The Early Christian Texts, Computational Methods and Cultural Evolution

Vojtěch Kaše (vojtech.kase@gmail.com; www.vojtechkase.cz)

August 12 2019, Warsaw

        Over the recent years, we witness a repeating call for consilience. Nowadays some scholars speak about its second wave. Consilience is understood as an attempt to build a unified science. While the first wave was primarily guided by natural scientists with bold claims concerning human sciences, in the case of the second wave we hear more a voice of social scientists and humanists dissatisfied by current developments in their disciplines. All proponents of consilience agree that it is deservable to form bridges between academic disciplines and to somehow move them closer to each other. What is sometimes mentioned as helping to build these bridges are the paradigms of evolution on the on hand and the adoption of quantitative and computational methods on the other. In this paper, I will try to demonstrate what this can mean in the context of biblical studies, namely in the case of the study of early Chriatianity.

The recent increasing interest in the buzz word “cultural evolution” might be viewed as one of the most promising pathways how to enhance evolutionary thinking in the human sciences and thus contributing to consilience. Broadly speaking, cultural evolution might be defined as a research of the long-term dynamics of human cultures informed by evolutionary theory. Recently, there has been established the Cultural Evolution Society, having its inaugural conference in 2017 in Jena. One of the first activities of the Society has been a survey among its founding members in an attempt to identify the major scientific questions and ‘grand challenges’ currently facing the study of cultural evolution. The results of this survey have been subsequently published in a journal article entitled “Grand Challenges for the Study of Cultural Evolution” and published in Nature Ecology & Evolution (Brewer et al., 2017). Instead of going further into describing what is and what is not “cultural evolution”, it might be useful to look at how is this buzz word employed in these challenges. The challanges are:

(1) Understanding the role of social adaptation in cultural evolution

(2) Understanding the role of cultural evolution in the context of organic evolution

(3) Modelling culture as a complex adaptive system

(4) Identifying processes of transmission and accumulation of cultural traits

(5) Integrating methods, data, and results across disciplines

(6) Creating new organizational and funding structures that support interdisciplinary research and teaching

(7) Identifying cultural evolutionary processes that address significant social, economic, and political problems

(8) Educating policymakers and the public about cultural evolution

        However, before delving deeper into these challenges as such, we have to ask why should we, attendants of a biblical studies conference, pay attention to these challenges at all? My answer is the following:

Most of us here in this room are some sort of experts on biblical literature in its historical environment. With that we can ask how can the historical data we work with help untangle any of the the above listed challenges? As someone working on topics concerned by the development of early Christianity in the Ancient Mediterranean at the Roman times in particular, I will narrow my question asking how can the study of early Christianity in its ancient Mediterranean environment contribute to address the grand challenges of cultural evolutionary research.

This question should not be confused with another question, which is, in my opinion, a little bit more intuitive to biblical scholars and historians in general: How can cultural evolution (or: cognitive science of religion or any naturalistic theorizing) contribute to solve the historical questions we have? I want to emphasize that I think that this question is complitely legitimate, but that my approach goes into a different direction and  put this question aside for a moment.

        From the point of view of cultural evolutionary scholarship, the history of early Christianity - and ancient Mediterranean in general - represents one particular historical environment among many others to be used to explore, elaborate, or test some hypotheses, theories or approaches. It is a field or laboratory. As such, it has its strengths[1] and weaknesses in respect to certain tasks associated with cultural evolutionary research[2], but will leave them aside for now.

Now we can ask what in particular does the early Christian research offer to address the grand challenges faced by the field of cultural evolution? Here I will focus on three of the challenges: (3) Modelling culture as a complex adaptive system; (4) Identifying processes of transmission and accumulation of cultural traits; (5) Integrating methods, data, and results across disciplines. Thus, my modified question now sounds how can the study of early Christianity contribute to solve challenges associated with modelling culture as a complex adaptive system, cultural transmission, and integrative methodology?

