Mapping the field of cultural evolutionary theory and methods in archaeology using bibliometric methods

The paper investigates how cultural evolutionary theory (CET) and associated methods have been applied in archaeology from 1981–2021. It situates archaeology within broader debates on Darwinian approaches to culture, tracing a shift from early selectionist evolutionary archaeology to dual-inheritance theory frameworks. The authors note questions about the appropriateness of biological analogies for culture, yet argue many cultural traits show evolutionary properties requiring tailored methods. The study’s purpose is to map the research front of CET in archaeology using bibliometric methods, identify clusters and trends, and assess methodological trajectories, with special attention to phylogenetics and geometric morphometrics. The motivation is to visualise development, reveal gaps, and evaluate archaeology’s contribution to cultural evolution scholarship.

The authors review the historical and conceptual background of applying evolution to archaeology, from 19th–20th century stage-progressive notions to late-20th century evolutionary archaeology that initially emulated genetics-like models (Dunnell) and then shifted to dual-inheritance theory (Shennan and others). They summarise debates on whether culture evolves in a way analogous to biology and highlight differences from Mendelian inheritance. Prior syntheses (Mesoudi et al. 2006; Mesoudi & O’Brien 2009) positioned archaeology as focusing on long-term macroevolutionary patterns. Existing reviews of CET in archaeology (Feinman 2000; Riede 2010; Garvey 2018; Prentiss 2019; Walsh et al. 2019) and a bibliometric study of cultural evolution broadly (Youngblood & Lahti 2018) are noted. The paper also remarks on the rise and fall of memetics and on the growing but still unrealised potential for archaeology to advance CET, particularly in macroevolutionary domains analogous to palaeobiology.

Data source and scope: The authors queried the Web of Science Core Collection (WoS) on March 14, 2022 for publications in archaeology and anthropology with topics including “cultural evolution” and “cultural transmission,” limiting to English-language records through 2021. The query yielded 674 entries (earliest 1981) with full bibliographic metadata and cited references exported in .bib format. They acknowledge WoS coverage biases for humanities/social sciences and English-language dominance. Data preparation: Using R 4.2.1 and the bibliometrix/bib2df packages, author names were standardised. Author-assigned keywords, titles, and abstracts were extracted; uni- and bi-grams appearing in ≥1% of articles were retained (WoS-assigned keywords were excluded). A hierarchical thesaurus was built mapping terms to first-order categories and broader second-order meta-categories (e.g., Methods, Topics). Occurrence and incidence matrices of thesaurified keywords by article were generated. Temporal trends were visualised as per-year fractions of articles containing second-order categories (loess smoothing in ggplot2 where appropriate). Bibliographic coupling network: A weighted, undirected network of articles was constructed based on shared references (bibliographic coupling). Edge weights used cosine similarity of shared reference vectors. No similarity threshold was applied; the largest connected component (630 articles) was retained. Community detection used the Louvain algorithm (igraph), with network and cluster densities computed (density = ratio of existing to possible edges). Closeness centrality was calculated for all nodes; one-way ANOVA tested for differences among clusters’ mean closeness (p < 0.001), followed by Bonferroni-adjusted pairwise t-tests. Cluster characterisation: To relate clusters to content, Pearson’s chi-squared test (χ² = 2394.7, df = 372, p < 0.001) was applied to thesaurified keyword incidences by cluster; standardised residuals identified keywords overrepresented per cluster. Weighted log odds ratios (tidylo; Monroe et al. 2008) further highlighted differential keyword usage across clusters. Visualisations used ggraph/ggplot2; additional R packages for analysis and plotting are listed. Sensitivity to edge-thresholding was checked (clusters remained coherent under thresholds). Analytical focus: The study reports corpus composition (authorship, outlets, geography), temporal trends in methods/topics, network structure (modularity, density), cluster sizes/densities/centralities, and semantic characterisations of clusters via statistical association of keywords.

