| Live from the Labs cnrs I international w 12 magazine Paleoanthropology BY Eddy Delcher w Using cutting-edge imaging technology, CNRS researcher Antoine Balzeau1 and fellow scientist Hélène Rougier2 have for the first time observed very thin bone structures found only in Neanderthal skulls.3 Singling out species-specific characteristics— known as autapomorphic traits—is essential for studying human evolution as it allows specialists to identify and classify the various types of prehistoric men. This task is made difficult by the rarity of these traits. One has nonetheless been identified recently in a small oval-shaped depression on the occipital bone forming the back of the skull of a Neanderthal specimen, believed to have died at about two years of age. Because the cranial bones of Neanderthal infants are only two to three millimeters thick (ten times thinner than adult skulls), researchers had so far lacked the equipment needed to carry out precise observations. Yet for this study, the team was given access to the AST-RX4 platform, which houses a high-performance scanner at the French National Museum of Natural History, in Paris. The researchers used microtomography, an imaging technique with micrometer resolution in which virtual cross sections of a sample are created via X-ray and then assembled into a 3D model. “Thanks to this unparalleled resolution, we were able to observe details we could not have seen before,” explains Balzeau. “We noticed a thinning of less than one millimeter of the occipital bone’s central layer, the diploë, which causes the depression.” “We had already analyzed dozens of adults, but it remained unclear whether this depression was specific to Brain Research By Eddy Delcher w CNRS researchers from the SBRI1 have developed an artificial brain able to learn languages and anticipate the end of sentences2. When listening to a sentence, the human brain processes the first few words and “guesses” what follows, based on speech patterns and visual cues. This accelerates the processing of information and improves communication. Based on research indicating that recurrent connections (loops) between neurons within the brain’s language centers play a crucial role in information processing, the research team has developed a simplified artificial brain composed of a neuronal network featuring locally recurring loops. This system enables the iCub, a childlike robot developed by the RobotCub Consortium,3 to predict the end of sentences and even learn new languages. “We programmed the recurrent network into the humanoid robot’s artificial brain so it could learn basic grammatical structures. When hearing ‘John throws the ball,’ it is able to identify the noun, verb, and object, and understands additional structures, such as ‘to Mary,’ that may follow,” explains Peter Ford Dominey, responsible for the project. iCub also identifies where the sentence could branch off into another direction, for example: “John throws the ball that broke the window.” As Dominey explains, “the system acquires the underlying grammar, and can generalize to new sentence types.” Because it only requires the basic grammatical features of a language to pick it up and understand it, the iCub can, for example, learn English, French, or Japanese. “We are now working on giving the iCub a notion of ‘self,’ a basic narrative identity that should make for more natural interactions,” concludes Dominey. 01. Stem Cell and Brain Research Institute (Inserm / Université Lyon-I / Inra). 02. X. Hinaut and P.F. Dominey , “Real-Time Parallel Processing of Grammatical Structure in the Fronto-Striatal System: A Recurrent Network Simulation Study Using Reservoir Computing,” PLoS One. 2013. 8:e52946. 03. www.robotcub.org Sorting Neanderthal Bones Contact information: SBRI, Lyon. Peter Ford Dominey > peter.dominey@inserm.fr q The iCub robot can learn languages in real time, simply by conversing with a human. Neanderthals,” says Balzeau. “Its identification in a child confirms its presence at an early developmental stage, making it an autapomorphic trait.” 01. Laboratoire Histoire naturelle de l’homme préhistorique (CNRS / MNHN). 02. California State University Northridge (US). 03. A.Balzeau and H. Rougier, “New Information on the modifications of the Neandertal suprainiac fossa during growth and development and on its etiology,” American Journal of Physical Anthropology, 2013. 151: 38-48. 04. Accès Scientifique à la Tomographie à Rayons X. Childlike Robot Learns Languages © P. Latron /Inserm Lyon Contact information: MNHN , Paris. Antoine Balzeau > abalzeau@mnhn.fr q Optical (upper) and X-ray (lower) images of a section of the diploë (left), the area of the occipital bone which is thinned in Neanderthal skulls. Paris © A. BALZEAU/CNRS/MNHN 1 cm
CIM30
To see the actual publication please follow the link above