Recent findings of activation of language networks in the brain during stone tool manufacture support hypotheses aboutthe co-evolution of language and technology. Our study replicates these findings and demonstrates that distinct toolmakingbehaviors and levels of expertise affect how reliably these networks are activated. Subjects, including expert toolmakers(n = 7) and untrained participants (n = 10), watched naturalistic videos of an expert toolmaker making three technologiesand imagined themselves performing the same actions as the toolmaker while being scanned. We performed event-relatedGLM analyses on our data, focusing on activation during observation and flaking. All technologies recruited networksinvolved in language production and comprehension, including IFG, vPMC, dPMC, SPL, IPL, and pMTG. Flaking en-gaged language networks more reliably than observation. Our study considers whether expertise is required for Oldowan,Acheulean, and Levallois comprehension by exploring the extent to which activation in language networks increases withtool complexity.

Our study gathered Diffusion Tensor Imaging data to compare anatomical connectivity in expert stone toolmakers with naïve subjects with no prior toolmaking experience. The introduction of stone tool technology marked a shift in the evolution of human cognition, as early hominins gradually developed their capacity for complex hierarchical action planning and coordination. It is hypothesized that other abilities requiring these same capacities, like language, co-opted this neurocognitive scaffolding. Similarities in connectivity between experts and novices thus may be explained by the involvement of these networks in language or by a ubiquitous human competence in everyday tool use. Differences are likely explained by the increased complexity of the tool types experts make and use. These differences would support findings from a previous analysis within this study that found tool types of varying complexity (Oldowan, Acheulean, Levallois) differentially activated language networks for subjects with different levels of expertise.