Résumé
Besides size increase, cortical reorganization is one of the most critical processes having affected human brain evolution. In the hominin fossil record, a virtual "early Homo endocast" is expected to display a characteristic neuroanatomical pattern including, among other features, a distinct Broca's area, the absence or only faint presence of a lunate sulcus positioned posteriorly, a unique orbitofrontal morphology, a relatively complex configuration of the middle meningeal network (Tobias, 1987; Holloway et al., 2004). Tracking the early appearance and full establishment of these derived features in the hominin fossil record should contribute the assessment and understanding of the inter-species evolutionary relationships within the human lineage and potentially allow the identification of the paleodeme from which early Homo emerged.However, together with an expected degree of intra-species variation having occurred during hominin radiation, the confounding effects of non phylogenetic-based resemblances (homoplasy), and the unpredictable as much as unavoidable impact of taphonomy on the fossilisation process, such research is complicated by the nature itself of the investigated structures. For example, the original description of the Australopithecus sediba (spec. MH1) highlights a developmental degree of the frontal lobe closer to the human condition in association with a rather australopith-like cranial capacity and convolutional pattern, thus suggesting a mosaic evolutionary pattern of the neuroanatomical diagnostic features (Carlson et al., 2011). However, a recent publication contests the identification of Homo-like features in A. sediba and rather suggests a common australopith endocranial organization (Falk, 2014).Paleoneurology has recently enlarged its traditional investigative toolkit by integrating descriptive morphology with advanced methods of high-resolution 3D imaging and computing suitable for the subtle characterization of the fossil morphoarchitecture (Bruner, 2014). In this perspective, we engaged in the computer-assisted revision of the endocranial structural organization of South African nonhuman hominin endocasts, with special attention to the quantitative assessment of the sulcal variation pattern and architectural asymmetries.The fossil specimens investigated so far represent three relatively well-preserved A. africanus endocasts: the Taung child and the specimens Sts 5 and Sts 60 from Sterkfontein Member 4. Based on virtual reconstructions obtained from high-resolution tomography (Taung, Sts 5) and surface scanning (Sts 60), we combined a method of landmark-free registration (Durrleman et al., 2012; Dumoncel et al., 2014) and a method of 3D endocranial shape asymmetries to investigate topographic differences in morphostructural organization and, more specifically, in lobe conformation, and quantitatively characterize intra- and inter-specific variability. Cerebral petalias and petalial patterns were also assessed. Finally, a semi-automatic methodology was developed to automatically detect, extract and compare the sulcal topographic organizations.Here we compare the evidence extracted from the three South African fossil endocasts to the figures from a representative sample of extant humans, bonobos and chimpanzees, and also discuss the value, limits and perspectives of our experimental analytical protocol.