Corpus and outlets: N = 674 publications (1402 authors); ~40% (n = 271) single-authored; multi-authored papers average 2.89 co-authors. First-author affiliations: ~39% USA, ~15% UK, ~6% Germany, ~5% China; others include Australia (4.3%), Canada (3.9%), Spain (2.9%), France (2.7%), Israel (1.8%), Italy (1.7%). Top outlets include Journal of Archaeological Science (n=30), Quaternary International (25), Journal of Anthropological Archaeology (25), Journal of Archaeological Method and Theory (24), Phil. Trans. R. Soc. B (24), PNAS (20), American Antiquity (14), Current Anthropology (14), PLOS ONE (13), Journal of Human Evolution (13). Temporal patterns: Publications grow markedly after 2000, with most in the last decade. Period focus: Palaeolithic terms are earliest (from 1994) and most frequent, with increasing fraction over time; Neolithic and Bronze Age follow. Keyword trends: “Cultural evolution” and “cultural transmission” dominate; “social learning” peaks circa 2010–2015. Gene–culture coevolution and evolutionary psychology occur mostly in early 2000s. Memetics appears briefly in early 2000s and ~2010, then disappears. Topic emphases include climate/catastrophes (peaking around 2016 to ~40% of articles), ecology/environment (~1/3 of articles), flora (~1/5), fauna (~1/10). Behaviour and cognition remain ~20% annually. Human evolution rises; genetics-related keywords decline. Greater focus on temporality than spatiality; many studies concern hunter-gatherers/foragers, mobility, subsistence, demography; metallurgy salient post-Palaeolithic. Methods trends (relative occurrence over years): Modelling/simulation keywords most frequent but decline in last five years. Phylogenetic/cladistic methods show early dynamism then sustained decline/stagnation. Geometric morphometrics remains steady to slightly rising in the last decade. Theoretical approaches trend downward slightly. Typology/taxonomy shows a positive trend. Quantitative/statistical methods increase from ~10% (2000) to at least ~25% by late 2000s and remain common. Network structure: The bibliographic coupling network (630 connected articles) has modularity ~0.155 and overall density 0.224. Louvain algorithm recovered seven clusters; main five analysed: n1=148, n2=92, n3=150, n4=154, n5=81 (n6=7, n7=2 not analysed). Cluster densities: d1=0.306, d2=0.306, d3=0.691 (highest), d4=0.496, d5=0.260 (lowest). Mean closeness centralities: c1=0.620, c2=0.571 (lowest; peripheral), c3=0.699, c4=0.710 (highest; core), c5=0.646. ANOVA shows significant differences (p<0.001); cluster 4 significantly higher than all except cluster 3 (Bonferroni-adjusted pairwise tests). Over time, network density declines in the most recent bin, indicating increased diversity (d1981–2001=0.204; d2002–2006=0.261; d2007–2011=0.321; d2012–2016=0.304; d2017–2021=0.212). Cluster characterisations (via keyword associations and log-odds): - Cluster 4 ("Complex human behaviour cultural transmission Early Stone Age"): Core of network; Africa/Europe; Palaeolithic/African Stone Age; lithics/handaxes; quantitative methods, morphometrics, typology, modelling; topics include symbolic communication, behaviour, art, primatology, fauna, demography, cognition, hunter-gatherers, human evolution, design. Sub-clusters: Early Stone Age lithics and behavioural ecology; experimental transmission; symbolic/ornaments and cognitive evolution; population size/connectedness and cultural complexity. - Cluster 3 ("(Foundational) cultural evolutionary theory methods"): Theoretical/methodological core; emphasis on cultural transmission (copying error, drift, innovation), population size effects; early foundational phylogenetic/cladistic approaches; later refinement and applications to specific questions; includes lithic morphometrics; smaller niches include ceramics and ethnoarchaeology. - Cluster 5 ("Ethnoarchaeology|cultural complexity chiefdoms early states"): Meso-/South America and islands; complexity, cultural evolution; topics include climate change, subsistence, agricultural intensification, state formation; aligns with social evolution traditions, loosely tied to evolutionary archaeology sensu stricto. - Cluster 1 ("Niche construction theory/psychology|gene-culture co-evolution"): Broad CET framework with many ethnographic/anthropological topics (evolutionary psychology, behavioural ecology, niche construction, social learning, cognition, ethics/morality, religion, linguistics, primatology, economics). Lowest density suggests heterogeneity and multidisciplinarity. Notably houses the most applications of comparative phylogenetic methods (often linguistic/cultural trait trees) rather than archaeological artefacts. - Cluster 2 ("Climate change/social adaptations population density Neolithic"): Neolithic focus; East/Central/Northern Asia; methods from geosciences and absolute dating; radiocarbon-based population proxies, paleoclimate impacts, diffusion of Neolithic, settlement distributions; peripheral to CET core, predominantly archaeological. Macro vs. micro emphasis and phylogenetics: Despite early prominence, phylogenetic applications in archaeology have declined relative to neighbouring fields (linguistics/anthropology). Within this corpus, most Bayesian/cladistic phylogenetic studies relate to languages or ethnographic material culture (cluster 1). Few studies apply advanced model-based Bayesian phylogenetics directly to artefacts; maximum parsimony approaches are more common. Recent palaeobiological advances—fossilised birth–death process, and Bayesian treatment of continuous characters—offer pathways to integrate geometric morphometric artefact data directly into phylogenetic analyses, potentially revitalising macroevolutionary archaeology.

