The research team, led by Kalangi Rodrigo, Eva Francesca Martellotta and Marco Peresani, assembled the corpus through a systematic literature survey using Google Scholar and Scopus, supplemented by manual screening of reference lists in key review papers and site monographs. The database spans 13 decades, from 1861 to 2026, and includes publications written in English, German, Italian, Spanish and French. Only peer-reviewed journal articles, edited volumes and monographs were included; unpublished theses, conference abstracts and non-peer-reviewed reports were excluded. Each publication was manually sorted into a relational database with binary marker columns recording bibliographic metadata, tool categories addressed, analytical focus and reported methodological approaches.

The authors are careful to frame the dataset as a defined sampling frame rather than an exhaustive census. They disclose that automated retrieval combined with reference chaining tends to under-sample pre-1980 and non-digitised works, particularly those embedded in general site monographs or published in languages other than English. Some regionally significant records from Croatia, Slovenia, Hungary, Slovakia and the Swabian Jura sit at or beyond the margins of the frame. Publication frequencies, they stress, capture where research activity has been concentrated, not the behavioural importance or actual distribution of bone tools in the past. The statistical analysis combined descriptive statistics, Fisher’s exact tests for two-by-two associations among binary variables, and logistic regression models with publication decade as a predictor. P-values were interpreted as continuous measures of evidence rather than binary thresholds, and where multiple tests were conducted on the same dataset, adjusted p-values were computed using the Benjamini–Hochberg false discovery rate procedure. All analyses were performed in the R programming environment.

The geographic pattern is markedly uneven. France accounts for 44 of the 137 studies, or 32.1 percent, and Italy for 23 studies, or 16.8 percent, meaning these two countries together represent nearly half of the corpus. Spain appears in 14 studies, or 10.2 percent, while Germany and Russia/Ukraine each contribute 10 studies, or 7.3 percent. The analysis also found that retoucher-focused research is not randomly distributed geographically: papers coded for Spain were significantly more likely to include retouchers, with an odds ratio of 3.92. The authors attribute this concentration partly to the well-stratified karstic cave sequences of France, Italy and Iberia, which favour bone survival, and partly to longer and more intensive excavation traditions in those regions.

Retouchers emerge as the continuously foregrounded discrete category within the Middle Palaeolithic focus of the corpus. In terms of stated keywords, retouchers are an explicit focus in 44 of 137 studies, or 32.1 percent, second only to the broad framing term ‘bone tools’ at 52 studies, or 38.0 percent. Studies that identify retouchers are more likely to report use-wear analysis, with an odds ratio of 2.45, although the association sits at the edge of significance after correction for multiple comparisons. The authors interpret this association as reflecting the suitability of retouchers for trace-based validation rather than evidence of a broad, integrated technological research programme, since retouchers bear discrete impact marks that are macroscopically recognisable and experimentally replicable.

The temporal analysis documents a genuine methodological transformation. The odds of reporting a functional analytical package—use-wear analysis, techno-functional analysis, experimental archaeology or residue analysis—increase by a factor of 1.54 per decade, while typological and descriptive approaches decline with an odds ratio of 0.85 per decade. Projected across the six decades from the 1960s to the 2020s, this corresponds to a cumulative odds multiplication of roughly 3.8, meaning that by the 2020s a study was nearly four times more likely to report functional methods than a comparable study from the 1960s. Notably, stratifying the models by tool focus shows that both trends are carried by the non-retoucher literature, whereas retoucher-coded studies show no temporal trend in either direction.

The historical trajectory behind these patterns is long and contentious. Early reports of bone artefacts in the nineteenth century were often made without rigorous criteria for distinguishing human modification from natural taphonomic alteration. The term ‘retouchoir en os’ appeared in 1889 when a specimen from Capitan’s collection was catalogued as a bone retoucher at the Exposition Universelle in Paris, and Henri Martin’s early twentieth-century work at La Quina reported large-herbivore bones bearing marks interpreted as human modifications. In 1925, Louis Siret’s experimental replication using equid phalanges helped reframe these ‘impressed bones’ as passive retouchers used in lithic edge modification. That early confidence provoked a powerful backlash. From the 1930s and 1940s, researchers used animal behaviour, experimentation and ethological analogy to show that carnivores can accumulate and modify bones in ways that mimic tools. Koby’s critique of the ‘Alpine Palaeolithic’ argued that polish and arrangement could result from dry transport and bear trampling, while Pei’s work at Zhoukoudian emphasised that many purported tools lacked clear anthropogenic modification once animal and environmental agents were considered. The taphonomic turn of the 1980s, driven by C. K. Brain’s cave taphonomy and Binford’s actualistic programme, led to the re-evaluation of many alleged early tools and produced an enduring default scepticism toward pre-Upper Palaeolithic bone artefacts, sometimes described as ‘Binford syndrome’. Re-evaluations of specific assemblages, such as d’Errico and Villa’s reassessment of material from Bois-Roche and Pech de l’Azé II, attributed previously identified tool modifications to carnivores and other natural agents.

