Evolving Fractal Drawings

1
Evolving Fractal Drawings

• Jon Bird
• Centre for Computational Neuroscience and
  Robotics
• University of Sussex
• Dustin Stokes
• Centre for Research in Cognitive Science
• University of Sussex

2
A couple of observations

• Analyzing creativity is not easy.
• Evolutionary robotics is not easy.
• So.lets do both at the same time!
• General research project
• With a broad interest in creative cognition, we
  are using techniques of
• evolutionary robotics (ER)to model creative
  (artistic) processes.
• Tradition start with high level creativity and
  work down
• Start with barely minimal notion of creativity
  and work up,
• attempting to build/evolve the simplest creative
  systems possible.

3

• Minimal creativity
• Evolutionary robotics framework
• On whats missing Value
• Fractal framework
• Why fractals?

4
I. Minimal Creativity
5

• Agency
• Philosophical sense?
• Broad agency Some behaviour, artefact, or event
  x is the product of the agency of A only if x
  would not have occurred had A not acted in some
  autonomous way.
• (autonomous behaviour ? intentional or
  cognitive behaviour)
• (autonomous action ? self-maintaining)
• autonomous behaviour ? self-moved behaviour
• no strings attached.

Cognitive science/AI/A-life sense?
6

• Novelty
• Absolute novelty/novelty simpliciter
• some F is novel only if F has never occurred
  before
• Historical novelty
• (Boden 2004)
• Relative novelty
• Psychological novelty
• (Boden 2004)

7

• Relative novelty but relative to what?
• extend from psychological/cognitive to
  behavioural
• At least two ways to understand
• 1) individual-relative
• A behaviour of some agent R may be novel
  relative to the behavioural history of R.
• 2) population-relative
• A behaviour of some agent R may be novel
  relative to a population of which R is a
  member.

8

• 2 conditions on minimal creativity
• Novelty x is creative only if x is behaviourally
  novel
• Agency x is creative only if x is the product of
  agency.
• Is this enough for creativity?
• Not likely
• So we leave open what conditions should be
  appended for a complete analysis.
• No problem novelty and agency are necessary
  conditions

9
(No Transcript)