Denmark

Many artificial intelligence systems are developed which interact with human beings and other such systems, sometimes embodied as robots. Their behaviour is driven by the knowledge they have on their environment. This knowledge may be learnt from experience. This is actually what advertised machine learning systems do. However, such systems will only behave properly if this knowledge, or culture, is shared with others. This raises the question: can they socially evolve their knowledge as assumed in cultural evolution?
Answering this question calls for developments in computational cultural (knowledge) evolution. It is at a crossroad between two threads:
- Cultural knowledge evolution in which cultural evolution is applied to the knowledge and beliefs of organisms.
- Computational cultural evolution in which computer capabilities are used to synthetise and analyse cultural evolution. It may be both experimental in which short-term small experiments are performed, or observational, in which observations can be made in the long run on large populations.

Computational cultural knowledge evolution may actually cover two types of activities: Agent-based simulation of cultural evolution in which agents are a model that can predict how a modelled system works or Artificial cultural evolution in which programs or agents are new `organisms&#39; exhibiting cultural evolution.
Both activities are worth pursuing to study the principles of cultural (knowledge) evolution. For instance, the discussion about other species failing to develop cumulative culture supports the conclusion that the human species has a unique feature: cumulativity. However, it may be more simply related to the capability to deal with complex cultural items which support cumulativity better or to rely on more elaborate adaptation mechanisms that provide this cumulativity. So far, it is not possible to decide. Similarly, it is difficult to provide direct evidence of the absence of replicator. Developing agents able to evolve cumulatively their culture could bring some insight in these debates.
We are experimenting with agents learning their knowledge from their environment and from other agents. They behave according to this knowledge and are able to adapt their knowledge in order to succeed in social games exploiting it. The advantage of using agents bearing an explicit representation of knowledge is that it is possible to assess the evolution of the properties of agents&#39; knowledge. In particular, the accuracy, the consistency, or the diversity of the knowledge can be directly measured.

Cultural Evolution Society 2022

Can AI systems culturally evolve their knowledge?

agent-based models

computational cultural knowledge evolution

artificial intelligence

The AGI conference series is the only major conference series devoted wholly and specifically to the creation of AI systems possessing general intelligence at the human level and ultimately beyond. By gathering together active researchers in the field, for presentation of results and discussion of ideas, we accelerate our progress toward our common goal.

Underline hosted the two main days of AGI 2020 on Wednesday June 24th and Thursday June 25th. Both days were **live-streamed** and you can read the full program bellow. All presentations are available for a **asynchronous viewing**. All presenters are grouped into panels. You can also hear guest talks by **Ben Goertzel** (SingularityNET), **Valery Tarasov** (Semenov Institute of Chemical Physics), **Tatiana Shavrina** (Sberbank) and **Joscha Bach** (MIT Media Lab). 

AGI-20

The AGI conference series is the only major conference series devoted wholly and specifically to the creation of AI systems possessing general intelligence at the human level and ultimately beyond. By gathering together active researchers in the field, for presentation of results and discussion of ideas, we accelerate our progress toward our common goal.

Many artificial intelligence systems are developed which interact with human beings and other such systems, sometimes embodied as robots. Their behaviour is driven by the knowledge they have on their environment. This knowledge may be learnt from experience. This is actually what advertised machine learning systems do. However, such systems will only behave properly if this knowledge, or culture, is shared with others. This raises the question: can they socially evolve their knowledge as assumed in cultural evolution?
Answering this question calls for developments in computational cultural (knowledge) evolution. It is at a crossroad between two threads:
- Cultural knowledge evolution in which cultural evolution is applied to the knowledge and beliefs of organisms.
- Computational cultural evolution in which computer capabilities are used to synthetise and analyse cultural evolution. It may be both experimental in which short-term small experiments are performed, or observational, in which observations can be made in the long run on large populations.

Computational cultural knowledge evolution may actually cover two types of activities: Agent-based simulation of cultural evolution in which agents are a model that can predict how a modelled system works or Artificial cultural evolution in which programs or agents are new `organisms' exhibiting cultural evolution.
Both activities are worth pursuing to study the principles of cultural (knowledge) evolution. For instance, the discussion about other species failing to develop cumulative culture supports the conclusion that the human species has a unique feature: cumulativity. However, it may be more simply related to the capability to deal with complex cultural items which support cumulativity better or to rely on more elaborate adaptation mechanisms that provide this cumulativity. So far, it is not possible to decide. Similarly, it is difficult to provide direct evidence of the absence of replicator. Developing agents able to evolve cumulatively their culture could bring some insight in these debates.
We are experimenting with agents learning their knowledge from their environment and from other agents. They behave according to this knowledge and are able to adapt their knowledge in order to succeed in social games exploiting it. The advantage of using agents bearing an explicit representation of knowledge is that it is possible to assess the evolution of the properties of agents' knowledge. In particular, the accuracy, the consistency, or the diversity of the knowledge can be directly measured.

technical paper

### Welcome to the fourth Cultural Evolution Society Conference, CES2022! 

As last year, Underline will support the online component of the conference. The ongoing COVID-19 pandemic is causing a great deal of uncertainty for all of us. Therefore, we seek to offer a hybrid format that will hopefully allow some folks to travel to Aarhus, others to meet up locally or regionally in CES hub-lets, and other yet to participate online. The field of cultural evolution is growing in size and scope; its theories, methods and applications - both academic and applied - are getting ever more robust. Do join us in what promises to be an exciting exchange of ideas and a celebration of our field.

