The comparative study of the nervous system rests on the determination of valid homologies among nuclei and fiber tracts in different species (Ebner 1969; Northcutt 1969,1981; Neary and Northcutt 1983; Bullock 1984). Since ancestral forms are extinct, the criterion of descent can rarely be satisfied (Colbert 1969). Hence, a number of collateral criteria are often used to support common ancestral origin (Simpson 1961; Ghiselin 1969, 1976). These include relative position and shape, developmental history, afferent and efferent connections, and function (Gans 1969; Gans and Northcutt 1983; Rose and Wilczynski 1984). Serious objections have been made to each of these or similar criteria (Campbell and Hodos 1970). Thus, the relative position and shape of nuclei and fiber tracts can be a consequence of mechanical factors, such as size and shape of the head, without a clear relation to common ancestry (Ariëns Kappers et al. 1936; Wake etal. 1983). Developmental history alone may not yield reliable conclusions about homologies, since the ontogeny of a particular structure may differ substantially even between species with a common ancestor (Pearson 1972; Brunjes 1983). Patterns of neuronal connections are not proof of ancestral affinity, and they may vary between closely related species (Shatz 1977; Haight and Neylon 1981). Function is an erratic index of ancestry, since convergent or parallel adaptations are common in evolutionary history (Bullock et al. 1983; Rodieck and Brening 1983). Thus, any such similarity between cephalopod (Horridge 1974) and vertebrate eyes (Walls 1942) is not indicative of common ancestral origin (Darwin 1872; see also Bullock 1945).