SCIPRESS FORMA
Forma, Vol. 11 (No. 2), pp. 103-114, 1996
Original Paper

A Neural Network Analysis of Navigational Learning in Homing Pigeons

Michael A. Jones and Verner P. Bingman

Department of Psychology, Bowling Green State University, Bowling Green, Ohio, 43403, U.S.A.

(Received October 25, 1995; Accepted January 10, 1996)

Keywords: Backpropagation, Cognitive Maps, Bird, Hippocampus

Abstract. Abstract. In addition to their well known navigational map, homing pigeons can also rely on familiar landmarks to navigate home. However, because of the difficulty in manipulating environmental landmarks normally used by homing pigeons, little is known about how pigeons use such landmarks for navigation or how they learn to do so. In an attempt to generate some hypotheses on navigational learning in homing pigeons, we have developed a neural network model to examine how landmark navigational learning may be influenced by specific environmental variables. The network quickly learned the spatial task it was presented. The network analysis suggested that landmark navigational learning may be influenced by both the distribution and the number of landmarks in space. The intuitive appeal of these predictions suggests that the type of neural network analysis presented here may indeed reflect natural learning processes normally carried out by animals and, at the very least, that such networks are a powerful tool for modeling learning processes that would be difficult to study experimentally.