##### ***For SoMe interaction and live tweeting please tag @CulturalEvolSoc and use #CESconf2022***


#### The CES 2022 local organising committee
* Luseadra Joy McKerracher
* Jesper Sørensen 
* Niels Nørkjær Johannsen 
* Benjamin Grant Purzycki 
* Mette Løvschal 
* Felix Riede
* Iza Romanowska 
* Karsten Olsen

Please click on the button bellow if you want to watch the videos

The field of cultural evolution is growing in size and scope; its theories, methods and applications - both academic and applied - are getting ever more robust. Do join us in what promises to be an exciting exchange of ideas and a celebration of our field.

**Author(s):** Pérez Velilla, Alejandro; Smaldino, Paul; Fedorova, Natalia Introduction: Risk-mitigation can be an important driver for the development of cultural practices and institutions. We concentrate on modeling the cultural evolutionary dynamics of sharing norms within clusters of individuals engaging in productive activities with associated risks, while sharing with each other at a cost. Through a process of growth and fission, sharing norms and sharer clusters evolve, leading to several possible configurations. Methods: Evolutionary game theory and insights from ergodicity economics are used to obtain results in a mathematical framework. Agent-based simulations are employed to generate simulated datasets with the future aim of interfacing with empirical data using Generative Inference. Results: Productivity increases the network carrying capacity of clusters as well as the optimal sharing norm, leading to more sharing in larger-sized clusters. Clusters can be at their densest in the regions of high variability, but sharing breaks down in both high-risk and high-security environments.

The evolution of sharing norms as cultural strategies for risk management

**Author(s):** Leon, X; Creanza, N Introduction 
Since songbirds use their songs for species recognition, we investigated whether two newly formed species could be distinguished by their songs. Differences in the songs of the Spotted Towhee (Pipilo maculatus) and Eastern Towhee (Pipilo erythrophthalmus) are evident; however, intermediate songs have been reported, particularly in the Great Plains where their breeding ranges overlap. 
Approach
We obtained recordings from Xeno-canto and performed syllable segmentation and song analysis using our recently published software, Chipper. Using log-transformed syllable features, we analyzed song variation across the geographic range of the two species and used a Random Forest Model to predict species classification. 
Results 
Our data showed several patterns, including a positive correlation between the longitude coordinates and the bout duration (rho = 0.612, p < 2.2e-16). Additionally, our model had an accuracy of 89.4% using the log-transformed syllable features. 
Conclusions
Our results indicate that the songs of the Spotted and Eastern Towhees exhibit geographic patterns, and these closely related species can be distinguished by their songs with high accuracy using our model.

Role of learned song in the evolution and speciation of Eastern and Spotted Towhees

**Author(s):** Snyder, K; Sellers, M; Creanza, N Introduction 
Environmental stress, especially during development, can cause impaired learning in individuals. Thus, learned behaviors, such as birdsong, may exhibit unique evolutionary dynamics after severe weather events relative to other traits. We predict that ecosystem-level stressors may cause population-level changes to birdsong as cohorts experience early-life stress, learn songs imperfectly, and tutor future generations. 
Approach
In 2016, an unprecedented drought affected western New York. We analyzed publicly available and our own birdsong recordings from 2010-2020, comparing two species in the drought-affected area and a nearby unaffected area. We also developed an agent-based model of song evolution after environmental stress. 
Results and Conclusions
We found drought-associated song changes in the species with more complex songs but not in the species with a simple song, implying that stress-induced deficits may disproportionately affect traits that are more difficult to learn. With our model, we found important factors for enduring cultural evolution after stress: population isolation and stressor duration. Together, these results demonstrate that climatic events can induce population-level trait evolution due to disruption in learning.

Cultural evolution after environmental stress: a case study of songbird populations affected by the 2016 New York drought

Introduction: Childhood is a period of life unique to humans. Childhood may have evolved through the need to acquire knowledge and subsistence skills. In an attempt to evaluate the importance of learning for the evolution of childhood, previous research examined the increase with age of returns to foraging. Any increase could be due to increases in knowledge or other factors such as body size and strength. Materials and methods: In a population of part time children foragers on the island of Pemba, Tanzania, we collected anthropometric measures such as size and strength (N: 202), estimated knowledge (N: 93) and observed a variety of foraging trips (N trips: 310, N individuals: 116). We estimate the contribution of the different elements to foraging success. Preliminary results: We find that knowledge is important for foraging success 
Clearly defining the relative importance of somatic versus cognitive traits of forager children provides clues about the selective pressures leading to the evolution of early life history traits in our species.

