Anthropology, University of Vienna, Statistics, University of Washington
Saturday 1:00-1:15, Galleria North
During the decade that the disciplinary home of geometric morphometrics has been physical anthropology, advances of pedagogy and software notwithstanding, our toolbox has ossified. I mean this literally: we typically exploit mainly information from surfaces of bones, as encoded by the location of "points that correspond," and typically we are content with ordinations. But the precision of our data sources is now too high to claim credibility for this computed homology -- the point-correspondences lack all biological realism -- and ordinations have become inadequate to our scientific context. For information from bony surfaces to be biologically relevant it must be filtered and interpreted using information from the other kinds of machines that assess biological function instead of dead, ossified form. Tomorrow's morphometric analysis will have little use for either Procrustes distance or bending energy, as neither has any biological meaning. They will be superseded by new pattern-theoretic approaches that deal with biological form as only one descriptive channel among the many others serving the disciplines of the living organism, such as bioengineering, physiology, auxology, or the cognitive sciences. These data fusions, in turn, require novel approaches to multivariate data superseding the previous century's emphasis on static covariance structures by random walks, random textures, and other spatiotemporal process models. In summary, it is time for a new morphometric toolkit subordinating the variable aspects of geometry to the more important channels whereby quantitative anthropologists of the future will assess the meanings of form for function, growth, and natural selection.
NSF grant A55204, "Fossils and Morphometrics," to J. Felsenstein and F. Bookstein.