Nostalgia for Adaptive Systems

After posting my thesis I felt a little nostalgic for the area of research for which I'd invested nearly four years. I went in search of a fix today, browsing around the impressive collection of videos on VideoLectures.net, and stumbled across a lecture on evolution and complex adaptive systems entitled "Adaptive Behaviour and Emergence" by Seth Bullock.

Seth's introduction of complex and adaptive systems was tight, reminding me of the relationship of this field with my own doctoral research and the months I spent climbing into it. Specifically, while taking notes, I noticed that the packaging of the field and the way it was presented was much the same as they way I discovered and navigated the material on my own, pushing out technical reports on: Complex Adaptive Systems (core CAS theory), Satisficing, Optimization, and Adaptive Systems (confusing influence of optimizing selectionist models), An Adaptive Systems Formalism (core adaptation theory), and Darwinism and Selectionist Theories (units of selection). These threads motivated large aspects of my research and appeared in my thesis in various refined forms.

Anyway, I thought I would capture my notes with links to motivate anyone generally interested in the topic to check out the video lecture.

• Dealing with complexity: system level behaviours (emergent), differentiate complex systems (weather), from complex adaptive systems (brain), from mixed systems (ecosystems). Differentiate nomenclature adaptation, adapt, adapted, etc
• Study of Complex and adaptive systems: spans domains, core features that such system exhibit such as scale, connectivity, nonlinear interactions. adaptive systems are typically complex, but not necessarily the reverse
• Units of selection: genes as replicators (classical Dawkins), genes as information, factors include: longevity, fecundity, fidelity of copy. Persist across generations. Gene focus dissolved the siren of group selection - the adaptive fit for the good of the group (species, sub-species group, ecosystem, etc)
  
• Game Theory: borrowed from economics and international relations, resulted in huge progress in the study of evolution. Classical games from such work that have stuck: tragedy of the commons, prisoners dilemma, hawk-dove. Lesson: short term gains do not pay-off.
  
• Evolution appears progressive: Another trap. Idea of evolution as a force for progress, evolution != optimization (adaptationism), apparent increase in complexity, simple models to demonstrate random walks can demonstrate similar properties, evolution as directed change with positive effect (for who?)
• Arms Race: addressing progressive arguments, more classic Dawkins, notion of local progress, no necessary winners, no long term strategies or goals, short term progress
• Major Transitions: more on evolution as progress, transition events that introduce major changes, open up new opportunities, new niches, can go back (be lost) but typically do not, Dawkins' proposition of watershed events and their preservation as a one-way ratcheting (examples like the cell, multi-cellular, etc)
• Conceptualization: fitness landscapes, from optimization perspective, hill climb landscape of potential arrangements, genetic operators and resultant neighbourhoods are far more complex in reality
• Open-ended evolution: fitness landscapes bound evolution, complexity perspective prefers an open ended perspective, reliance on an environment under which adaptation is promoted, replicators capable of creating more complex replicators (grammar)
• Apparent Design != Evolution: structures that seem as as though they were designed may not be adaptations, they may be the result of natural processes (physics), for example ordered structures, galaxies, solar systems, snow flakes
• Scope: Physics defines the scope of stable forms as well as evolvable forms, constrains evolutionary processes
• Self Organization: Introduces the importance of self-organization to complexity theory and adaptive systems, classic Kauffman self organizing structures as the seeds for natural selection (morphospace)
  
• Edge of chaos: complex adaptive systems cannot be either too stable or too chaotic, sweet spot 'on the edge of chaos', classic Langton
• Power Laws: scale free properties of complex systems, difference between normally distributed and power law distributed, solar flares, earth quakes, classic self organized critically
  

You have got to love Wikipedia these days, it has so many technical topics!

I enjoyed Seth's direct presentation of the material and expect to watch some of his remaining video lectures that seems to focus around simulation and evolution, not surprising given his research into simulated evolution.