. In the jargon of artificial
evolution, a range of possible forms for a given
system is sometimes described in terms of a
‘fitness landscape": imagined initially as a flat,
two-dimensional plane, different areas can be
assigned a ‘height’ which corresponds to the
fitness of the forms at that point. The flat plane
becomes a hilly landscape, one that may have
numerous individual peaks of various heights,
separated by valleys. In utilitarian applications
of genetic algorithms, the central aim to arrive at
a solution which has maximal fitness — which
occupies the highest peak on that hilly landscape.
One of the main problems for these
systems is that without exhaustively searching
the space of possible forms — and such spaces
are often very large — it may be impossible to
tell whether a certain evolved solution sits on
the highest peak in the entire space, or only on a
medium-sized foothill. The random mutations
of artificial evolution can be thought of as
exploring the area around a certain solution —
its locality, in this imaginary landscape. If a
process searching for ever-better solutions, like
the one in Breed, finds itself on a local fitness
maximum — the highest hill in the immediate
neighbourhood — it will stop. In all likelihood,
subsequent mutations will result in forms with a
lower fitness value, and since this process cannot
‘climb down’, it will stay stuck at that hilltop. In
utilitarian applications of genetic algorithms,
the phenomenon of evolution halting at a ‘local
maximum’ is a problem (see for example
Flyvbjerg and Lautrup (1992), or for a more
general discussion Kauffman (1993)). In Breed
Driessens and Verstappen turn this problem
into a virtue, for it is in finding local maxima —
backing itself inexorably up the nearest hill —
that this automated evolutionary process can
produce a variety of forms which meet these
open spatial criteria. Rather than defining a
fixed formal or aesthetic optimum, the artists
frame the fitness criteria in a way that involves a
kind of loose control over the results, where the
details of the evolved forms are unguided even
as global attributes such as volume and surface
area are firmly specified.
DigitalCreativity,Vol.14,No.1
Whitelaw
46
Driessens and Verstappen have brought
this project to a sculptural conclusion by
fabricating a selection of the evolved forms.
Originally the artists used plywood, which was
hand-cut and assembled into forms with 32 0.6centimetre
voxels to a side. More recently, they
have used a computer-controlled rapid
prototyping process (selective laser sintering) to
generate forms which are more delicate and
intricate, occupying 10cm. cubes with 64 voxels
per side (Figure 1). Even here, the resolution of
the forms is limited by the physical realisation:
as Verstappen (2002) writes,
it would be nice if you could hardly see the voxel
elements, to see that come parts really get an
organic structure like coral or broccoli.
3 Tuboid
Another experiment in computational morphogenesis,
and a similar transition from the
immaterial to the material occurs in Tuboid
(1998–99). Here, the spatial template is a tube,
rather than a cube, though as in Breed it is a
form that shapes its own development through
space and time using a simple artificial evolutionary
process. Here, a worm-like shape is
formed from a sequence of cross-sections — two
dimensional slices — which accumulate over
time. The structure which underpins those slices
is ingenious and elaborate. Each slice is defined
by a group of up to 256 articulated spokes; each
spoke has four segments, which can vary in
length, and the outer three segments of each
spoke can also pivot through any angle. The
outline of each slice is formed by joining the
endpoints of these elastic, articulated spokes, as
illustrated in Figure 2. The configuration of the
spokes is defined in turn by a set of parameters
controlling the length of each spoke segment
and the rotational speed of the three pivoting
outer segments.
Of course the encoding of a parameter
for speed implies a structure that changes with
time — and this is where Tuboid is most
interesting. A sequence of slices accumulates
over time to define a three-dimensional form;
each new slice, however, is a mutation of the
previous one, so that the spoke lengths and
pivot rates will vary slightly from one slice to the
next. These values are only constrained by a
single criterion: that the outline of the slice must
not intersect itself; it must always describe the
perimeter of a single two-dimensional form.
This mutational sequence of profiles is projected
through a third dimension to define a single
smooth, wobbly, organomorphic form. The
complex inner architecture of articulated spokes,
and the encoding of rotational speed in the
genome, combine with this spatial constraint to
generate sequences (and thus tuboid forms) that
are highly coherent, with smoothly undulating
bulges and ridges. As in Breed, these forms are
not quite specified by the genome but rather
emerge from the interaction of that genome —
and in this case its mutational sequence — with
its past activity and with a spatial constraint.
