Regarding compass response functions for modeling path integration

Recently Haferlach, Wessnitzer, Mangan and Webb (2007) reported producing a novel neural model of the animal behaviour known as path integration, a navigation process requiring a compass and odometer. Here we comment on comparisons that that paper makes with our earlier work, (Vickerstaff & Di Paolo, 2005) where we also evolved a neural model of path integration. The authors make prominent mention of the similarity of their compass sensors to the known properties of the polarisation-sensitive (POL) neurons found in insects. They give the impression that their compass response function, expressed as a dot-product equation (p. 274, eqn. 1, see below), has a closer similarity to POL neurons than does the cosine function we employed AgResearch Ltd, Lincoln Research Centre, Cnr Springs Road and Gerald Street, Private Bag 4749, Christchurch 8140, New Zealand Department of Zoology, School of Biological Sciences, University of Canterbury, Private Bag 4800, Christchurch, New Zealand. [email protected] Department of Informatics, School of Science and Technology, University of Sussex, Brighton, BN1 9QH, UK. [email protected]