"How Evolution Learns From Past Environments To Adapt To New Environments

... suggest that in environments that vary over time in a non-random way, evolution can learn the rules of the environment and develop organisms that can readily generate novel useful traits with only a few mutations."

QUOTE "learn the rules" UNQUOTE ... haha! Semantic tricks again....

Ah - I am reminded of 'the matchbox computer to play hexpawn', which I built once.... :-)

You get lots of matchboxes with the various pictures of all the possible playing positions of the pieces on the outside. Inside each one you place various counters or pieces of paper that reflect all the possible moves from the position on the outside of the box.

You then play the game many times. Many, many times. Many, many, many times. (Sorry - old BBC Radio Comedy Show gag line...)

Each time that you strike a position when the 'computer' loses, you 'prune the tree' of that possible outcome by removing the losing move selector. Eventually, the computer will only win, or draw. You can also prune paths leading to the draw position if you wish to 'force' (as per magician's forcing) the outcome.

Theoretically, this paper pontificatesguesses that if 'natural selection' prunes out the unsuitable outcomes, then the 'organism had learned' - interesting semantic trick ....


"employed computer simulations of evolution of simple computational 'organisms'. "

Ha! - Pilots trained on computer simulations of planes still crash and die when the real physical entity does things that were not pre-programmed - thus one must take great care before pontificatingguessing how the real world will behave...