In-depth reading:
Steve Grand: Creation - Life and how to make it (Orion Phoenix, 2001)
Steve Grand is the creator of Creatures, the first computer game to use genuine artificial life. In this book, he not only takes the reader step-by-step through combining the ingredients for making an artificial creature with lifelike characteristics, but also discusses many of the social, philosophical and ethical issues surrounding this field of endeavour.
Introduction: A Latter Day Frankenstein
Philosophical discussion - what is life? Problem of vitalism and the need to let it go (e.g. Penrose's quantum theory of minds). Grand favours a qualified materialist approach and stresses the danger of explaining something (minds, life) away, like a magic trick, instead of explaining it for greater understanding.
Chapter 1: Failing the Test
Discussion of AI and its problems, Turing Test etc. … there is still a long way to go. There has been a recent paradigm shift from heirarchical, top-down, linear processing towards parallel, bottom-up, emergent models. Chapter ends with an interesting comparison of assumptions from the old model and how they have changed with the new model.
Chapter 2: Lies, Damn Lies and Linguistics
Figure & ground; tangible & intangible - the problems caused by linguistic denigration of the intangible. Things we care about most are intangible e.g. "life", "mind", "poverty", "society". Vitalism seeks a special kind of spiritual "stuff" while Materialism says matter is all. "There is no such thing as a thing" - the brain carves up the visual field into disctrete objects: put boundaries around things. There are intangible things, e.g. clouds, persistent patterns in which the constituent matter is constantly changing. Living organisms are in the same category: "Living organisms are systems in flux".
Chapter 3: A Guide to the Intangible
You are a persistent phenomenon, even though the substrate from which you are made is in constant flux. "Things that persist, persist. Things that don't, don't". "Phenomena persist because they are either inherently stable in an absolute sense, or because they are more stable than other phenomena with which they are somehow competing" (p42). (a good definition of natural selection) Things are not things as such, but rather disturbances in something, like waves or ripples. Certain configurations of matter produce emergent properties such as life or consciousness. A discussion of Conway's Game of Life. "The universe is not made of stuff but of events and relationships".
Chapter 4: Levels of Being and the General Scheme of Things
Simple persistent phenomena can give rise to more complex ones. If the resulting arrangement is more stable it will persist and the system will tend to become more complex. Anything stable in itself will persist unless there is competion for resources. Living systems are autocatalytic networks, "taking in raw materials from outside and making more of itself, completely automatically as a consequence of the nature of its ingredients". Autocatalysis between different elements within a system leads to more complex persistent phenomena. Because they compete for resources, they need to be not only stable, but more stable than their neighbours. Life = catalytic closure. Encapsulation, e.g. natural bubbles, keeps the ingredients together. Bubbles will divide if they get too big (reproduction), occasional damage (mutation) leads to a varied population. DNA (copy the recipe, not the food) amplifies the effect of mutations and opens up the possibility of duplications - "spare" DNA, where variation can flourish. Heirarchy of persistent phenomena: photons —> particles —> atoms —> molecules —> autocatalytic networks —> self-reproducing systems —> adaptive systems —> intelligence & mind —> (society).
Chapter 5: The Importance of Being Emergent
Definition of emergent phenomena - complex results arising from simple interactions between members of a population - results which can only be "predicted" by actually running the system - e.g. gliders in Conway's Game of Life. Emergence is self-perpetuating and bottom-up. Cause and effect happen in webs, not chains. There is no top-down control, just a web of feedback loops - does your brain control your thoughts or do your thoughts control your brain? In order to make life, we will "look into the souls" of persistent phenomena and extract their essence in order to build new things with it.
