**Abstract:**

Research in cultural evolution demonstrates that humans have biases to attend to, remember, and transmit some types of content over others. These biases influence the evolution of diverse cultural phenomena across cultures, including folktales, religious myths, literature, film, and online misinformation. Given the widespread influence of these biases we can anticipate that they also influenced the training data of Large Language Models (LLMs). As the use of LLMs grows, it is important to examine whether they exhibit biases in their output. In five preregistered experiments using material from previous studies with human participants, we use a transmission chain-like methodology to test for content biases in the output of the LLM ChatGPT-3. We find that ChatGPT-3 shows biases analogous to humans for content that is gender-stereotype-consistent, social, negative, threat-related, and biologically counterintuitive, over other content. The presence of these biases in LLM output suggests that such content is widespread in its training data and could have consequential downstream effects, by magnifying pre-existing human tendencies for cognitively appealing and not necessarily informative, or valuable, content.

**Speaker's social media:**

Twitter: @JStubbersfield; Bluesky: @jstubbersfield.bsky.social

RECORDING - Large language models show human-like content biases in transmission chain experiments

**Abstract:**

In this submission, we explore the concept of "machine culture", describing culture that is mediated or generated by machines. In particular, we review existing and foreseeable instances where intelligent machines are transforming evolutionary processes of culture at the levels of variation, transmission, and selection. Empirical evidence highlights that machine culture is already occurring: models using reinforcement learning are enhancing variation when they inspire humans to develop novel behaviors, as has been demonstrated for the ancient game of Go (Shin et al. 2023); Large Language Models are facilitating the transmission of cultural knowledge and redefining the value of human creative skills (Epstein et al. 2023); and recommender systems are rewiring pathways of cultural transmission by selecting what and from whom humans learn. Compared to humans, machine intelligence is structurally different, trained differently, and subject to different computational constraints (Gershman et al. 2015, Griffiths 2020). We investigate how these idiosyncrasies of machines might reshape cultural variation, transmission, and selection. Ultimately, our contribution presents a research agenda and a conceptual framework that can be leveraged for deeper investigations of machine culture.

**Speaker's social media:**

@LevinBrinkmann

LIVE - Machine culture

**Abstract:**

In many real-world contexts, individuals face collective action problems at multiple scales, effectively choosing between local and global cooperation. Unlike isolated public goods scenarios, which can be solved if enough players are willing to cooperate, multilevel collective action problems present an additional difficulty of coordination in working towards the same goals. Through a behavioural experiment, we explore how people navigate these multiscale cooperation dilemmas and investigate the interaction between cooperative behaviour and perceptions of social identity. We find that introducing a local cooperation option undermines global cooperation efforts. Moreover, the presence of a local option exacerbates parochial biases, partly connected to a performance-cohesion effect, whereby successful cooperation within groups is associated with social identification with that group. While increasing the relative saliency and rewards of global cooperation can recover global cooperation levels, parochial biases persist. Our results highlight how complex group structures can influence behaviour and beliefs, potentially hindering global cooperation efforts and introducing complicated social cohesion dynamics.

**Speaker's social media:**

@charliepilgrim

LIVE - Competing scales of cooperation

**Abstract:**

Archeological findings indicate that around 400KYA-200KYA in the eastern levant, ancient humans followed a new type of habitation style, basecamp cave dwelling. This entailed multiple small groups living in the same environment for long stretches of time, foraging for resources in different parts of the environment during the day and meeting back at the cave for shelter. We suggest that this habitation pattern could support the gradual development of complex social learning skills. Initially, basecamp habitation allowed tracking different group members’ expertise in resource attainment, and the development of prestige bias. In addition, based on their knowledge of the common environment, individuals could infer where other individuals foraged that day, and gain counterfactual information about the richness of the environment. Intentional signaling using distinct local markers could increase information transmission fidelity. Using simulations of computational models, we show that material benefits gradually increase with the level of sophistication of social learning strategies. This increase was more distinct in larger environments with multiple foraging opportunities. Our work highlights the way habitation patterns and physical settings which humans occupied could support the development of complex social learning skills, and the way environmental characteristics could serve as a common framework for communication.

**Speaker's social media:**

@uhertz

RECORDING - Ancient humans’ basecamp habitation pattern and the transition from experience-based to information-based social learning

**Abstract:**

Mathematical and computational approaches have made important contributions to cultural evolutionary theory by highlighting the effect of factors such as population size, social network structure, and transmission fidelity on cultural evolution. However, some cultural phenomena have so far been more difficult to capture using agent-based and formal models. In particular, determining how personality, background, and cultural context shape how individuals select, transform and transmit social information has remained out of the reach of traditional modelling approaches, and has thus mainly been studied empirically. We here propose that leveraging the capacity of Large Language Models (LLMs) to “role-play” and impersonate specific profiles may be fruitful to address this gap. Indeed, these generative models have been found to exhibit quite satisfying accuracy and relatively good consistency when used to simulate different characters. We present a framework for simulating and analyzing cultural evolution in populations of role-playing LLMs, along with results from use-cases that illustrate the kind of hypotheses that can be generated using this approach. The software we developed for conducting these simulations is open-source and features an intuitive, code-free user-interface, which we hope will help to build bridges between the fields of cultural evolution and generative artificial intelligence.