Since the 2000s, an interdisciplinary standard has converged on combining taphonomic screening, microscopic trace studies and experimental replication to establish bone tools in early contexts. Advances in microscopy and imaging, including scanning electron microscopy, computerised tomography, X-ray microtomography and 3D surface methods, have expanded the analytical toolkit. At Neanderthal sites such as Chagyrskaya in the Altai and Abri Lartet in France, assemblages of approximately 1,200 modified bone artefacts were analysed using X-ray microtomography to visualise internal deformations and crack patterns. Publication output reflects this acceleration: the 2010s account for 44 of the 137 studies, or 32.1 percent, and the 2020s a further 31 studies, or 22.6 percent, so more than half of the corpus has been published since 2010. Yet the deeper recording practices tell a more selective story. Taphonomic analysis is reported in 98 of 137 studies, or 71.5 percent, rising from 38.1 percent before 1980 to 90.3 percent in 2020 to 2025. By contrast, only 55 studies, or 40.1 percent, employ the broader functional analytical package, and high-resolution recording—metric characterisation of functional areas, 3D models, geometric morphometrics or angle and edge documentation—is present in just 43 studies, or 31.4 percent. Metric characterisation of functional areas appears in 28.5 percent of studies, 3D models in 5.8 percent, geometric morphometrics in a single study, and angle or edge recording in 2.9 percent. ZooMS, a biomolecular species identification method, appears in only about 4 percent of the corpus, partly because it involves destructive sampling that may alter microscopic surface features used for functional inference, and partly because its success depends on uneven collagen preservation.

The authors argue that the issue is not simply one of bias but of epistemic filtering: certain behaviours become more visible because they align with dominant validation strategies, while others remain under-represented. Over time, they warn, this imbalance may narrow reconstructions of Palaeolithic bone technology toward activities closely aligned with lithic production, inadvertently downplaying other domains of bone use such as hide processing, plant working, composite tool manufacture or percussive tasks unrelated to stone-tool maintenance. Retouchers, they suggest, have functioned as a methodological centre of gravity, drawing advanced analytical techniques toward them while other osseous categories such as bevelled tools, smoothers, pièces esquillées, unworked intermediate tools and bone points, which may involve different operational chains and more ambiguous use traces, remain comparatively under-analysed.

To counter these tendencies, the study proposes a minimum set of reporting recommendations: documentation of recovery and screening protocol including mesh size and the treatment of unidentifiable long-bone fragments, taxonomic and skeletal-element identification, blank state distinguishing unmodified from modified supports, systematic recording of use-wear and impact scars with the imaging method specified, metric characterisation of functional areas, and an explicit mapping of terminology to established synonyms. The authors also call for extending high-resolution functional analyses beyond retouchers, broadening geographic and chronological representation, and adopting more standardised reporting practices. They acknowledge concrete obstacles, including the uneven digitisation of older and non-English literature, the dependence of biomolecular methods on still-incomplete reference databases, and the continued absence of shared reporting standards across the field.

Subject of Research: Attention bias and methodological change in European Middle Palaeolithic bone-tool research

Article Title: Retouchers at the centre: quantifying attention bias and methodological change in middle palaeolithic bone-tool research in Europe

Article References: Rodrigo, K., Francesca Martellotta, E., & Peresani, M. (2026). Retouchers at the centre: quantifying attention bias and methodological change in middle palaeolithic bone-tool research in Europe. Archaeological and Anthropological Sciences, 18(11), Article 222. https://doi.org/10.1007/s12520-026-02588-1

Image Credits: AI Generated

DOI: 10.1007/s12520-026-02588-1

Keywords: bone tools, retouchers, Middle Palaeolithic, Neanderthals, use-wear analysis, taphonomy, archaeology, lithic technology, research bias, Europe, bibliometric analysis, experimental archaeology