Chapter 6: Looking Glass World
How can binary numbers and logical operations produce virtual reality? Keep things simple and use an inside —> out approach, not an outside —> in approach.Don't try to emulate the appearance of a behaviour, try to emulate the physical processes that produce that behaviour. Example: bouncing ball using just simple arithmetic. "A simulation of a living thing is not alive and a simulation of intelligence is not intelligent. On the other hand, intelligent, living things can be made out of simulations" (p91). E.g. a simulation of an atom is not an atom, but a molecule made from simulated atoms could be called a real molecule. Second- and higher-order simulations are real things in a way that first-order ones are not, as long as the first-order simulations from which they are made are sufficiently realistic. Some systems are "brittle" - tiny mistakes in the first-order simulations get magnified in subsequent layers, but lots of systems are not brittle, but "robust", and produce similar enough emergent properties regardless of small differences in the first-order simulations. We need to create a virtual space for our virtual lifeforms.
Chapter 7: They Call Me Legion For I Am Many
Don't tell a computer what to do, tell it what to be. The first law of thermodynamics states that energy/matter cannot be created or destroyed. In the real world, this provides stresses that drive biology and evolution, because of finite resources. Computers, however, don't really move data, they copy it from one place to another, so it is necessary to impose constraints artificially. Intelligence or life emerge from the interactions between lots of parallel processes, but computers are serial machines. Parallel processing must be simulated in order to create the right conditions using time-slicing loops and buffered memory, for example, in Conway's Game of Life each cell decides on its next state, storing the answer in memory, before any of the cells change. "Intelligence is not the ability to follow rules, it is the ability to develop the rules in the first place" (p110). Discussion of topology and dimensionality in relation to programming: serial processing is more or less one-dimensional (a line), with limited use of a second dimension (sideways jumps in the code). Parallel processes that interact with each other are more like 3D lattices. Grand asserts that 3-Dimensional parallel processing is necessary for intelligence or life to emerge.
Chapter 8: On the Balance of Nature
Feedback loops enable organisms to stay in their ideal place, e.g. swarms of flies riding a thermal. "Feedback upon feedback means that one species' poise creates the predictability upon which another species' survival depends (like starlings "knowing" that food is to be found inside thermals) (p120). Positive feedback is something that tends to exacerbate change. Negative feedback is something that tends to counteract change. A thermostat is an example of a negative feedback system, turning heat on when it gets cold and off when it gets hot. If it was a positive feedback system the room would get hotter and hotter or colder and colder. Nature is a complex network of multiple feedback loops. Adaptation is a feedback mechanism - e.g. astronauts lose muscle mass in space because it is not needed in low gravity. Eyes adapt to local lighting conditions so that the brightest objects in view will appear bright even in low light, neurones will reduce their response to an unchanging signal (e.g. a constant noise that becomes unnoticeable until it stops). Feedback loops can be visualised as a landscape of hills (positive feedback) and valleys (negative feedback) with balls (organisms) rolling around. Negative feedback keeps a ball near the bottom of a valley while positive feedback makes the ball accellerate down a slope. The bottom of a valley and the very top of a hill or ridge are points of balance - null points - in the system. Positive feedback loops have end points (the bottom of slopes on each side of the hill) as well.
Other Reading …
Manuel DeLanda: Intensive Science and Virtual Philosophy (Continuum, London, 2002)
Deleuze & Guattari: A Thousand Plateaus (Continuum, London, 2004)
Mark Ridley (ed.): A Darwin Seclection (Fontana, HarperCollins, London, 1994)
Aimee Weber, Kimberly Rufer-Bach and Richard Platel: Creating Your World - the official guide to advanced content creation for Second Life (Wiley, Indianapolis, 2008), especially chapters 6-8 on LSL scripting.
Theorizing A-Life, Art and Culture (chapter 6 in Whitelaw, M ~ Metacreation: Art and Artificial Life, MIT, 2004) - I found this useful to help contextualize my project for my learning log. This book looks very very relevant and I intend to go back and read the whole thing.
Gladwell, M: The Naked Face, New Yorker Annals of Psychology, August 5, 2002 - Article about the work of Paul Ekman, inventor of the Facial Action Coding System, and about rare individuals who are naturally extremely skilled in reading facial expressions.
Marc Cavazza et. al.: Modelling the Upokrinomena: Artificial Physiology for Artificial Life