Challenge (3) is concerned by Modelling culture as a complex adaptive system.  In terms of complexity science, complex adaptive systems are characterized by four major features: (i) parallelism; (ii) conditional action, (iii) modularity; and (iv) adaptation and evolution) (Holland, 2006). Complex adaptive systems normally consist from a large number of similar elements (mutuallism).  The behavior of these elements can be interpreted in terms of receiving signals and responses to these signals (conditional action) and is commonly combined into larger units of subroutines (modularity). As complex systems in general, because of various feedback loops, this produces emergent properties of the system as whole. Finally, such a system is adaptive when its elements change over time.[3]

In  his recent book, Pascal Boyer spent a lot of ink warning against any essentialistic approach to culture. In my opinion, viewing culture or religion as a complex adaptive system  along the lines suggested, for instance by Richard Sosis, appears to be in a risk of such essentialism. According to Boyer and his colleagues from the Parisian school of cultural evolution (esp. Morin, 2015; Sperber, 1996; cf. Sterelny, 2017), strictly speaking, there is no such thing as culture, there are just many individual but interacting human beings with mind sometimes containing very similar mental representations, which might be, for analytical reasons, considered as identical (while they are not).  What does this imply for approaching culture or religion as a complex adaptive system? Let’s demonstrate it on the example of early Christinity.

What this does perspective imply for the study of early Christianity? From this perspective, we have to start with a population of inhabitants of the ancient Mediterranean in the first three centuries. At some point in time, in this population, there firstly begins to circulate a set of representations referring to Jesus of Nazareth, his life, teaching and later on also of the life of his disciples. These representations were transmitted through the population like any other cultural innovation. Over time, these representations started to substantially affect individual’s embedeness in their social networks. There started to emerge local clusters, representing what we are used to describe as communities. Among members of these communities, the representations referring to Jesus became also more and more coupled with their in-group-out-group intuitions and thus formed basis for their group identity.

I argue that this approach is empirically much more promising than its still very popular alternative, starting to approach religion on the group level. It is because it is much more sensitive to individual differences and boundary phenomena. For instance, it enables to capture people as somehow influenced by Jesus’ teaching but still participating in many other religious or magical activities, while only slowly modifying their social networks.

This approach requires complexity thinking suggested above. The components of the system are all members of the ancient Mediterranean population potentially entering into contact with mental representations referring to the Jesus of Nazareth et cetera. If we adopt this perspective, we should attempt to be heard in the cultural evolutionary debates about modeling culture as a complex adaptive system and a cultural evolution of religion and to emphasize our understanding of religion.

        Another grand challenge of crucial relevance to us is the one of (4) Identifying processes of transmission and accumulation of cultural traits.  I suggest that, in our case, this challenge can be addressed properly only when we accept the above sketched approach to religion. I have just sketched an approach accessing Christianity as a more or less open set of representations and behavioral innovations diffusing over a population, and not as a bigger and bigger group of people covered by system of belief and practices. The advantage of this approach is that makes as very sensitive to the internal dynamics of religion over time.

The early Christian material enables us to explore a number of cultural evolutionary hypotheses concerning cultural transmission. First of all, there is a debate concerning the role of content versus in context biases in transmission of cultural traits. In our data, we can trace cultural transmission of many belief- and ritual innovations over a very long time period and among a very large population of people (e.g. early Christian meals in the first three centuries were shaped by cultural transmission processes over more than 10,000 sundays). Thus, we can gain insights hardly accessible to any one working only with nowadays living participants in an psychological experiment or an anthropological field research.

As a third grand challenge to be more closely discussed here is the one of Integrating methods, data, and results across disciplines. Here I see that a lot of things can be done. The field of cultural evolution is driven by hypothesis-testing by means of quantitative methods. Therefore, I suspect, that the potential contribution of early Christian studies to the research on cultural evolution will be respected only when complemented by adoption of quantitative methods.  But what quantitative methods are available to historians working mainly with historical data?

Here I perceive the biggest potential in the methods of quantitative text analysis. By quantitative text analysis I mean methods, algorithms and tools developed primarily in the computer science field of natural language processing and information retrieval and computational linguistics. These methods are increasingly more and more common also in the humanities when associated with the buzz word “distant reading” introduced by Franco Morretti.

However, the application of QTA methods in the humanities contexts, especially in project under the umbrella term of digital humanities, often lacks the ideal of a hypothesis-driven research. Further, in these context, is it not common to associate these methods with a naturalistic framework referring either to the universals of human cognition or evolution. Thus, an adoption of QTA does not in itself imply a relevance for the consilience project attempting to bridge natural sciences, social sciences and humanities.

There is already a handful of studies adopting the QTA in the cultural evolution framework. For instance, Alberto Acerbi and Olivier Morin analyzed a large corpus of literary novels from 19th and 20th century to capture cultural evolution of emotions in the respected populations. Other scholars used very similar data to explore culture evolution of norms as reflected by language dynamics.

