Ways to test stochastic dynamic programming models empirically

Stochastic dynamic programming (SDP) models are widely used to predict opti mal behavioural and life history strategies. We discuss a diversity of ways to test SDP models empirically, taking as our main illustration a model of the daily singing routine of birds. One approach to verification is to qua ntify model parameters, but most SDP models are schematic. Because predicti ons are therefore qualitative, testing several predictions is desirable. Ho w state determines behaviour (the policy) is a central prediction that shou ld be examined directly if both state and behaviour are measurable. Complem entary predictions concern how behaviour and state change through time, but information is discarded by considering behaviour rather than state, by lo oking only at average state rather than its distribution, and by not follow ing individuals. We identify the various circumstances in which an individu al's state/behaviour at one time is correlated with its state/behaviour at a later time. When there are several state variables,the relationships betw een them may be informative. Often model parameters represent environmental conditions that can also be viewed as state variables. Experimental manipu lation of the environment has several advantages as a test, but a problem i s uncertainty over how much the organism's policy will adjust. As an exampl e we allow birds to use different assumptions about how well past weather p redicts future weather. We advocate mirroring planned empirical investigati ons oh the computer to investigate which manipulations and predictions will best test a model. (C) 2000 The Association for the Study of Animal Behavi our.