Tuboid exists in both virtual and physical
manifestations. In virtual form, these tubes can
be viewed either from the ‘outside’, as solid
extrusions, or from within, generated in real
time as enfolding, continuously unfolding
tunnels (Figure 3). Once again, Driessens and
Verstappen have fabricated a selection of these
forms as physical objects — in this case weird,
slightly eerie, shiny white towers around a metre
high (Figure 4). These are built up from 4
Figure 2.
Diagram
showing the
articulated
spokes (in
grey) which
generate the
outline (in
black) of a
single layer
of a Tuboid
structure.
Image
courtesy of
the artists.
47
DigitalCreativity,Vol.14,No.1
Morphogenetics
millimetre slices of fibreboard, hand-smoothed
and sprayed with glossy automotive lacquer.
They almost resemble stalagmites cast in plastic,
but with their bulbous protrusions and rippling
ridges they also have more bodily connotations
— like tapering sections of intestinal tract.
4 Ima Traveller
Ima Traveller (1996-8) is another work which
uses a morphogenetic process derived from a-life
techniques, though where Breed and Tuboid are
both form-machines, Ima Traveller is an imagemachine.
Like Breed, Ima Traveller uses a virtual
‘cellular’ structure: initially a single cell — a
pixel at the centre of the screen — births new
cells, neighbours. These split in turn, and the
reproductive cycle continues until the screen is
quickly filled with proliferating masses of pixels.
As in Breed, a set of rules controls the way in
which each cell reproduces, and here too those
rules draw on each cell’s own state and its
current environment — its neighbours. In Breed
a cell could have only two states — on or off,
solid or void. In Ima Traveller however, each cell
is a pixel, a point of colour, and it can have one
of millions of possible colour states. In Breed the
process of cellular differentiation works inwards,
refining the original volume in ever-finer detail;
here, the cellular space itself grows, expanding
beyond the edge of the display; cells crowd each
other out in an ever-spreading pixelated mass.
The artists have engineered the splitting rules so
that the dividing cells form blossoming masses
of colour which differentiate endlessly, opening
up into ever-greater detail: the visual effect is of
a relentless zoom, a sense of diving into a
continuously-unfolding picture plane. The
nearest visual analogy is with the ‘fractal zoom’
— that computer-graphic cliché which tunnels
into the filigreed coastlines of a Mandelbrot set
image, revealing endless ever-tinier coastlines
and curling filaments. The graphic quality of
Traveller is less slick, more pointillistic —
suggesting coloured clouds or variegated mats of
lichen (Figure 5). The software interface also
allows this diving zoom to be steered with the
mouse, so a path can be woven through the
most interesting zones. The interactive pleasure
of this process is in the way these zones continually
explode, expand, differentiate and refine
themselves. An apparently uniform patch of skyblue
is soon peppered with darker specks, one of
which opens into an inky void, which in turn
lightens to a cloudy grey mass, which in turn
sprouts islands of green…. Emanuelle Lequeux
Figure 3. Erwin Driessens and Maria Verstappen, Tuboid
(1998-99), detail from real-time ‘tunnel’ mode.
Figure 4. Erwin Driessens and Maria Verstappen, Tuboid
(1998-99), detail showing three physical models. Image
courtesy of the artists.
DigitalCreativity,Vol.14,No.1
Whitelaw
48
responds with justified lyricism:
Images give birth to other images, pixels to other
pixels, without end. At each tremble of the mouse
another universe is born.
(Lequeux 1999)
5 Tickle and other generative pleasures
At first glance, Tickle (1995-2000) appears to
bear little relationship to the rest of Driessens
and Verstappen’s work. Tickle is a small, autonomous
robot: a blank aluminium box about the
size of a cigarette packet, it is fitted with a pair
of nubbed rubber caterpillar tracks (Figure 6).
True to its name, the robot’s purpose is to tickle;
with basic mechanical sensors and motors
powered by rechargeable batteries, it crawls over
a reclining body, reversing and pivoting to avoid
falling off and tracing a wandering, unpredictable
path. Based on the context of its documentation,
and on its presence in venues such as the
Life 2.0 competition for a-life art, we can infer
that Tickle is an art object; however at the same
time, Driessens and Verstappen allow it to be
interpreted as a purely utilitarian object, a pet
robot massager — perhaps even a vaguely erotic
plaything. Their web site discusses the prospect
of limited commercial production — an idea
that only arose, Verstappen says, in response to
the interest generated by online documentation
of the work. Interestingly, Verstappen also
reports that the artists have experimented with
automatic ticklers for some 13 years, trying out
other solutions such as a motorised blade of
grass suspended over the ticklee’s back. Tickle
refines a solution to this delicate sensory/spatial
task; as Verstappen (1999b) writes, “an important
aspect of good tickling is that it has to be
unpredictable”. While based on simple ‘finite
state’ mechanical sensors — sensorimotor
switches connecting the robot’s orientation to its
motion, designed solely to prevent it falling off
the subject’s body — Tickle’s behaviour is
complex enough to be (reportedly) pleasurable.