Quantitative Text Analysis and the Cultural Evolution of Moralistic Religions in the Ancient Mediterranean

In what follows, I attempt to demonstrate the relevance of these methods when dealing with some crucial hypotheses concerning religious cultural evolution and the challenges above.

One of the biggest empirical questions faced by cultural evolutionists is the problem of the phenomenon of ultrasociety: What were the main factors that enabled humans to live in large-scale societies which started to emerge over the last ten millenia or so? On the one hand, this is a question of technological development associated with the emergence of agriculture et cetera. But, on the other hand, it is also associated with  challenges on a psychological level: we seem to be mentally well equipped  to live in small-scale groups, but what about living in a city of hundred of thousands of inhabitens and to meet there on a daily basis with people we have never met before?

Some cultural evolutionists argue that to live in these large-scale societies requires adoption of new cultural norms supporting cooperation among strangers and that emergence and spread of new types of religions (putting higher emphasis upon morality) over the last couple of millenia might be associated with that. As proponents of the “Big Gods Hypothesis” claim, From cultural evolutionary perspective, as "communities increase in complexity and size, the gods’ powers and moral concern also become greater" and "by the time we get to state-level societies, Big Gods predominate and religion becomes intensely intertwined with public morality” (Norenzayan, 2016). Other scholars are sceptical about the role of religion in that process and instead suggest that the spread of new forms of religion was instead associated with economic devolopment. In these debates, Christianity is often used as a textbook example of the new form of religion.

For instance, Mullins and his colleagues from the Seshat project led by Peter Turchin suggests that early Christianity represented a movement which "introduce[d] a stronger moralizing component than previous local religions, as well as the adoption of supernatural beings overtly concerned with morality, which were largely absent in earlier ideologies" (Mullins2018). These authors here emphasize a difference between Christianity and the ancient Greek religion. In their view, “Greek gods were certainly not omniscient and, crucially, did not much care about what Greek people did from a moral standpoint, provided only that they continued to participate in the proper rituals. Nor was there any widespread punishment for moral transgressions” (Mullins2018).

But is this really the case? Let’s use the quantitative text analysis methods introduced above to look at this in more detail. Nowadays, there are severely well known resources of digitalized ancient Greek texts, like Perseus Digital Library or Thesaurus Linguae Graecae. However, for the type of analysis introduced below, one work in progress is of special interest. It is the corpus of  Lemmatized ancient Greek Texts created and made publicly availible by the Open Greek and Latin Project in Leipzig. This corpus contains 25,522,507 of words from works from the most important Greek texts from the period period from 8th century BC to 4th century CE.  This corpus is especially useful because each Greek word in it is coupled with its lemma (i.e. its dictionary-like form) and morphological information (especially word type category). I was working with these texts whilre writing scripts in the programming language Python and now I will shortly introduce to you some of my analyses.

After some intial cleaning and preprocessing of the data, I ended up with a corpus of 79 authors or authorship communities, associated with a corpus of 512 documents containing 2,775570 words. These words are only substantives and adjectives in their dictionary-like format, since this the most useful format for the below introduced tasks. These texts were splited into four groups:

(1) archaic texts (8BC-6BC),

(2) classical texts (5BC-4BC)

(3) Roman texts (1CE-4CE)

(4) Christian texts (1CE-4CE of Christian provenience)

        In these texts, there has been identified in total 17,299 instances of the Greek term theos, 7,841 in the Christian texts and 9,116 in the “pagan” texts (in pagan texts, there has been further identified 5,051 instances of the term ζεύς (Zeus). These instances has been extracted from the text with their immediate textual neighborhood (what is known as a concnordance row or a keyword in context) consisting from 3 words (ie. substantives and adjectives) on the left side and 3 words on the right side.

I was interested what kind of words these neighboring terms are and how they change over time and in dependence on provenience of the source text. Therefore, I focused on terms  which could be interpreted as indicating an association between the term god and some other conceptual domain: namely morality, power and impulsivity or violence. Thus, I formed three lists of indicator terms. For time being, let’s focus on the domain of morality only.

