• Aucun résultat trouvé

Complex dynamical systems: society and individuals

Dans le document 1.2 Plan of the thesis (Page 24-27)

The idea that these patterns of collective behavior are shaped by vary-ing selection pressures at both the individual and the population level has been largely put forward by several theorists in the social, behavioral and cognitive sciences (Boyd & Richerson, 1985; Lass, 1997; Lewis, 2008;

Young, 1993). Social evolutionary theories maintain that cultural evolu-tion is constrained by individual traits and social structures (Labov, 2011;

R. C. Lewontin, 1983). However, although there is a widespread inter-est in approaching the study of the evolution of culture form an integrated perspective, the complexity of the social process has prevented the devel-opment of models of dissemination of culture that include both a compre-hensive set of cognitive traits and specific features of the social structures at the population level.

Here I ground my exploration in two main sources of variation: Society and individuals. Firstly, the transmission of cultural features in a popula-tion (e.g. language variants) depends on a number of characteristics of the society. One of these features is the arrangement of connections among in-dividuals in the social network (Centola, 2018). For example, who speaks to whom, how often people exchange information or how many neighbour-hoods there are in a population are all important characteristics of a social network. Social networks shape the spread of cultural information by con-necting or disconcon-necting people over time and in turn they can affect the dissemination of cultural variants and the emergence of cultural conven-tions. Previous models have successfully shown how the co-evolutionary dynamics of social influence and network structures can affect cultural change (Becker et al., 2017; Centola, 2010; Centola & Baronchelli, 2015;

Centola et al., 2007; Muthukrishna & Schaller, 2019). Inspired by those models and expanding on previous work in cultural evolution (Fay et al., 2010; Tamariz et al., 2014), here I focus on a specific manipulation of the social network: The order in which connections between individuals un-fold over time. I present a micro-scale model of dissemination of culture and I apply my simulations to a number of specific social learning scenar-ios.

Secondly, societies are formed by individuals. And individuals may be disproportionately in favour or against an idea, a norm or a word (Boyd &

Richerson, 1985). For example, individuals can be biased towards adopt-ing one particular word just because it is more iconic or easy to remember, or against one particular institutional norm just because it was promoted by an unpopular government. People also have personal beliefs or value systems, and they can be more prone or reluctant to use new informa-tion, such as scientific informainforma-tion, institutional information or religious information to update their prior beliefs or value systems. In this thesis, I am particularly interested in the construction of models that include a comprehensive set of cognitive biases that have been shown to be crucial mechanisms underlying the adoption of cultural variants at the individual level, and therefore critical for explaining cultural diversity at the

popu-lation level. In particular, throughout this thesis I explore the effects of a general content bias, or a preference for variants with high value (Boyd

& Richerson, 2008; J. Henrich & McElreath, 2007; Tamariz et al., 2014), an egocentric bias, or a preference for self-produced variants (Tamariz et al., 2014), an allocentric bias, or a preference for others-produced vari-ants (Garrod & Pickering, 2007; Pickering & Garrod, 2004; Tamariz &

Kirby, 2015), a compliance bias, or a tendency to conform to institutions (Cialdini & Goldstein, 2004), aconfirmation bias, or a tendency to process new information in a way that confirms one’s prior beliefs (Del Vicario, Scala, Caldarelli, Stanley, & Quattrociocchi, 2017; Nickerson, 1998; Quat-trociocchi, Scala, & Sunstein, 2016) and a memory-frequency bias, or a preference for variants that are more frequent in the agents’ history (Ferdi-nand et al., 2013; Hudson Kam & Chang, 2009; Hudson Kam & Newport, 2005; Tamariz & Kirby, 2015). My models, thus, require also the imple-mentation of memory, which is operationalised as the maximum amount of history that can affect agents’ variant choice at a given moment. All these cognitive traits are, therefore, important selection pressures that can affect the way in which cultural variants spread out in a population. Throughout my exploration, I examine the impact of these parameters using different versions of my agent-based models.

Human societies, however, are not just a number of individuals or groups learning and sharing information within a social network. Humans have an extraordinary capacity to build developmental environments and com-plex niches. Niche construction is particularly relevant to human evolution because by modifying the environment, humans create artifacts that act as additional sources of biological and cultural selection (Laland, 2017a;

Laland et al., 2000). Human institutions are a paradigmatic example of these human artifacts (Bowles, 2000). Institutions have been defined by re-searchers on institutionalism as ‘integrated systems of rules that structure social interactions’ (Hodgson, 2015, p. 501). Similarly, Boyd and Rich-erson (2008) define social institutions as norms and conventions that give durable structure to social interactions within a population. These institu-tions are now recognised to have played important roles in the co-evolution of individual cognition and human culture by creating norms, spreading be-liefs, establishing language rules or promoting ideas. In the later versions of the computational models I present in this thesis, I pay special attention to the emergence of representative institutions and how they interact with human cognition, value systems and the selection and production of cul-tural variants in a population. On a more conceptual level, the notion of

niche construction will be also used in the last part of this thesis to propose a conceptual model applied to the case of the emergence of language regu-larities. I gather evidence from iterated learning models (Chater, Reali, &

Christiansen, 2009; Kirby, 2017; Kirby, Dowman, & Griffiths, 2007; Smith et al., 2017; Zuidema, 2002) and ecological evolutionary developmental bi-ology (Gilbert & Epel, 2009, 2015; R. C. Lewontin, 1983; M¨uller, 2020;

Sultan, 2015, 2017) to show that these two frameworks are compatible on the basis of recent interdisciplinary studies that are stressing the necessity to construct more integrated and less simplistic models of cultural evo-lution (e.g. Balari & Lorenzo, 2013; Charbonneau, 2016; Deacon, 2010;

Fisher & Vernes, 2015; Laland, Odling-Smee, Hoppitt, & Uller, 2013; La-land, Odling-Smee, & Turner, 2014; Mesoudi, Whiten, & LaLa-land, 2006, among others).

The processes of cultural transmission are therefore shaped by social structures (e.g. network connectivity dynamics and emergent institutions) and by individual cognitive traits (e.g. cognitive biases, value systems and memory). All these factors take part in complex dynamic systems that require an integrated treatment of the different levels of the evolutionary process.

Dans le document 1.2 Plan of the thesis (Page 24-27)