Algorithm for Classiication of Neurons with Application to Amphibian Data Zentrum F Ur Kognitionswissenschaften Universit at Bremen Algorithm for Classiication of Neurons with Application to Amphibian Data

Zusammenfassung We present a general algorithm to classify neurons according to their ring rates. We apply the algorithm to ring rates of tectum-opticum neurons of the salamander Hydro-mantes italicus. These data were classiied by inspection by Roth 7]. Whether separate groups exist at all remained uncertain. Our algorithm now provides a good survey over the data and quantitative results concerning the data-structure. We show that it is advantageous to transform the data appropriately before classiication. Valuation criteria suggest that the data form a continuum. This is in good accordance with the most recent biological results. Furthermore, in the continuum we nd accumulations. The algorithmic classiication of the accumulations basically connrms the classiication by Roth. To every neuron group of Roth, there is a corresponding group of the algorithm. Moreover, the algorithm nds two new groups.