The study maps CET in archaeology as a growing, diversifying field with distinct intellectual clusters. Archaeology remains somewhat peripheral to CET’s core journals and conversations, often publishing in archaeology-focused venues. The corpus reveals three article classes: (1) theory/method development adapting biological frameworks to culture (often simulations and formal models), (2) archaeology-tailored CET methods applied to archaeological questions (showing a growing empirical base), and (3) studies using CET as a broad framework to address primarily archaeological issues, reflecting a vernacular use of “cultural evolution” for change over time. While microevolutionary approaches (experiments, transmission models) are well represented, macroevolutionary archaeology—expected to parallel palaeobiology—appears underdeveloped, particularly in cultural phylogenetics applied to artefacts. The decline in archaeological phylogenetic applications may stem from lack of standardised artefact classifications, methodological conservatism, and the recency and complexity of Bayesian phylogenetic tools. However, palaeobiology’s methodological innovations (e.g., fossilised birth–death models, continuous-character Bayesian phylogenetics) and the maturation of geometric morphometrics create opportunities to integrate shape data into phylogenetic analyses, aligning archaeological practice with advances in evolutionary linguistics and enabling macro-scale CET inquiries.

This mapping review provides a systematised bibliometric overview of CET in archaeology (1981–2021), identifying core clusters, temporal trends, key methods, and thematic foci. Quantitative and computational approaches are central, but the specific suite of phylogenetic methods foundational in palaeobiology has stagnated in archaeological applications, while geometric morphometrics has risen. The authors argue that recent Bayesian phylogenetic advances—especially incorporation of continuous shape characters and fossilised birth–death processes—can bridge morphometrics and phylogenetics, potentially enabling a renaissance in artefact phylogenetics and strengthening macroevolutionary research in archaeology. Future research should: (1) adopt model-based Bayesian phylogenetics for artefact data, (2) integrate geometric morphometric descriptors directly into phylogenetic inference, (3) develop shared artefact classification standards to facilitate comparative analyses, and (4) engage more with CET’s core venues to enhance cross-disciplinary exchange.

- Data source bias: WoS underrepresents humanities/social sciences and non-English publications; coverage begins in 1975 and omits many key books/chapters, especially in archaeology and CET. Results reflect general trends rather than exhaustive coverage. - Language restriction: Only English-language publications analysed; non-English CET literature is underrepresented. - Search-term scope: Limited to “cultural evolution” and “cultural transmission,” potentially missing relevant CET work using different terminology. - Methodological constraints: Bibliographic coupling assumes shared references reflect topical similarity; may connect articles citing different aspects of a source. Community detection (Louvain) and lack of edge-thresholding can yield large, weakly connected clusters; interpretation requires caution. Topic reconstruction and clustering algorithms have known limitations. - Analytical scope: Two smallest clusters (n=7, n=2) not analysed in detail. Keyword assignment relies on author terms and text-mined n-grams; WoS-assigned keywords were excluded due to inaccuracy. - Generalisability: Findings focus on anglophone archaeology and may not capture CET usage beyond this community or outside WoS-indexed venues.