Interaction, self-reference and contingency in computational neuroscience: analytical descriptions and information theoretic consequences

We will attempt to show that we can use self reference in conjunction with signal theory (vis. electrical systems theory) to construct an artificial organism which complies with the demands of autopoietic systems. The organism is able to develop autonomous behaviour due to sensor-motor learning which is inspired by Piaget [Piaget (1971)]. Learning is achieved through the experience the organism gets while interacting with its environment. This interaction is characterised by closed sensor-motor loops and can not be reduced to a stimulus-response scheme moreover any interruption of the sensor-motor loop makes the system dysfunctional. In this sense our model is non reductionistic. The results of this investigation is expected to be useful for the understanding of social systems since in the organism's environment there are other organisms which also learn. This leads to mutual interaction between them. Thus, the environment is contingent to each individual organism since there are other organisms in the environment which exhibit unpredictable behaviour. As a consequence this is directly related to Luhmann's double contingency of the communication process. The underlying information theory differs from the typical input/output paradigm and is a selfreferential information theory or an information theory of the information which is gathered to improve behaviour (in this case to build up predictive structures).