Still, the challenge of an automatic tickle
remains: at the time of writing, the artists were
working on a prototype for TickleSalon, a far
more elaborate robot which revisits earlier
tickling machines and suspends a small, moplike
stroking appendage over the subject’s back.
Four digitally-controlled stepper motors move
the mop gently but precisely; the system uses
tension detection to create a live, three-dimensional
map of the body surface which the robot
Figure 5.
Erwin
Driessens and
Maria
Verstappen,
Ima Traveller
(1996–8),
detail. Image
courtesy of
the artists.
49
DigitalCreativity,Vol.14,No.1
Morphogenetics
will use to refine its tickling trajectory.
Of course it is the pleasure of
unpredictability — as well as an underplayed
humour — which connects Tickle very clearly to
the duo’s other work. Once again, following alife’s
basic philosophy, a formal system with very
simple constraints gives rise to a complex
behavioural and phenomenal result. Tickle is a
system for delivering tactile pleasure — and in
the same way Ima Traveller is a kind of visual
massage, an endless, automated proliferation of
visual variety. A catalogue essay on Ima Traveller
quotes Verstappen:
even if a work is conceptual, it must be good, to
allow you to flee and to escape yourself. To
submerge you.
(Lequeaux 1999)
Perhaps what makes Driessens and Verstappen’s
work both appealing and elusive is its balance of
dry ‘conceptual’ constraint with the immersive
pleasure of morphogenic and behavioural
variety. There is a quite careful, deliberate
structure to the processes undertaken: constraints
are set, a minimal number of parameters
placed in variation. There is a setting-aside of
stereotypical artistic ‘creation’ in favour of a
metacreation which is more controlled, and
cooler, but which consequently addresses the
limits of a creative process more directly.
Driessens and Verstappen (1999) describe this
approach very clearly:
We see a challenge in the question ‘how can we
express the longing for an activity that explores
the unseen, the unthought and the unknown?’
Not influenced by taste, style and meaning but
also avoiding complete unpredictability.
In this context, we are interested in dynamic
processes: each change in a process is a reaction to
the previous change, so it is really matter of
feedback. On the one hand, this creates cohesion,
on the other hand the system remains unpredictable
as a result of coincidences. Small changes
can sometimes lead to dramatic transformations.
This quote emphasises an investigation
of formal mechanisms, and a straightforward,
blank approach to the ‘unknown’. It indicates a
paring-back of art practice to an impersonal,
systemic exploration which might be taken as a
naïve ignorance of the complexities of art’s social
and cultural involutions, its histories and
discourses, and its contemporary political
dynamics. However work such as Exhibition
(1990) strongly counters this impression. In this
multi-stage project, the duo firstly selected a set
of photographs of European and American
gallery spaces that had appeared in major art
magazines such as Flash Art and Kunstforum.
Based on those images, the artists created a set of
physical models of the gallery spaces (minus the
artworks) — detailed down to minute power
points. The spatial punch line is that only the
portion of the space visible in the photograph is
Figure 6.
Erwin
Driessens and
Maria
Verstappen,
Tickle
(1995–6),
image
courtesy of
the artists.
DigitalCreativity,Vol.14,No.1
Whitelaw
50
modelled: only the wedge-shaped section of the
room defined by the camera’s field of view exists.
As part of the project, the models were carefully
lit and re-photographed from the original
viewpoint, re-constituted as coherent spaces in
the form of an image. These re-made images
were then published in another art magazine
(Artscribe no. 84). The other half of the project
— Model spaces — involved an exhibition of the
maquettes themselves at the Museum Fodor,
Amsterdam. Here, they reveal themselves as
partial, three-dimensional reproductions —
spaces truncated by a field of view. Despite its
characteristic coolness, Exhibition takes on the
institutional mechanisms of contemporary art
practice very directly, pointing to the crucial
circuit between space and image, gallery and
magazine, work and reproduction, which
sustains the contemporary art world.
