Neural network able to distinguish between various Stone Age tools
Image credit: Dreamstime
A study led by Max Planck Institute for the Science of Human History researchers has demonstrated that machine learning algorithms can be used to identify what differentiates prehistoric artefacts from different periods.
The Stone Age spanned millions of years of human development, with the transition between the Middle Stone Age (MSA) and Later Stone Age (LSA) believed to have marked a major cultural shift. However, distinguishing between these two periods is not straightforward.
MSA toolkits began to appear around 300,000 years ago – around the same time as the first Homo sapiens fossils – and continued to be used 30,000 years ago. However, from 67,000 years ago, changes in stone tool production indicated a shift in behaviour which continued into the recent past. These distinct toolkits are labelled LSA.
Despite this difference, the transition between MSA and LSA is not considered a linear process, but rather one that occurred at different times in different places. Despite the importance of understanding this transition in understanding our cultural history, distinguishing between MSA and LSA is challenging.
“Eastern Africa is a key region to examine this major cultural change, not only because it hosts some of the youngest MSA sites and some of the oldest LSA sites, but also because the large number of well-excavated and dated sites make it ideal for research using quantitative methods,” said Dr Jimbob Blinkhorn.
“This enabled us to pull together a substantial database of changing patterns of stone tool production and use - spanning 130,000 to 12,000 years ago - to examine the MSA-LSA transition.”
Blinkhorn and his colleagues examined the presence or absence of 16 alternate tool types across 92 stone tool assemblages. Rather than focusing on these individually, however, emphasis was placed on the “constellations” of tool forms which frequently occur together.
The researchers used an artificial neural network to train and test models which differentiate MSA and LAS toolkits. While the use of neural networks is widespread in research, it is so far been more limited in archaeological research.
“[Artificial neural networks] have sometimes been described as a 'black box' approach, as even when they are highly successful it may not always be clear why,” said Dr Matt Grove, a University of Liverpool archaeologist. “We employed a simulation approach that breaks open this black box to understand which inputs have a significant impact on the results. This enabled us to identify how patterns of stone tool assemblage composition vary between the MSA and LSA and we hope this demonstrates how such methods can be used more widely in archaeologist research in the future.”
The study demonstrated that MSA and LSA assemblages can be distinguished by a neural network, based on the constellation of artefact types found within an assemblage.
“The combined occurrence of backed pieces, blade and bipolar technologies together with the combined absence of core tools; Levallois flake technology; point technology, and scrapers robustly identifies LSA assemblages, with the opposite pattern identifying MSA assemblages,” said Blinkhorn. “Significantly, this provides quantified support to qualitative differences noted by earlier researchers that key typological changes do occur with this cultural transition.”
The researchers also used neural networks to examine the chronological differences between older and younger MSA assemblages, with 94 per cent accuracy.
Next, they plan to expand the use of artificial neural networks to dig deeper into the regional trajectories of cultural change in the Stone Age in Africa. “The approach we’ve employed offers a powerful toolkit to examine the categories we use to describe the archaeological record and to help us examine and explain cultural change amongst our ancestors,” Blinkhorn said.
Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.