Landmarking endocasts

Abstract

The brain and the braincase are tightly integrated during growth and development, and the cerebral cortex leaves morphological traces on the internal cranial surface. Paleoneurology studies the endocranial moulds – endocasts – in order to make inferences about brain evolution. The use of shape analysis based on landmark data is a current standard in morphometrics, but placing landmarks on endocasts involves inherent challenges. Here, we address endocast shape analysis focusing on the problems associated with landmarking their anatomical areas. First, we review the application of shape analysis in paleoneurology, from stereoplotting to sliding semilandmarks. Then, we address the problematics of landmarking endocasts’ smooth surfaces. Finally, we present a case study comparing the uncertainty in landmarking physical and digital endocasts. The mean absolute intra-observer error calculated on a digital sample is 0.86 mm, which is comparable with the error obtained by other authors on cranial, cerebral and endocast analyses. Landmarking on digital replicas, overall, displays larger uncertainty when compared with the physical casts, but the differences are not significant, and both methods give similar results when dealing with Hominoids’ interspecific variation. Although patterns of uncertainty seem to be largely idiosyncratic, larger errors are usually found on the parietal surface. This study concerns only intra-observer error. Anatomists can have a different perception or interpretation of the cortical references, and landmarking can be improved by performing joint sampling based on the agreement of different experts.