**Speaker's social media:**

@Jeremy__Perez

LIVE - Extending cultural evolutionary theory to machines

**Abstract:**

Context
The cultural evolution of quantification systems is a recent phenomenon traced back to the Neolithic. Yet, research shows that basic mathematical ideas are part of human core knowledge and likely developed early in evolution. How such core knowledge gave way to complex, culturally diverse mathematical ideas is not fully understood. 
Purpose
This paper focuses on the potential mathematics implicit in material culture, including finished artefacts, the production of tools, and processing of materials, using fibre technologies as a prime example of how material culture preserves, transmits, and transforms mathematical knowledge. 
Methods
A comparative approach linking the ethnomathematics of fibre technologies to cognitive research and relevant archaeological evidence of string, basketry, and weaving, provides a way of testing hypotheses on the origins and development of mathematics in the remote past.
Key results
Results support the premise that weaving practices contain and afford operations such as geometry, measurement, ordering, grouping, and counting. Fibre technologies likely provided a context for the development of mathematical ideas over the past 100,000 years, intensifying during the Upper Palaeolithic. 
Implications
Archaeological material culture is an underused resource that can inform about the development of mathematical knowledge prior to the emergence of formal systems of quantification. 

**Speaker's social media:**



LIVE - The cultural evolution of mathematics through fibre technologies

**Abstract:**

Social learning is a key driver of the success of the human species (Henrich, 2015), enabling the preservation of technologies that individuals could not possibly have acquired on their own (Tennie, Call, & Tomasello, 2009). Presently, we are witnessing the rise of intelligent machines – “AI systems” – that are capable of making discoveries beyond the human horizon (Wang et al., 2023). This raises the question of whether social learning enables humans to also adopt and preserve inconceivable discoveries made by intelligent machines. In our behavioral experiment, participants’ success in a strategic task hinged on adopting a solution that humans struggle to discover due to a characteristic bias to avoid large losses. We contrast populations of humans transmitting solutions over multiple generations on their own with populations that had the opportunity to learn from AI agents in their first generation. Our findings show that an algorithmic model succeeds in discovering the adaptive strategy where humans fail on their own. Most importantly, humans are able to adopt this machine-introduced strategy and preserve it over multiple generations. Taken together, our results showcase the potential of intelligent machines to transform human cultural evolution, and an opportunity for designing them towards benefitting human progress.

**Speaker's social media:**

Twitter: @ThomasF_Mueller
Bluesky: @thomasfmueller.bsky.social

RECORDING - Humans can learn and culturally preserve inconceivable strategies discovered by intelligent machines

**Abstract:**

Humans adapt, not just through organic evolutionary processes but also through innovation, technology, communication, social organization, environmental exploitation and alteration, developing complex skills passed on within and between social groups. Evolutionary theorists recognize our species has a distinctive capacity to learn across generations and share that knowledge in dynamic and cumulative ways. However, much of the theorizing in evolutionary approaches are too reductive for most socio-cultural anthropologists. We argue that approaches to culture in cultural evolutionary theory can be augmented by envisioning culture as a dynamic assemblage of entangled processes at multiple levels, not necessarily reducible to constituent parts. Using such a framework, more anthropologists would be able to engage constructively with cultural evolutionary theory, and evolutionary theorists would have access to a wider range of anthropological resources. We offer Concepts in dynamic assemblages as an approach. We argue that focussing on Concepts and the dynamic assemblages within which they interface can augment exchange between cultural evolutionary theory and anthropology. Our goal is to foster dialogue and center the nuance that anthropologists require in evolutionary approaches and encourage recognition by evolutionary scholars that the cultural resources they are modelling can be more complex and yet tractable in an anthropological sense. 

**Speaker's social media:**

Twitter: @Anthrofuentes

LIVE - More than “Information in the head”: Culture, cultural evolution, and concepts in dynamic assemblages

**Abstract:**

Few cultural practices beyond language are as widespread as string figure games. Their global
distribution and potential to yield insights into cultural transmission and cognition have long been
noted. Yet, it remains unknown how or when this behavior originated and to what extent shared
motifs are signals of repeated innovations or deep cultural transmission. Here, we combined a
global cross-cultural inventory of string figures with a novel methodology based on knot theory,
which enables the unequivocal numerical coding of string figures. We performed a computational
analysis of a sample of 826 figures from 92 societies around the world. Across these societies, we
found 83 recurring string figure designs, some of which are regionally restricted while others
display a global distribution. The cognitively opaque nature of string figure designs and their clear
geographic distribution reveals processes of cultural transmission, innovation, and convergent
evolution. Most strikingly, the global distribution of some figures raises the possibility of shared
ancient origins.