This project offers an interesting perspective
on the duo’s later work, and their use of
a-life morphological techniques. Verstappen
(1999a) describes the transition from Exhibition
into the later generative and morphological
works:
In ... works like Exhibition, we focused on the
preconditions of the art system. ... At that time
we felt ourselves so much confronted with [the
social-political] reality of the art practice, that
we had to deal with it in a very direct way. Then
we thought, if all these spaces have necessarily to
be filled up with art each month again and
again, we can think of [how to] automate this
production. Somehow very nihilistic in its
approach we started to make our first attempts to
make automatic artificial art. But very quickly
we understood this was not an easy job, and then
it became really interesting, an exciting adven-
ture.
Here, then, is the balance of generative
pleasure and dry wit: what began as an attempt
to satirise the art world’s endless appetite for
novelty, became an engaged investigation of the
generative processes that might supply such
novelty. Verstappen positions the artists’ practice
as post-Duchamp, post-readymade; she observes
that even after Duchamp’s work opened the way
for “an aesthetic interpretation of everything”,
subsequent creative practice has dealt only with
limited segments of this unimaginable ‘everything’
— “somehow they all end up defining
rules [for] how to interpret reality”. Of course a
framework, a set of constraints is essential, and
Verstappen does not suggest that this work aims
to somehow access this aesthetic “everything”
directly (as Steven Rooke’s work perhaps does).
Instead, it entails the systematic deployment of
frameworks and constraints, the testing of
different morphogenetic processes. By re-casting
these constraints as computational processes and
parameters, the artists make them transparent,
manipulable and malleable; the computer is a
kind of meta-framework, a framework for
generating and testing frameworks which in
turn pursue specific slices of this ‘everything’,
“the unseen, the unthought and the unknown”.
6 Tweaking artificial life: technique and
concept
Of central interest here is the way that Driessens
and Verstappen use a-life techniques towards
this end, and the way their practice reinterprets
and repositions those techniques. Their use of
artificial evolution is particularly interesting,
especially in comparison with the more generic
forms of ‘breeder’. The evolutionary processes in
Breed and Tuboid are hybrids which complicate
the usual transparency of the artificial genome,
linking its action closely to an ongoing
phenotypic process. The importance of morphogenetic
time in these works — of a sequence of
development — distinguishes them from the
instantaneous expressive processes of most
breeders. Further, where expression in those
systems occurs once, in a single act of translation
which specifies the entire phenotype, Driessens
and Verstappen show that it is possible to have
51
DigitalCreativity,Vol.14,No.1
Morphogenetics
expression occurs repeatedly and locally, so that
the artificial genome is not an explicit ‘blueprint’,
but one element in a more complex,
dynamic system. The use of a kind of threedimensional
cellular automaton in Breed is a
perfect example of this extra complexity: rather
than the gene acting on the form, this process
has the gene specifying the form’s ongoing
interactions with itself. The artists bend the
conventional model further in Tuboid, where the
resultant form is not so much a ‘phenotype’ as a
three-dimensional temporal record of a selfconstraining
evolutionary process. Rather than
springing fully-formed from a single inner
‘code’, these blobby towers are accretions of
slices in a sequence of mutated individuals —
more analogous to colony structures such as
stromatolites or coral reefs than individual
organisms. The clean, simplistic, a-temporal
notions of genetic expression which a-life
generally adopts are thoroughly complicated, as
is the taken-for-granted correspondence of code
with genome and form with organism.
Richard Dawkins’ Biomorph software,
the ancestor of all form-breeders, uses a digital
genome to generate forms made from linear
elements; Dawkins (1986) used the rich variety
of forms produced to support his arguments for
the power of Darwinian evolution in The blind
watchmaker. However Stuart Kauffman has
argued that
there’s less there than meets the eye … it’s clear
you can generate varieties of morphologies, if you
have something called a genotype that makes
something called a morphology. … The part I
tend to dislike in what he’s done is that there’s
nothing natural or self-organised or robust about
the development mechanisms and morphologies
that Richard posits. He simply has computer
programs arbitrarily draw stick figures or
whatever. That’s not how real development
works.
