Tracking and averaging in variable environments: A transition rule

Sprague-Dawley rats were exposed to variable patches in 2 controlled laboratory settings where duration and magnitude of yield were systematically varied along with the time intervening between foraging experiences and choice tests. Following relatively short delays, rats chose the patch that was best at the end regardless of its average value. However, as delays lengthened, rats chose the patch with the best average irrespective of its final performance. Despite widely differing procedures, preference shifts were an orderly function of time and patch quality and were predicted with precision by the temporal weighting rule (TWR), a biologically based quantitative model that discounts patch outcomes as a hyperbolic function of time. The laboratory phenomenon of spontaneous recovery is best understood in terms of the dynamic averaging process prescribed by TWR.