The following terms are often used in describing artificial life and the|
process of Evolutionary Art, at least as employed by .
aesthetic fitness score: a number assigned to the image produced by an evolved organism that determines the organism's likelihood to breed and have offspring (see aesthetic selection).
aesthetic selection: natural selection is the built-in process by which Nature chooses survivors and hence the shape of future generations; aesthetic selection is the external process by which an artist steers the course of image evolution and development.
algorithmic art: computer-generated artwork that derives from mathematical functions or programs, as opposed to being drawn or painted by hand or by scanning in photographs. In "true" algorithmic art, the artist does not retouch the image in any way, e.g. by using software tools to add or remove features from the image. All the work goes into setting up the process that produces the image. Evolutionary art is algorithmic.
algorithms: groups of instructions for performing specific mathematical or computational tasks.
archetype: the notion proposed by Carl Jung that certain powerful images or concepts have a genuine existence or reality - somewhere (I call it hyperspace for lack of a better term) - hence the collective or universal unconscious. This idea cannot help but resonate with me when a powerful but unexpected image jumps out of the computer screen as it were. I think of Joseph Campbell's description of common themes in origin myths around the world when this happens.
archetype camera: my fanciful notion of constructing an apparatus employing a visual image evolution system coupled with a realtime source of random numbers from nature, with a human observer in the loop that might produce visual imagery widely recognizable as universal archetypes. The system of co-evolving image generators and image commentators I still hope to have time to build might be a step in this direction.
artificial life: (or alife) in this context, a system grown in software inside a computer that exhibits many of the properties of living systems, including birth, mating, death, and adaptive change over many generations; artificial life is often characterized by self organizing systems evolving from the bottom up, as opposed to artificial intelligence, which is usually thought of as being completely designed by humans from the top down. My evolutionary art system is not artificial life in the sense of Tom Ray's Tierra (click here to go to its website) from a New York Times article in 1991, which can be argued to be genuinely alive within its boundaries, because in my case the artist is loosely or strongly guiding the evolution of the images. However, a long-term research goal is to construct an architecture of co-evolving image generators and image recognizers (like pollinating flowers and bees), which could lead to an open-ended evolution of images with the artist's hand guiding the evolution from ever higher levels further removed from the details of genetic variation and selection.
breeding cycle: the process of aesthetic fitness based mating and mutation in a population of digital organisms that leads from one full generation of individuals to the next.
chaos: dynamical systems displaying sensitive dependence on initial conditions. The weather is chaotic, a healthy heartbeat is chaotic. Fractal equations lead to chaotic bifurcations and an infinity of self-similar structures as in the Mandelbrot Set. In recent years mathematicians and scientists have discovered much hidden order in some chaotic systems - behind the scenes as it were, in spacetime patterns of phase space. Hundreds of books have been written, but a good starting point is James Gleick's Chaos.
chaoscope: a term coined by Ralph Abraham to describe the use of computer graphics in visualizing dynamical systems; the idea is to get people thinking of such computer simulations as revealing aspects of the way the world works. Telescopes let us see large things very far away; microscopes let us see very small things; chaoscopes let us follow trajectories of dynamical systems, make pictures of phase space attractors.
co-evolution: the evolution of two or more species, where the development, activity, and fitness of each species mutually affect those of the others. Many flowering plants and pollinating insects have co-evolved to the point where each is entirely dependent on the other. For an accessible description of co-evolving adaptive fitness landscapes, see Stuart Kauffman's excellent book At Home in the Universe.
complex adaptive system: a self organizing system of interacting components that evolves by constantly adapting to its changing environment and itself; you are a complex adaptive system.
complex plane: the mathematical plane described by one real axis (usually taken to be the x-axis) and one "imaginary" (square root of negative 1) axis. In evolutionary art it is sometimes advantageous to treat the canvas for the image as a portion of the complex plane, at other times as part of the real plane (simply two real-number coordinate axes).
