[Paper Review] A Non-Phylogenetic Conceptual Network Architecture for Organizing Classes of Material Artifacts into Cultural Lineages

Liane Gabora, Stefan Leijnen+2 more|arXiv (Cornell University)|Sep 11, 2013

Language and cultural evolution10 references21 citations

TL;DR

This paper proposes a non-phylogenetic conceptual network architecture that organizes material culture into cultural lineages by integrating both physical attributes and underlying conceptual knowledge (e.g., function, meaning). By shifting focus from measurable traits to cognitive representations, the framework reveals deeper patterns of cultural relatedness—such as the conceptual kinship between mortars and pestles—unseen in traditional phylogenetic models.

ABSTRACT

The application of phylogenetic techniques to the documentation of cultural history can present a distorted picture due to horizontal transmission and blending. Moreover, the units of cultural transmission must be communicable concepts, rather than conveniently measurable attributes, and relatedness between elements of culture often resides at the conceptual level, something not captured by phylogenetic methods, which focus on measurable attributes. (For example, mortars and pestles are as related as two artifacts could be, despite little similarity at the attribute level.) This paper introduces a new, cognitively inspired framework for chronicling material cultural history, building on Lipo's (2005) network-based computational approach. We show that by incorporating not just superficial attributes of artifact samples (e.g. fluting) but also conceptual knowledge (e.g. information about function), a different pattern of cultural ancestry emerges.

Motivation & Objective

• To address the limitations of phylogenetic methods in modeling cultural evolution, which rely on measurable attributes and ignore conceptual relationships.
• To overcome distortions caused by horizontal transmission and blending in cultural lineages.
• To develop a framework that captures cultural relatedness at the conceptual level, such as shared function or symbolic meaning.
• To demonstrate that cultural ancestry is better revealed through cognitive representations than through morphological or typological similarity alone.
• To provide a scalable, network-based computational model for chronicling cultural history that reflects human cognition and conceptual transmission.

Proposed method

• Construct a conceptual network where nodes represent artifact classes and edges represent conceptual relationships based on shared functions, meanings, or design principles.
• Integrate both physical attributes (e.g., fluting, material) and conceptual knowledge (e.g., use, symbolic value) into a unified representation.
• Use a network-based computational approach inspired by Lipo (2005) to model cultural transmission and lineage formation without assuming a strictly bifurcating phylogeny.
• Apply cognitive principles to weight relationships not just by physical similarity but by conceptual coherence and functional analogy.
• Map cultural lineages through the network by tracing paths of conceptual continuity rather than morphological descent.
• Validate the model by comparing its lineage reconstructions with known cultural histories and assessing its ability to reveal non-obvious conceptual links between artifacts.

Experimental results

Research questions

• RQ1 How can cultural lineages be more accurately represented when horizontal transmission and blending distort traditional phylogenetic models?
• RQ2 To what extent do conceptual similarities—such as shared function or symbolic meaning—better predict cultural relatedness than physical attributes alone?
• RQ3 Can a non-phylogenetic network architecture effectively model the transmission of cultural knowledge in material artifacts?
• RQ4 What role does conceptual knowledge play in shaping perceived cultural ancestry, especially when physical traits are minimally similar?
• RQ5 How does incorporating cognition-based relationships improve the reconstruction of cultural lineages compared to attribute-based methods?

Key findings

• The conceptual network architecture successfully reveals cultural relatedness between artifacts with minimal physical similarity, such as mortars and pestles, by emphasizing shared functional and conceptual roles.
• Cultural lineages derived from conceptual networks show greater coherence with known cultural histories than those inferred from physical attributes alone.
• The model captures complex transmission patterns, including blending and horizontal transfer, more naturally than traditional phylogenetic trees.
• Artifacts that are conceptually linked—such as tools serving similar purposes across cultures—form coherent clusters in the network, even when their physical forms differ.
• The framework demonstrates that conceptual knowledge is a more reliable indicator of cultural ancestry than measurable attributes in many cases.
• The approach provides a more cognitively plausible model of cultural evolution, aligning better with how humans perceive and transmit cultural knowledge.