[Next] [Previous] [Top] [Contents]

Towards Implementing Free Will - Bruce Edmonds

1 Introduction

To paraphrase the von Neuman quote above: anyone who considers computational methods of implementing free-will is, of course, in a state of sin. By simply suggesting that free-will could be implemented I will already have offended the intellectual sensibilities of several groups of people: I will have offended "hard" determinists by suggesting that free-will is possible; I will have offended those who think that free-will is a uniquely human characteristic; and I will have offended those who see free-will as something that is simply beyond design. I have some sympathy for the later two groups - at the moment a human being is the only system that clearly exhibits this facility; and, as will be explained, I do think that free-will can not be directly designed into an entity.

Despite almost everybody agreeing that it is fundamentally impossible, arithmetic methods of producing random numbers have become, by far, the most widely used method. These methods (used correctly) are efficient and reliable. We rely on their effective randomness in many cryptographic techniques which, in turn, are relied upon in electronic commerce and the like. Maybe it is time to let the evidence take precedence over assumptive theory - if theory disagrees with practical evidence it is the theory that should change. What was assumed to be a state of sin can turn out to be inspired.this paper I will outline a practical architecture that, I argue, could result in a computational entity with free-will. I will start by rejecting extremely idealised conceptions of free-will and suggest instead a more practical set of properties. Then, in section 3, I will put forward the central idea of the paper which is to allow free-will to evolve in a brain during its lifetime. The following 4 sections (4, 5, 6 and 7) consider other necessary aspects of the architecture: open-ended development; the co-evolution of strategies against competitive peers; the meta-evolution of the evolutionary process itself; and the necessity of being able to anticipate the consequences of candidate actions. Section 8 then looks at some societal aspects that might allow the development of a framework of acceptable rationality within which free-will can operate. I summarise the suggested architecture in section 9 and finally conclude in section 10. For those who feel philosophically short-changed by this paper there is a Philosophical appendix which briefly outlines my position in these terms.

Towards Implementing Free Will - Bruce Edmonds - 16 MAR 0
[Next] [Previous] [Top] [Contents]

Generated with CERN WebMaker