Darwinian: a process of individual variation from generation to generation and natural selection leading over time to a system evolving along the lines proposed by Charles Darwin.
daughter population: if the group of organisms alive at a particular moment in time is thought of as the population, its daughter population is the one that comes after it by breeding, mutating, and copying its individuals.
digital amber: a metaphor for how the genetic code of an individual organism is preserved in the computer after a run is terminated, and whereby it may be resurrected by inoculation into the primordial soup of a future run (the notion of preserving the genetic information of dinosaurs in real amber was popularized in the movie Jurassic Park).
digital organism: a virtual organism described and implemented inside an artificial life computer program.
diversity: in Evolutionary Art, as in Nature, a healthy system or population is one which exhibits a broad range of characteristics, or diversity; if all organisms are identical or very similar, there is little scope for evolution and change, and the entire population is at risk, say from the arrival of a new, virulent parasite .
DNA: the molecules that carry the genetic code for all living organisms on Earth (deoxyribose-nucleic-acid); the term is sometimes used loosely in artificial life to refer by analogy to the genetic code of digital organisms in the computer. Purists may argue against using `DNA' at all in this context, but `virtual DNA' is fine with me.
DNA base pairs: the smallest effective building blocks out of which DNA molecules are constructed; used as an analogy in artificial life or genetic programming to refer to the simplest building block functions in an organism's genetic code. One may say that the most elementary mathematical functions used in evolutionary art are akin to DNA base pairs, while more complex functions are more akin to genes.
dynamical function: a function that describes the trajectory of states of a dynamical system, which is outside the scope of this glossary. In evolutionary art dynamical functions add a more chaotic (natural) character to images. For websites containing much information about visualizing spacetime patterns or dynamical systems, see the Visual Math Institute | homepage.
emergent property: a property of a system that emerges spontaneously from within, as opposed to being designed or imposed from without.
Eukaryotic Transition: the symbiotic merger two billion years ago of formerly independent genetic lineages arising from bacteria to become the nuclei, undulipodia, mitochondria, and chloroplasts of the ancestors of all of today's plants, animals, fungi, and protists. See symbiogenesis.
evolutionary art: imagery produced by a process of simulated evolution inside a computer, guided by an artist's aesthetic fitness selection; influential pioneers of the process of generating evolutionary art are Karl Sims and Richard Dawkins .
exon: a portion of DNA, or genetic art code, in the genotype that exerts an observable effect on the phenotype, or adult organism. By contrast, much of both biological DNA and the genetic art equivalent appears to be `trash' (introns), not encoding for any obvious characteristic, yet still quite important in ongoing evolution.
fitness: the degree to which an organism is fitted, or adapted, to its environment and other organisms. Higher fitness means superior ability to survive and reproduce.
fitness assignment: the means by which an organism's ability to survive and reproduce is determined; in aesthetic image selection this is the numeric aesthetic fitness score assigned by the artist to each individual in the population.
fitness landscape: the structure of fitness hills and valleys of a species in its environment, or of many species, whose individual successes and failures in adapting to change alter that structure itself. See Stuart Kauffman.
fractal: coined by Benoit Mandelbrot, a fractal structure means one that exhibits a constant amount of detail no matter how highly it is magnified. Fractal structures exhibit scaling self-similarity, e.g. a form that looks like a seahorse turns out upon closer examination itself to be composed of an infinite number of smaller and smaller seahorses. I often include fractal functions in my genomes. See also Clifford Pickover and Ken Musgrave.
functions: John Koza introduced the notion of functions and terminals in genetic programming; functions take one or more inputs and produce a single output, while terminals take no inputs - their output is simply their value. Koza's GP paradigm utilizes a tree-like branching structure, where functions are like forks in a tree, and terminals are like leaves. In programming an evolutionary art computer software system, one may employ both spatial (geometric, trigonometric, image processing) and dynamical or iterative functions.
