ABSTRACT
In this chapter, I recast some of the phenomena of conditioning into the language of dynamic systems. In that mapping, reinforcers and UCSs are cast as fixed-points, or attractors, thus respecting the defining attribute of such stimuli—that they attract (or in the case of aversive stimuli, repel) behavior. Such goal direction occurs in a landscape dimpled with the potential wells of appetitive stimuli and the potential hills of aversive stimuli. Conditioning involves moving neutral stimuli (including those issuing from the animal’s own behavior) from ground level close to the tops or bottoms of these features. Behavior is viewed as the trajectory of a marker as it rolls through this landscape, avoiding hills, veering toward wells, and occasionally being captured by them. The relief of the landscape is determined by the organism’s motivation and may be affected by emotional manipulations, with arousal steepening the features (and forcing behavior to the nearest attractor), and relaxation leveling the features (and freeing behavior to meander). Asymptotic conditioning generates geodesics: paths of minimum length that move away from hills and into wells. Because there is no more direct path than a geodesic, further conditioning is impossible, unless the elevation of the relevant well/hill is affected by manipulations of the reinforcer. This dynamic metaphor accommodates the notions of associative strength, surprise minimization, species-specific defense, and appetitive behaviors; it treats various types of conditioning from a unified perspective. The dynamic systems approach to conditioning is consistent with neural models of attention and categorization and may provide a common language for behavioral and cognitive phenomena.
