Integration of microscopy, geometric morphometrics and machine learning classification algorithm for the identification of hand preference from stone tools - CNRS - Centre national de la recherche scientifique Accéder directement au contenu
Communication Dans Un Congrès Année : 2018

Integration of microscopy, geometric morphometrics and machine learning classification algorithm for the identification of hand preference from stone tools

Résumé

Hand preference is related to areas of the brain linked to several critical functions such as language. Therefore, identifying hand preference in fossil hominids allows describing hand preference evolution but is also essential to characterize the origin and development of complex cognitive capabilities during human evolution. Stone tools, used during all periods and found extensively in archaeological sites, are likely to bear information about the hand which held them during repetitive activities. They are therefore of prime interest to provide new data about hand preference during evolution. However, only few studies, mainly focused on stone tool production, have been carried out with this purpose. Here we propose to focus on a repetitive activity which is very likely to embed better the information concerning hand preference: tool use. We aim at establishing an experimental protocol and a reference collection to build a model allowing determining the hand holding the stone during use. Monitored experiment has been carried out and both classical use-wear analysis procedures and new quantitative method based on geometric morphometrics on used edges were performed. These techniques provided complementary information to select the best parameters to be used for hand preference inferences. Applying k-nearest neighbors algorithm, we were able to build a model with more than 76% accuracy in classifying stone tools used by right or left handed subjects. The method is still perfectible but already promising and the addition of new stone tools in the reference collection and of new parameters in the model is likely to increase the good classi cation rate.
Fichier non déposé

Dates et versions

hal-03970688 , version 1 (02-02-2023)

Identifiants

  • HAL Id : hal-03970688 , version 1

Citer

Alice Rodriguez, Maria Gema Chacon, Raphaël Cornette, Emmanuelle Pouydebat, Marie-Hélène Moncel, et al.. Integration of microscopy, geometric morphometrics and machine learning classification algorithm for the identification of hand preference from stone tools. XVIII° Union International des Sciences Préhistoriques et Protohistoriques (UISPP) World Congress, Jun 2018, PARIS, France. ⟨hal-03970688⟩
20 Consultations
1 Téléchargements

Partager

Gmail Facebook X LinkedIn More