(Kauffman in Brockman 1996 72)
In the work of Driessens and Verstappen
we find formal models for morphogenesis which
are self-organised, and to that extent, perhaps
more ‘natural’ than those typically found in
form-breeders. While this artwork never claims
to mimic ‘how real development works’, it does
begin to suggest an artificial embryology;
simple, locally-interacting architectures which
unfold into rich forms and differentiated spaces,
and which tightly articulate the code of the
genotype, with the material specificity of the
phenotype.
The discourse and philosophy around
the techniques of artificial evolution is also quite
different here. In general, evolutionary artists
have emphasised generative potential — the
vastness of an image- or form-space — rather
than the constraints inherent in the language
which allows access to that space (see for
example Todd and Latham (1992)). The
constitutive structures of each proceduralgeometric
grammar or mathematical image
definition are largely ignored in the rhetoric
around these works — instead, the dominant
language of Darwinian evolution lends them an
air of grand totality. Moreover, the presence and
agency of a human aesthetic ‘selector’ is often a
crucial element in that rhetoric. Driessens and
Verstappen offer a striking contrast: here,
artificial evolutionary processes act primarily to
constrain morphological outcomes in a balance
between unpredictable novelty and spatial
coherence. As outlined above, the automated
evolution in Breed uses a simple, self-limiting
technique which, with successive runs, gives a
variety of results for a given set of criteria.
Rather than a desire-driven amplification of
creative agency, the artists deliver a ‘blind’
process, a quietly automated factory for novel
forms which meet a set of specific criteria. The
forms themselves are invested with no special
significance; they simply indicate that blank
automatic process. Other breeders tend to figure
evolution as akin to a manned spacecraft, a
propulsive process steered by human aesthetic
will which traces sweeping arcs through a vast
hyperspace. Here, by contrast, ‘evolution’ is set
DigitalCreativity,Vol.14,No.1
Whitelaw
52
up to self-organise, to coalesce and converge.
The artists also make some interesting
innovations in their use of cellular automata. As
they are conventionally deployed in a-life and its
popular rhetoric, cellular automata (CAs) are
understood as extensive matrices, spatial grids
which define an artificial universe. In the best
known cellular automaton, the Game of life,
coherent formations within the lattice of cells
are readily identified as autonomous, living
entities (see for example Emmeche (1994)).
‘Gliders’, for example, are formations which
traverse this grid on diagonal paths, continuing
endlessly unless they collide with another group
of cells. An entire catalogue of other characteristic
formations — native life-forms — have been
identified, such as the ‘blinker’, a small symmetrical
oscillating structure, and the ‘rock’, a
stable block of cells. Thus the Game is figured as
a terrain for life, a deterministic grid which,
remarkably, gives rise to a host of characteristic,
coherent, emergent temporal and spatial
patterns.
Driessens and Verstappen use CAs in
very different ways. Notably, they rupture the
continuity and constancy of the cellular grid
itself: they take the ‘cellular’ aspect of the CA
literally, and treat the cells as dynamic, divisible
entities rather than as symbolic elements
supporting transitory higher-level patterns. In
Breed and Ima Traveller, the cells which constitute
each structure continually divide, refining
their locality. The interesting patterns which
arise in both these works are marks of an
ongoing morphogenetic process rather than
discrete metaphorical ‘life form’s. In a sense
Driessens and Verstappen turn the cellular
automaton inside-out (or rather, outside-in): in
a conventional CA the emergent ‘life forms’
exist inside the fixed cellular grid. The plastic,
differentiating cellular grids in Breed and Ima
Traveller only exist inside their resultant forms.
This change fits with the artists’ leanings
towards the material: in a conventional CA, the
emergent forms are mere epiphenomena —
while they appear to be coherent, mobile and
autonomous, they are simply patterns of
activation travelling over a static array of formal
elements. In the theories of Edward Fredkin this
property of CAs is expanded into a speculative
cosmological theory which he calls Finite nature
(Fredkin 1992; see also Wolfram 2002). Under
this hypothesis, the cosmos is itself a gigantic
CA: our three-dimensional universe of matter,
space and time is merely an emergent pattern
generated by a formal array with a higher
dimensionality. Matter, and life, are held to be
ultimately manifestations of an underlying
process of computation, an immanent logical
substrate. Driessens and Verstappen, by contrast,
use CAs in a way which puts matter, and
morphogenesis, ahead of the logical array. In a
process such as Breed there is no given array,
only a certain level of detail, a certain number of
cellular subdivisions. There is no underlying
granularity: space continues to open up. In Ima
Traveller the same process is at work, though
here the array sprawls outwards endlessly,
growing like a puddle of bacteria.
Thus even as the artists make use of core
a-life techniques such as artificial evolution and
cellular automata, they reconfigure and reengineer
those techniques. At one level, the
changes that they make may seem to be simply
arcane technical ‘tweaks’; these tweaks also entail
important metaphorical and conceptual transformations,
however. In reconfiguring artificial
evolution, Driessens and Verstappen move away
from the metaphorical structure which so
dominates the use of that technique, both
within a-life science and a-life art. Here there is
no grand evolutionary trajectory, no sexual
reproduction or family tree — not even a
biomorphic aesthetic ‘organism’. While other
breeders draw heavily on these familiar ideas,
both as conceptual models and explanatory
devices, Driessens and Verstappen apply the
techniques without their figurative baggage.
These notions are still present, but as Verstappen
(1999a) explains their role is strictly explana-
tory:
53
DigitalCreativity,Vol.14,No.1
Morphogenetics
In some works we use metaphors like celldivision,
evolution, genetic code, and so on, to
explain what happens. But on the level of the
internal process, we work with this machine, the
computer has its own physical rules.
What is most significant about this work
is that in disregarding the conventional metaphorical
structures around a-life techniques and
treating those techniques as elements in a more
abstract computational process, Driessens and
Verstappen create systems which are more
elegant and richly generative than many more
conventional creative applications of a-life. They
offer a powerful demonstration that a-life
techniques (and their attendant metaphors)
need not be simply received and applied by
artists. Driessens and Verstappen also show that
a-life’s generative processes can themselves be
unpacked, refashioned and explored more fully.
In part these processes operate here as abstract
morphological machines; however in rearranging
a-life’s templates the artists are also illustrating
alternative notions of space, form, morphogenesis
and evolution.
References
Brockman, J. (1996) The third culture. Simon and
Schuster, New York.
Dawkins, R. (1986) The blind watchmaker. Longman
Scientific and Technical, London.
Driessens, E., and Verstappen, M. (1999) Private
Email. 2 September 1999.
Driessens, E., and Verstappen, M. (2002) Notnot.
http://www.xs4all.nl/~notnot (accessed July 2002).
Emmeche, C. (1994) The garden in the machine: the
emerging science of artificial life. Trans. Sampson, S.
Princeton University Press, Princeton.
Fundación Telefónica (2002) Art and Artificial Life
International Competition. http://
www.fundacion.telefonica.com/at/vida/english/
index.html (accessed July 2002).
Fredkin, E. (1992) A new cosmogeny. http://
www.digitalphilosophy.org/new_cosmogony.htm
(Accessed July 2002).
Flyvbjerg, H., and Lautrup, B. (1992) Evolution in a
rugged fitness landscape. Physical Review A 46
Mitchell Whitelaw is a writer, educator and
artist, working in areas of new media art and
creative digital cultures. He has specific interests
in generative techniques and creative applications
of artificial life, and completed a PhD on
art and artificial life in 2000. More recently his
critical and creative work has focused on
experimental digital audio, and its articulation
of data and matter. He currently teaches in the
School of Creative Communication at the
University of Canberra, Australia.
6714–18.
Kauffman, S. (1993) Origins of order. Oxford
University Press, New York.
Langton, C. G. (1992) Preface. In Langton, C. G.,
Taylor, C., Farmer, J. D., and Rasmussen, S.,
(eds.) Artificial life II. Volume X of SFI studies in
the sciences of complexity. Redwood City, CA. pp.
xiii-xviii.
Lequeux, E. (1999) Ima Traveller. Trans. Holmes, B.
In Driessens, E., and Verstappen, M., Ima Traveller.
Le Carre, Centre culturel des Ursulines, Chateau-
Gontier.
Rooke, S. (2002) The evolutionary art of Steven Rooke.
http://www.azstarnet.com/~srooke (accessed July
2002).
Sims, K. (2002) Karl Sims home page. http://
www.genarts.com/karl (accessed July 2002).
Todd, S., and Latham, W. (1992) Evolutionary art
and computers. Academic Press, London.
Verstappen, M. (1999a) Private Email, 28 October
1999.
Verstappen, M. (1999b) Private Email, 2 November
1999.
Verstappen, M. (2002) Private Email, 22 July 2002.
Wolfram, S. (2002) A new kind of science. Wolfram
Media, Champaign, IL.