Towards Implementing Free Will - Bruce Edmonds
Those who insist that free-will requires prior free-will (arguing that
otherwise the choice process would also be determined) can follow the chain
of causation (and indeterminism) backwards until it slowly diminishes down
a limit of nothing (determinism). In this model the gradual emergence of
free-will in the brain is analogous to the emergence of life - it can start
from infinitesimal amounts and evolve up from there. This requires that
free-will can come in different degrees - that circumstances can constrain
behaviour to different extents from totally (determinism) to partially
(some degree of indetermination). The artificiality of an all-or-nothing
division into having it or not makes as little sense with free-will as
it does with life, especially if one is discussing mechanisms for its appearance
(as must occur somewhere between the newly fertilised embryo and the adult
human. As Douglas Hofstadter said [8]:
"Perhaps the problem is the seeming need that people have of making black-and-white cutoffs when it comes to certain mysterious phenomena, such as life and consciousness. People seem to want there to be an absolute threshold between the living and the nonliving, and between the thinking and the "merely mechanical,"...Thus a situation where free-will evolves in increasing effectiveness during the development of the brain satisfies the first of my criteria. Not only can the actions be free, but also the deliberation that resulted in those actions be free and the process to develop those deliberations be free etc. The fact that the chain of free-will disappears back into the internal evolutionary process can be expressed as a closure property.selective advantage that this feature confers upon us (as a species) is primarily that of external unpredictability (combined with an internal rationality). That is in a competitive environment, if an opponent can predict what you will do then that opponent would have a distinct advantage over you. Such competition in a social setting has been posited as one of the evolutionary selective factors that promoted intelligence in our species [2]. Unpredictability can be evolved has been shown by Jannink [10]. He developed a simulation with two separate populations which were co-evolved. The first of these populations was allocated fitness on the basis of the extent to which its programs successfully predicted the output of programs from the other and individuals from the second were allocated fitness to the extent that it avoided being predicted by individuals from the first population. Here the two populations are involved in a basic evolutionary `arms-race'.the basic architecture I am suggesting is composed of the following elements:
Generated with CERN WebMaker