gene: in biology, a segment of DNA whose code determines specific features or groups of features in an organism; in evolutionary art, groups of mathematical functions that produce the specific features in an image.
gene bank: in evolutionary art, a digital repository of thousands of genetic sequences representing successful images from previous runs, essential for seeding new runs.
genetic code: the set of instructions that determines the growth, type, shape, and other characteristics of a living or artificial organism.
chaoscope: a term coined by Ralph Abraham to describe
genetic cross dissolve: Invented by Karl Sims in 1991, a means of generating a trajectory through any number of theoretical genetic relatives between any two given individuals. You line up a number of images, and GenCross produces video sequences passing through them, but visiting previously unseen and unpredictable intermediate forms.
genetic structure: the shape or characteristics of the instructions themselves; in living systems the genetic structure is that of DNA molecules, while in Evolutionary Art the genetic structure is at the whim of the artist / programmer.
genetic programming: a system for programming a computer popularized by Professor John Koza of Stanford University, wherein the genetic structure of an organism, produced during evolution, both determines the nature of the organism and allows it to reproduce, and is a set of instructions for carrying out a task; in Evolutionary Art the task is to create a visual image.
genetic variation: the variation of characteristics from one individual to the next within a species; an evolving system must possess such variation for selection, natural or guided, to have a foothold for implementing change.
genotype: the genetic code that characterizes a species of organisms; one often distinguishes between the genotype (coded instructions) and phenotype (resulting physical body or, in this case, a picture or image) of an organism.
geologic record: the evidence in rocks, minerals, fossils, the shapes of continents, etc., that tells the story of the Earth's birth, history, and evolution up to the present time.
guided selection: in Evolutionary Art, the artist guides the selection of digital organisms, determining their fitness to reproduce; this guidance may be loose (probabilistic) or more tightly constrained - in Nature, selection is natural rather than guided.
high resolution: an image has high resolution if it is very sharp, clear, and detailed even upon magnification; the more pixels, the higher the resolution (see pixel).
host organism: a host organism is one which harbors parasites, smaller organisms that use some of the host's body or genetic material for their own life cycle and reproduction.
hyperspace: an abstract multidimensional space in either the mathematical sense or philosophical: dynamical systems operate in high-dimensional phase spaces; consider the hyperspace of all possible two dimensional colored images.
image generators: computer generated, evolved organisms whose phenotypic expression is to produce images, whether viewed on a computer screen or reproduced as prints. The fitness of an image generator as a breeding organism is determined aesthetically by an artist -- or, in principle, by other evolving image recognizers.
image recognizers: hypothetical computer-evolved organisms that "observe" the images produced by a population of image generators, and "comment" on them; over evolutionary time their comments then become the basis for determining fitness of the image generators in a co-evolutionary cycle.
incest: in Evolutionary Art the populations are relatively small, around 100 individuals, and it is not uncommon for siblings to mate and produce offspring (see also parthenogenesis).
inoculate: to insert something into a growing medium; in Evolutionary Art one may inoculate the starting point for evolution (the primordial soup) with genetic information from preserved individuals or species from earlier runs.
intron: a section of DNA, or a portion of a genetic art coding sequence, that appears not to have any observable role in determining the characteristics of the adult organism, hence sometimes called `garbage'. In genetic art, it is clear that the effects of such unexpressed virtual DNA in, say, one's 5th-great-grandparent, can have a large impact on the appearance of the individual in question, perhaps supporting the case of those evolutionary theorists who posit greater value to introns than `garbage'.
IRIS print: a popular printing technology for computer generated images. They can currently print on many kinds of art paper up to 36"x46" in size, at 300 dots per inch, and the quality and stability of inks improves yearly.
iterative function: a function, like the Mandelbrot Set, in which an equation including the same quantity on both sides of itself is operated repeatedly until some termination condition is met. In genetic programming for evolutionary art, iterative or dynamical functions appearing inside the genome alongside geometrical ones introduce a complex mixing of spatial and dynamical or temporal characteristics, resulting in more interesting images.