**Speaker's social media:**

Twitter: @roopekaaronen, Bluesky: @roope.bsky.social

RECORDING - A global cross-cultural analysis of string figures reveals evidence of deep transmission and innovation

**Abstract:**

Previous cross-cultural research suggests that climate hazards may have profound effects on a society’s culture.  For example, in non-state societies serious food-destroying hazards (typically droughts, some floods, and pest infestations) strongly predict higher warfare frequencies and more customary beyond-household sharing.  The research question addressed here is whether non-food-destroying hazards (such as storms and seismic activity) have similar or different effects on a range of cultural traits hypothesized to be adaptive responses in a worldwide sample of nonindustrial societies. For the past two years our research team has coded the type and frequency of hazards for 132 societies in the ethnographic record using eHRAF World Cultures. Preliminary results suggest that the contrast between frequent food-destroying and frequent non-food-destroying hazard profiles generally make very different predictions. For example, societies with more non-food-destroying hazards tend to have tighter cultures (stronger norms and severe punishment for norm violations), but cultural tightness is not predicted by food-destroying hazards. Food-destroying hazards in non-state societies predict more warfare, but non-food-destroying hazards predict less warfare. We present theory and causal models suggesting explanations for these divergent findings. Understanding how humans have adapted to hazards in the recent past may help us better understand the hazards faced today. 

**Speaker's social media:**



LIVE - Food-destroying versus Non-food-destroying hazards: A worldwide comparison of differential effects on culture

**Abstract:**

The study of gene-culture coevolution, reconciling both nominally “social” and “natural” sciences into one comprehensive framework, has been broadly applied to understand the evolution of a wide range of “species-defining” human phenotypes; from language to large-scale cooperation. Despite this appeal, the empirical evidence for gene-culture coevolution is routinely perceived to be limited to a handful of “classic” examples, most notably evolution of lactase persistence. How could such a gap between conceptual weight and empirical evidence emerge? We propose that the issue here might be two-fold – there are both conceptual ambiguities in the idea of gene-culture coevolution itself, and formidable practical challenges to discerning its signal in natural data. 

The present talk will attempt to cut through some of these difficulties, re-defining gene-culture coevolution into a broader framework, and giving insight into an ongoing project focused on re-conceptualising an empirical science of gene-culture coevolution. Using methods from landscape genomics to quantify adaptive genetic responses to variation in subsistence across the Pacific islands, we present evidence for adaptation in genes related to energy metabolism and metabolic disorder. We conclude that closer cross-disciplinary collaboration may yield a stronger empirical science of gene-culture coevolution which promises to be of both evolutionary and medical significance.



**Speaker's social media:**

Twitter: @svenkasser

LIVE - How to stop beating a dead cow: Gene-culture coevolution revisited

**Abstract:**

Enquist et al. (2023) suggested that human cumulative cultural evolution was enabled by enhanced abilities to process and remember sequential information, leading to cultural and cognitive achievements unprecedented in other animals. This hypothesis mainly rests on results of sequence discrimination learning tasks, showing that unlike for humans, learning to discriminate between sequences is difficult for non-human animals, requiring them hundreds, often thousands of trials. It is unclear whether animals’ difficulty in these tasks reflect difficulties with remembering sequences per se or is mainly due to the arbitrary ordering of the sequence items, thus underestimating their sequencing abilities. To disentangle these effects, we investigate sequence discrimination learning with a physical task in which we contrast an “arbitrary” condition with one in which we use causal relations between stimuli to indicate which sequence yields a reward, aiming to facilitate encoding of sequential information. We present results from zoo-housed primates (23 squirrel and 24 capuchin monkeys, 13 chimpanzees), 77 human adults and 239 children (3-11 years) in an “AB vs BA, BB, AA” task, providing the first sequence discrimination study with chimpanzees and children, demonstrating that a lack of meaningful relations between sequence stimuli impedes learning in humans and potentially also great apes.

**Speaker's social media:**



Underline digital video library

RECORDING - Large language models show human-like content biases in transmission chain experiments

LIVE - Machine culture

LIVE - Competing scales of cooperation

RECORDING - Ancient humans’ basecamp habitation pattern and the transition from experience-based to information-based social learning

LIVE - Extending cultural evolutionary theory to machines

LIVE - The cultural evolution of mathematics through fibre technologies

RECORDING - Humans can learn and culturally preserve inconceivable strategies discovered by intelligent machines

LIVE - More than “Information in the head”: Culture, cultural evolution, and concepts in dynamic assemblages

RECORDING - A global cross-cultural analysis of string figures reveals evidence of deep transmission and innovation

LIVE - Food-destroying versus Non-food-destroying hazards: A worldwide comparison of differential effects on culture

LIVE - How to stop beating a dead cow: Gene-culture coevolution revisited

RECORDING - Investigating the evolutionary and developmental origins of sequence learning in primates

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES