Ashby, W. Ross

The mechanism of habituation


W. ROSS ASHBY SUMMARY THE Phenomenon of habituation, in which the response to any regularly repeated stimulus decreases, has not so far received any general mechanistic explanation. It is here shown that when any system is subjected to a regularly repeated stimulus or disturbance, the successive responses, if they change in size, do not in general tend to become larger or smaller with equal probability: there is a fundamental bias in favour of the smaller. The atm of this paper is not, in any case, to relate habituation to concepts of physiological or psychological type such as "fatigue" but to relate it to basic concepts of mechanistic type. As will be shown below, the basic phenomenon of habituation can be identified over a very wide range of systems, and only a language that can range equally widely is appropriate.

Mechanical Chess Player


The next principle is that the machine, as it faces a position, must form a great variety of transformations from the position, must form a valuation from each transformation, must follow the line of action that these valuations suggest, and then, when the game is over, must go back to modify the transformations and valuations simply accordin to whether the game was won or lost. A Geiger counter suitably worked up will provide variety and may lead to such rules as: always keep rooks as far apart as possible, or, always keep one bishop on a white square and one on a black. However complicated and subtle the really good transformations are, Brownian movement can provide them. I think if you put any limitations on the ability of either machine or human operator to analyze, you put an exact equivalent limitation on the ability to form strategy.