Using face recognition software to understand Pilbara rock art

In a paper just published in the Journal of Archaeological Method and Theory, researchers at CRAR+M UWA have used some novel techniques to understand how people across the Pilbara depicted a special type of kangaroo – the fat-tailed macropod – in their rock art.  Using software that looks for shape differences, we have discovered many new things about this ancient art form. Working with our Indigenous collaborators, we were able to better understand what some of these differences might mean!

A research collaboration between Wintawari Guruma Aboriginal Corporation (WGAC) and CRAR+M recorded over 130 of these special, very old, kangaroo engravings – across low hills on either side of a major creek.  WGAC elder Michael Hughes showed the Nardang site to the researchers and helped explain many of the depictions. Jo McDonald from CRAR+M has led multiple visits to this site complex – where the recording work continues: we still have not found the site boundaries! Archaeology student Shiqin He, undertook the analysis for her Honours thesis at UWA, and the team collaborated with computational specialist Zulqarnain Gilani, who helped with the implementation of face recognition software he usually uses for medical research. Patrick Morrison’s love of statistics drew him into this project.

This analysis included many motifs recorded with Murujuga Aboriginal Corporation during field schools on the Burrup over the last 12 years. Here we describe the innovative research conducted to understand the stylistic characteristics and geographical distribution of these ancient depictions, using cutting-edge multivariate and geometric morphometric analyses (GMA).

Revealing the Fat-Tailed Macropod Motif

We aimed to identify the key attributes of the fat-tailed macropod motifs, explore stylistic similarities and differences across the Pilbara, and determine how these motifs are distributed across and placed within the landscape. By combining conventional archaeological approaches with advanced morphometrics, our researchers uncovered fascinating insights into the cultural and artistic expressions of early Pilbara artists:

What do they look like? We found a broad similarity across the Pilbara in how the fat-tails were depicted, suggesting a culturally proscribed schema, i.e. a widespread cultural practice or communication, among different groups. But there were also distinct stylistic differences between what is now coastal Murujuga and inland Pilbara assemblages, particularly in terms of size, technique, and internal decoration.

We identified 30 landmarks on each fat-tailed motif to allow the analysis to compare all the motifs.

Geometric Morphometric Analysis (GMA): The GMA provided a systematic, repeatable method for quantifying the variability in the shape of the motifs. This approach highlighted that the Pilbara artists depicted a range of macropod postures and gaits, reflecting their intimate knowledge of kangaroo behaviors.

While GMA couldn't find regional shape distinctions, it did find that differences in gait and posture were depicted across the whole region.

Some of the variations in posture and gait present in the fat-tailed macropod assemblage.

Principal Component Analysis (PCA): This multivariate approach showed that body length and ear length contributed significantly to how the motifs looks. Interestingly, the inland motifs, particularly those at the Nardang site complex looked more similar to each other, suggesting the influence of a single artist or a small group. This raises the possibility of a master-apprentice relationship in the production of these artworks.

Landscape and Visibility: The study found that most fat-tailed macropod motifs are located near water sources and are associated with other occupation evidence. Murujuga motifs were generally larger and more visible in the landscape, possibly serving as landscape markers or signals. In contrast, the smaller, less visible Nardang motifs might have been used for intimate, inter-generational learning or storytelling.

Murujuga motifs (top) are bigger and placed to be highly visible. The inland Pilbara fat-tails (bottom) are generally small and not easily visible in the landscape.

This analysis has continued investigations of a deep time Pilbara motif that has intrigued archaeologists for decades. Ken Mulvaney suggested this may be an extinct species, while Steve Brown suggested the tail was a stylistic convention which indicted its importance to the artists (see article for all references). Many of the motifs are also associated with geometric shapes, sparking further curiosity about potential cultural meanings. Many are also seen in hunting scenes, others with emu and echidna in the same style - revealing that this is a deeply storied landscape.

This image highlights the intriguing association between macropod motifs and geometric elements.

Cultural and Temporal Variations: The inland motifs appeared older than those at Murujuga, but many, particularly at Murujuga, show that subsequent artists continued to engage with this distinctive image through time. This temporal aspect provides insights into the changing use of rock art styles and cultural practices over time, but also shows how once rock art is 'in place' it continues to hold meaning and invoke a response from its audience.

This research collaboration has enhanced our understanding of the cultural significance and stylistic use of Pilbara rock art, highlighting ancient human-animal relationships and ongoing communication through art.  We are excited about what our ongoing research with Murujuga and Wintawari Guruma will discover next!