This essay is in two parts. In the first part, consideration is given to specific issues raised in the preceding article, stressing two considerations. First, although Dr. Hughes disclaims an essentialist position, he in fact argues consistently for a classification of nerve cells based on key, essential features; except where, briefly, he argues for a numerical taxonomic approach. He apparently has not understood the hypothetico-deductive approach proposed by us. Second, he has not given consideration to important biological aspects of the problem of cell classification, in particular to the significance of variation of properties both within groups of cells and between corresponding groups. In the second part we argue that underlying the exchange of specific criticisms are differences in scientific methodology. We attempt to identify these differences, and to relate them to the present discussion. We hope that this will help focus subsequent discussion on the central issues involved and we argue for increased awareness on the part of neurobiologists of our own presuppositions about how science works.