  1. for morality, there were terms ἀγαθός (good), ἀλήθεια (truth), ἀληθής (unconcealed, true), ἀρετή (virtue, moral excellence, perfection), δίκαιος (just, righteous, impartial), δίκη (justice, judicial hearing, punishment)
  2. for power, there are δύναμις (might, power, marvelous works), δυνατός (powerful, able, possible), εἷς (one), μέγας (large, great), μόνος (only, solitary, desolate)
  3. for impulsivity, there are βία (force, violence), δεινός (fearful, terrible, dread, dire), θυμός (an outburst of passion, wrath), μάχη (strife, contention, quarrel), πόλεμος (a war, battle, strife), χαλεπός (hard, harsh, fierce)

With these in hand, I could then measure the number of co-occurrences of the key term and an indicator term in a text in proportion to the frequency of  the indicator term alone in the rest of the text. These values can then be statistically compared accross our four groups of texts. What are the results?

        First, what my analysis revealed, contrary to the findings of Mullins and his colleagues, that the understanding of god in the Greek period has been far from being stable. Between the archaic and the classical period, we witness a substantial increase in the strength of association between the term “theos” and various terms associated with “morality”, like agathos or dikaios. Further, when we take all Christian texts in our corpus and compare them with all non-Christian texts, the Christian god appears to be on avarage more associated with morality. However, this result does not persist when we compare the Christian god with the God of the classical period.

These findings are in agreement with the affluence hypothesis introduced by Nicolas Baumard, Pascal Boyer and their colleagues. According to them, the emergence of new forms of religions, emphasizing extended cooperation, was not so much associated with increasing social complexity, but was instead a by-product of economic development causing some changes in human behavioral preferences. As the Greek society in the classical period reached certain level of affluence (even higher than the Roman Empire few centuries later), a substantial number of individuals from the upper segment of the society also naturally slowed their life-history strategies and turned attention from immediate goals to long-term goals, widened their social networks et cetera. This could, in turn, also elevate into popularity new forms of religions, characterized also by morally-concerned deities.

Conclusion

I started this paper mentioning that we know face to a second wave of consilience in a sense of a unification of knowledge accross the disciplines. I mentioned evolutionary thinking and computational or digital methodologies as two important pillars of this project. I tried to demonstrate that the study of early Chriatianity and the study of ancient Mediterranean cultures in general, can substantially contribute to the field of cultural evolution, which can be viewed as on of the most promising school of evolutionary thinking concerning human behavior and culture. Further, I suggested that our contribution to cultural evolutionary research can be especially sound when combined with computational methodologies. Since the primary data of biblical studies research are texts, it is quite natural to start with them and to complement our close reading of the sources with their distant reading as well. My impression is that only this way can the biblical scholarship become an important voice in cultural evolutionary research, substantially helping in addressing its challenges and influencing future direction of this research. Otherwise, I worry, we will become only secondary consuments of theories produced by others, eternally dissatisfied by superficial interpretations of our data made by non-experts.

The Grand Challenges with Short Commentary

(1) Understanding the role of social adaptation in cultural evolution. This challange focuses on “possible adaptive roles of social structures and behaviours such as sharing, kinship and capital punishment” in the dynamics of cultures. (2) Understanding the role of cultural evolution in the context of organic evolution. For critiques, this challenge is commonly viewed as as the focal point of cultural evolutionary research: if the research does not contribute to solve puzzles at the level of organic evolution by natural selection, why to speak about evolution at all? However, this is misleading. as we should realize the complexity behind interactions of population of many biological species at the same time, each of them somehow modify the environment in which they live and therefore also change selective pressures. Last 10 thousand years since the advent of agriculture might be evolutionary too short time span for novel mental adaptations to develop, but we should not overlook the fact how we impacted other species, especially microorganisms and plants. Therefore the research of cultural dynamics and biological evolution is closely associated. (3) Modelling culture as a complex adaptive system. Across disciplines, we face an increasing popularity of complex systems thinking. We should be aware that “complex system” is a technical used mainly by physicists studying systems marked by complicated feedback loops, both positive and negative. Interacting parts of these systems produce emergent behavior, unpredictible by knowledge of the behavior of the parts alone. These systems are difficult to study, but their study is not impossible. Especially recently, the research in complex systems is marked by substantial progress, especially due to computational modeling. These methods have been increasingly popular in the study of cultural dynamics and are more and more popular in archaeology. (4) Identifying processes of transmission and accumulation of cultural traits.  Here we can list these questions: “What are the cognitive and behavioural processes underlying cultural transmission? How are innovations selectively transmitted over existing technologies or behaviours? How do differentiated social statuses, roles and educational systems impact cultural transmission? How can we most usefully conceptualize the units of cultural transmission? What does it mean when we say a culture evolves?”. In the context of the academic study of religion, these questions have been frequently discuss in the context of the cognitive science of religion, and later on I will return to them. (5) Integrating methods, data, and results across disciplines. It was a surprise for me during the inaugural conference in Jena, that most of the attendants were psychologists working mainly with experimental methods typical for their discipline. Psychological experiment is a very useful method. It is sometimes presented in social sciences as an ideal of science. However, considering the challenges above, it appears that psychological experiment is hardly suitable to solve many of them. Instead of that, the challenge number 3 adequately refers to modeling. We should realize that mathematical and computational models have at least the same importance in sciences as experiments.  The last three challenges can we only shortly list: (6) Creating new organizational and funding structures that support interdisciplinary research and teaching; (7) Identifying cultural evolutionary processes that address significant social, economic, and political problems; (8) Educating policymakers and the public about cultural evolution.