Age specific foraging success and importance of ecological knowledge in Pemba, Tanzania

 Introduction: When the children of one ethnic group are educated by adults of a different ethnic group, cultural dynamics in one or both groups may be affected in ways that are poorly understood. Such a situation characterizes Western-style education received by, or imposed upon, many indigenous peoples. Here I examine changes in distributions of distinctive cultural norms across a range of domains among indigenous Matsigenka students and their Mestizo teachers in three boarding secondary schools in Amazonian Peru.
Methods: I use Bayesian item-response theory models to characterize the distributions of norms among students and teachers interviewed in 2017 and 2019, and compare them to such distributions among non-academic Matsigenka and Mestizos.
Results: The analyses suggest that, in some boarding schools, Matsigenka students may be adopting some of the distinctive norms of their Mestizo teachers. The extent of this pattern appears to be mediated by, among other things, the particular boarding school environment.
Discussion: I discuss these results in light of new theory predicting when and why members of a disempowered minority group will adopt norms of a powerful majority.

A longitudinal study of cultural dynamics in Amazonian boarding schools

**Author(s):** Defelipe, R. P.; Mendes, D.; França, I. F.; & Resende, B. Our goal was to provide a how-to guide to move infant observational research from lab to home. We searched four databases (Scopus, PubMed, Google Scholar, Frontiers in Psychology Special Issue) from January to February 2022, using ‘infant/children’, ‘online/web-based/tele-observation’, ‘pandemic/Covid-19’ as descriptors. Eligible papers were in English or Portuguese, investigated children up to 13 years, used remote technology, and applied observation as methodology. Most of the 60 papers were from Developmental Psychology, from North America (e.g. USA), investigated Caucasian infants up to one year, used supervised online platforms (e.g. Zoom), used computers instead mobiles or tablets, and put the child facing the screen or in face-to-face interaction with the parents. The studies mostly concluded that remote infant observational research is feasible, especially, if you: observe infants up to one year old; design a short (e.g. 5 minute-session) and simple (e.g. free-play) experimental tasks; has already met your participants in person; give financial feedback (e.g. Amazon gifts); and are not dealing with low-income population with unsuitable internet and/or technological devices. The extent to which in-person research can be adapted for remote settings needs future investigation to promotes targeted investment in innovation, design, implementation, and effectiveness.

Moving infant observational research from lab to home: partial results from a systematic review

**Author(s):** Greenhill, Simon J. Forkel, Robert Introduction:
The last few decades have seen the publication of many language phylogenies. These phylogenies have proven to be incredibly useful for testing hypotheses about linguistic and cultural evolution. However, they can be hard to get hold of and require detailed phylogenetic knowledge to maximise their re-use. 
Materials and Methods:
Phlorest is a database of published language phylogenies that aims to standardise the outputs of these analyses to make them FAIR (Findable, Accessible, Interoperable, and Reusable). First, each analysis is preprocessed to a consistent format, providing a summary tree and a posterior tree sample, linked where possible to the raw data. Each taxon in the analysis is mapped to catalogues like Glottolog and D-PLACE. Currently, Phlorest contains 42 phylogenies, covering a total of 3470 taxa from 1974 languages. 
Results/Findings:
In this talk I will present Phlorest and discuss the benefits it provides. First, phlorest enables replicability and reuse of these trees. Second, having these phylogenies aligned in time and space enables us to compare patterns and processes across the globe. Third, phlorest allows us to scale up to bigger questions by combining trees into super trees. Finally, phlorest allows us to highlight interesting big picture findings from historical linguistics to the wider public, providing a highly visible resource that brings this research to a wider audience.

Phlorest: A database of consistent and reusable language phylogenies.

Introduction: Focusing on clothing and adornment (dress), this worldwide cross-cultural comparison asks why people in some societies appear to dress in uniform or standardized ways, whereas in other societies individuals display considerable variability in dress. Materials/Methods: Hypotheses are tested on 80 societies drawn from the Standard Cross-Cultural Sample (SCCS). The central hypotheses consider the impact of general societal tightness or looseness, degree of egalitarianism, societal complexity, and resource stress. Exploratory methods identify four latent constructs of dress from newly coded variables, one latent construct for tightness/looseness, and one latent construct for resource stress. Results: As expected, increased societal tightness was positively related to increased standardization and rules regarding dress and increased resource stress is generally related to more standardization of dress and rules regarding adornment. Societies mid-range in complexity have more standardization. Contrary to expectations, the predictors of tightness-looseness differ from the predictors of dress. Discussion/Conclusion: While some of the theorized relationships are supported, at the end of paper we discuss some puzzling findings, speculate about possible explanations, and suggest further lines of research.

Premium content

Can AI systems culturally evolve their knowledge?

Downloads

Next from Cultural Evolution Society 2022

The evolution of sharing norms as cultural strategies for risk management

Similar lecture

Post-Turing Methodology: Breaking the Wall on the Way to Artificial General Intelligence