Mandelbrot Set: the widely known fractal set discovered by Benoit Mandelbrot that produces an infinite variety of forms revealed by computer graphics. The M-Set is the grand-daddy of deterministic chaos sets produced by iterating a simple equation in the complex plane many many times. In evolutionary art, the M-Set is of limited but occasional use when caught up in a whirlwind of genetic reassemblage of functions.
mates: in Evolutionary Art, if random chance plus their aesthetic fitness scores permit, two individual organisms may mate to produce offspring in the next generation or daughter population; the offspring generally exhibit some of the visual features of the images produced by their parents, but not always.
mathematical functions: adding two numbers is the mathematical function add; in Evolutionary Art the basic building blocks that produce organisms capable of generating visual images may be simple functions like "add" all the way up to complex mathematical functions composed of hundreds or thousands of computer instructions.
meme: a term coined by Richard Dawkins to represent the basic unit of replication of ideas in the primordial soup of human culture, analogous to the gene of biological evolution. See The Selfish Gene, p.192 2nd ed. 1989.
mutation: an organism's genetic code may be changed, or mutated, in some random way before or during reproduction; this is one of the sources of variation in the appearance or characteristics of individual organisms from one generation to the next (as opposed to sexual reproduction).
natural selection: Nature's builtin process of determining which individuals in a population survive and reproduce; contrasted with aesthetic selection, wherein an evolutionary artist guides such selection.
organism: a single functioning entity in a population; you are a human organism, while in Evolutionary Art an organism is the structure inside a computer that mates with others and produces a single image.
parasite: an organism that lacks sufficient genetic information to reproduce on its own, instead relying on other, host, organisms for parts of its life cycle and reproduction. One can imagine the eventual evolution of true artificial life-generated images, with "image parasites" and "aesthetic immune systems" evolving.
parthenogenesis: the process, not uncommon in Evolutionary Art but rare in Nature, wherein an organism reproduces by, in effect, mating with itself. When certain rare animals reproduce by parthenogenesis, their offspring are generally identical clones; in evolutionary art this is not often the case, since their genetic codes may be joined at different crossover points.
phenotype: the physical expression of an organism in its environment: the human body is the phenotypic expression of our genotype (DNA), while an artistic image is the phenotypic expression of its genotype (virtual DNA) in evolutionary art.
phylum: a fundamental grouping of charasteristics of living organisms; everything with a spinal chord - fish, reptiles, amphibians, mammals, and a few others - are in Phylum Chordata. In evolutionary art it often seems that major groups of images likewise represent separate phyla, as they appear quite different and are difficult if not impossible to crossbreed into anything successful.
pixel: an individual tiny dot of color in an image or on the computer screen; a good computer screen may have a million pixels, while a high resolution image reproduced on film or other medium may have several hundred million pixels; in some forms of Evolutionary Art a pixel is also analogous to a single cell in biological organisms.
population: the group of all individual organisms alive at a given moment or during a given generation.
primordial soup: in Earth history, the oceans that existed before any life had evolved are often called the primordial soup, containing the building blocks that would eventually yield living organisms; in Evolutionary Art, this is the collection of mathematical building blocks and their environment before any organisms have been assembled.
probabilistic: loose, chancy, partly randomly determined, as opposed to definite or precisely determined; in Evolutionary Art a certain amount of random chance is essential to the successful evolution of interesting images - the artist is always balancing between too much control of the process and too little.
programming paradigm: a characteristic method or scheme for programming a computer; the genetic programming paradigm is that method wherein a genetic structure is both a set of building blocks that can mate and reproduce, and a program for doing something, such as creating a visual image.
pseudorandom number generator: the most common means of introducing randomness into a digital computer program. They generate long, but finite, strings of numbers whose sequence statistically yields no observable pattern (when working properly...). They all have the property that when started with the same "seed", they yield precisely the same sequence of numbers. This is an advantage when trying to reproduce results in evolutionary art runs, but it may introduce a level of determinism distasteful to the artist.
random: in genetic algorithms and simulated evolution in a computer, one needs a plethora of "random" choices: for choosing mates probabilistically based on their fitness scores, for selecting sites along the genetic code for sexual crossover, etc. How does one introduce such randomness into that bastion of determinism, the digital computer? Most commonly through the use of a pseudorandom number generator (above). But Rupert Sheldrake has suggested the use of genuine realtime random noise generators in such simulations to test an intriguing though highly controversial theory of causation called morphic resonance. Generating evolutionary art with such a realtime random number generator may be more satisfying to an artist because no work could ever be repeated, unlike restarting a run with a preserved seed from a pseudorandom number generator.
resolution: image resolution is the amount of detail relative to size. Even the newest, largest computer screens are only 1280 to 2048 pixels wide; typical screens and television have only 640x480 pixel resolution. A 30" x 40" IRIS print at 300 dots per inch is 12,000 pixels wide, around 300 Mb compressed, while the highest resolution images I have made are 19340x15240 pixels (890 Mb), on a Kodak LVT film recorder using 16"x20" sheet film. To produce any large artwork, you need much larger images (more pixels) than you see on a computer screen. Special computer software is often required to handle very large images.
runts: in a litter of baby animals, the smallest or slightly deformed are often pejoratively called runts; they may or may not survive to adulthood to reproduce - in Evolutionary Art, as in Nature, the runts may not appear very successful in a given generation, but sometimes their offspring many generations removed may inherit and turn to advantage some of their hidden characteristics.
self organization: a process exhibiting emergent properties whereby certain systems achieve a degree of order from within, as opposed to having it imposed from without; living systems are self organizing within the broader environment in which they are embedded, while a computer is assembled from without by humans and machines. See Stuart Kauffman.
sexual reproduction: in the context of Evolutionary Art, sexual reproduction is the process wherein two organisms (more accurately, their genetic codes) mate, producing offspring who inherit some of the characteristics of each parent (as opposed to mutation, which is a different method of supplying variation from one generation to the next).
simulated evolution: in artificial life, simulated evolution is a system usually implemented inside a computer for abstract modeling of the evolutionary process; some researchers point out that even though the conditions set up by humans in the computer appear artificial, the evolutionary processes themselves are nevertheless real (see Tom Ray).
species: loosely, organisms of a particular type that are capable of mating and producing viable offspring; in Evolutionary Art species are determined by the set of building blocks available to them and the maturity of the computer program that produced them at a particular time.
spore: the means of reproduction of certain plants and fungi, functioning like seeds in flowering plants.
spore from space: a reference to the theory of Panspermia, wherein it is proposed that spores or other microscopic, durable forms of life or seeds of reproduction might be blown about interstellar space, eventually perhaps landing on a living planet and bearing their genetic information from a distant point of origin; in Evolutionary Art spores from space are occasionally brought in to introduce new genetic variation into a population that has lost diversity.
symbiogenesis : the process, championed by Lynn Margulis, whereby formerly independent genetic lineages of organisms merge symbiotically to produce new composite organisms at a higher hierarchical level of organization, as did the ancestors of the mitochondria, chloroplasts, and undulipodia in all organisms other than bacteria, beginning at the Eukaryotic Transition.
terminal: in John Koza's Genetic Programming paradigm, functions are like forks in a tree, while terminals are like leaves. In this style of evolutionary art a terminal may be a constant number, an x or y coordinate, or a random number.
virtual: an implicit or abstract representation of something; a virtual organism in Evolutionary Art refers to a non-physical collection of computer instructions and memory that behaves as an entity and generates visual images, as opposed to a real, physical living organism.
virulent: a virulent parasite is one which reproduces rapidly in an individual and in a population, often to the detriment of its host organism.