Predictability or contingency lies at the heart of the associative process, the CS's ability to predict the US determines the strength of learning. Important candidate properties for encoding predictability may lie in the US pathway. Previous work in this lab has demonstrated that consistent -5 prediction of a US prevents """"""""dropout"""""""" of a critical component of one US pathway, the Retzius cells, the dominant serotonergic neurons of the leech CNS. Throughout training predicted USs continue to elicit a brisk barrage of action potentials in these cells. Recurring unpredicted USs degrade both the learning and the response of the Retzius to the US. A battery of cell ablation, physiological, biochemical and structural methods will be applied to identify and characterize the cellular and network interactions which encode predictability in these cells. The causal contribution of learning related changes to the behavior will be assessed in a semi-intact preparation. A significant recent finding from our lab is that at least two US pathways contribute to associative learning. A second Nocioceptive (N) cell pathway persists after the elimination of the Retzius cell pathway. The observation of multiple US pathways raises a host of issues concerning CS- US convergence and the functional significance of distinct US pathways. Experiments are proposed to further identify and characterize US pathways and assess the role of their modulation in associative learning. The simple nervous system of the leech and the increasingly well defined CS and US pathways underlying learning make the leech particularly attractive for modeling studies. This is particularly timely since recent experiments in our lab have indicated that network properties are important for ecoding contingency. To begin this analysis we will use the realistic modeling approach, MARIO, to test the functional contributions of learning related changes to associative learning. The special constellation of experimental advantages presently available in the leech for the study of associative learning, especially the ability to connect learning related changes with their behavioral consequences, will likely yield new insights into the associative process and its underlying mechanisms.