. Synthetic ethology is proposed as a means of conducting controlled experiments investigating the mechanisms and evolution of communication. After a discussion of the goals and methods of synthetic ethology, two series of experiments are described based on at least 5000 breeding cycles. The first demonstrates the evolution of cooperative communication in a population of simple machines. The average fitness of the population and the organization of its use of signals are compared under three conditions: communication suppressed, communication permitted, and communication permitted in the presence of learning. Where communication is permitted the fitness increases about 26 times faster than when communication is suppressed; with communication and learning the rate of fitness increase is about 100 fold. The second series of experiments illustrates the evolution of a syntactically simple language, in which a pair of signals is required for effective communication. Keywords: artificial lif...

Distributed representations are attractive for a number of reasons. They offer the possibility of representing concepts in a continuous space, they degrade gracefully with noise, and they can be processed in a parallel network of simple processing elements. However, the problem of representing nested structure in distributed representations has been for some time a prominent concern of both proponents and critics of connectionism [Fodor and Pylyshyn 1988; Smolensky 1990; Hinton 1990]. The lack of connectionist representations for complex structure has held back progress in tackling higher-level cognitive tasks such as language understanding and reasoning. In this thesis I review connectionist representations and propose a method for the distributed representation of nested structure, which I call "Holographic Reduced Representations " (HRRs). HRRs provide an implementation of Hinton's [1990] "reduced descriptions". HRRs use circular convolution to associate atomic items, which are rep...

Abstract not found

We propose certain non-Turing models of computation, but our intent is not to advocate models that surpass the power of Turing Machines (TMs), but to defend the need for models with orthogonal notions of power. We review the nature of models and argue that they are relative to a domain of application and are ill-suited to use outside that domain. Hence we review the presuppositions and context of the TM model and show that it is unsuited to natural computation (computation occurring in or inspired by nature). Therefore we must consider an expanded definition of computation that includes alternative (especially analog) models as well as the TM. Finally we present an alternative model, of continuous computation, more suited to natural computation. We conclude with remarks on the expressivity of formal mathematics. Key words: analog computation, analog computer, biocomputation, computability, computation on reals, continuous computation, formal system, hypercomputation,

this paper we outline the general characteristics of continuous formal systems

The performance of symbolic inference tasks has long been a challenge to connectionists. In this paper, we present an extended survey of this area. Existing connectionist inference systems are reviewed, with particular reference to how they perform variable binding and rulebased reasoning and whether they involve distributed or localist representations. The benefits and disadvantages of different representations and systems are outlined, and conclusions drawn regarding the capabilities of connectionist inference systems when compared with symbolic inference systems or when used for cognitive modelling.

It has been argued that neural networks and other forms of analog computation may transcend the limits of Turing computation; proofs have been oered on both sides, subject to diering assumptions. In this report I argue that the important comparisons between the two models of computation are not so much mathematical as epistemological. The Turing machine model makes assumptions about information representation and processing that are badly matched to the realities of natural computation (information representation and processing in or inspired by natural systems). This points to the need for new models of computation addressing issues orthogonal to those that have occupied the traditional theory of computation. Keywords: computability, Turing machine, hypercomputation, natural computation, biocomputation, analog computer, analog computation, continuous computation 1.

This article describes a series of experiments to study the emergence of inherently meaningful communication by synthetic evolution in a population of artificial agents. By "inherently meaningful" we mean that the communication is meaningful and relevant to the agents themselves, independent and regardless of any meanings we (as observers) may attribute to the communications. (We discuss elsewhere [7, 13] the relevance to the study of intrinsic intensionality of these experimental techniques, which we call synthetic ethology.) Briefly, we may say that communication is inherently meaningful if it has some actual or potential relevance to the agents.

This article will attempt: (1) to explain the importance of new information processing technology, (2) to discuss its potential impact on computing, and (3) to suggest research necessary to its successful implementation. 1.2 Artificial Intelligence: Old and New

As defined in MacLennan (1987), a field computer is a (spatial) continuum-limit neural net. This paper investigates field computers whose temporal dynamics is also continuum-limit, being governed by an integro-differential equation. Such systems are motivated both as a means of studying neural nets and as a model for cognitive processing. As this paper proves, even when they are purely linear. such systems are computationally universal. The "trick" used to get such universal nonlinear behavior from a purely linear system is quite similar to the way nonlinear macroscopic physics arises from the purely linear microscopic physics of Schrödinger's equation. More precisely, the "trick" involves interpreting the system in a non-linear way. That is, the meaning of the system's output is determined by which neurons have an activation exceeding a threshold (which in this paper is taken to be 0), rather than by the actual activation values of the neurons. (This