[1] The main strength of early Christian environment is that it covers an extensive time period of several centuries and concerns several generations of millions of people. This is much longer time span than which is typically available to an anthropologist doing fieldwork among an exotic tribe or to an experimentalist working with her or his participants for a half an hour. Our other advantages are that we are concerned by a period without the disturbing influence of print and mass-media and by a period marked by relative freedem in religious matters. We are dealing with both highly diverse but also somehow culturally unified environment - mainly the Hellenized Roman Empire at its peak. That means we are dealing with a high extent of variation in cultural behavior but also some sort of cultural koiné. Further, since, for the most part, the religious matters were only partially a subject of state authority, the cultural innovations in religious domain and its success does not depend so much upon individual decisions of an elite, but might be productively studied as a population process. In comparison to any modern environment, we are dealing with a period in which issues as cultural transmission operate without influence of the disturbing forces of print and digital media. Thus, there is a much bigger correspondence between real spread of a cultural innovation and the population level demand for it. Finally, this brings to the early Christians as very prolific producers of texts, a behavior which, with an exception of the Jews, strikingly contracts with behavior of devoters of other religious traditions in the given cultural contexts.  These five points might be viewed as features making the study of Early Christianity a promising for untangle some cultural evolutionary challenges.

[2] The most important disadvantage is the scarcity of available evidence. It is nice that we have thousands of pages of early Christian texts, and, of course, to gain a good knowledge of these texts represents a respectful achievement of any early Christian scholar. But we are well aware but these texts are a product of only a small literate fraction of the population with sufficient resources enabling to invest time into producing texts. Therefore, it is problematic to make any generalizations on the basis of these texts concerning the population as a whole. Further, coupling the literary evidence with the anecdotal archaeological evidence produces more problems than solves, since it appears that the core of early Christian iconography lies completely elsewhere than we perceive it be on the basis of our knowledge of early Christian texts. The limitation of early Christian texts might be well demonstrated when compared with textual data from nowadays modern social network. Look at Instagram, for instance, which has recently crossed the boundary of 1 billion active users. For a computationally competent scholar of religion, there is now almost 30 millions of posts containing the hashtag #jesus and about 5,700,000 posts. All these posts are downloadable and available for computational analysis. In other words, it seems that online social networks like Instagram are treasures of highly relevant data concerning human behavior relevant for cultural evolutionary research. Of course, these data also have many disadvantages, but indoubotely there at least much bigger and come from substantially bigger segment of the population than our ancient early Christian data.

[3] Richard Sosis and his colleagues developed an approach to religion according to which “religion may be best understood as an adaptive complex of traits incorporating cognitive, neurological, affective, behavioral, and developmental elements” (Sosis, 2019, p. 47). They argue that “The religious system… is a complex adaptation that serves to support human social life” (ibid., p. 48). As such, “[r]eligious systems typically maintain eight core elements: authority, meaning, moral obligation, myth, ritual, sacred, supernatural agents, and taboo” (ibid., p. 48). Sosis’s approach might look like as a perfect application of complexity science to the domain of religion. However, under a closer scrutiny, it becomes clear that the usage of the term complex adaptive system goes into a quite different direction than the one which common in the complexity science sketched above.# First of all, I would welcome if it would be specified what is the ontological status of such a system? Further, in Sosis’s scheme, we see a number of very different and specialized components, but not the many similar components revealing mutualism typical for other complex systems.