The first chapter, titled “The Evolution of Productive Organizations”, attempts to break new ground in our explanations of the nature of guilds, partnerships and other pre-modern firm-like organizations. I use the theory of Cultural Evolution to develop a foundation for the evolutionary origins of firms. In extant theory, a historically rooted explanation for the evolution of firms is largely absent. I argue that Cultural Evolution theory can address this challenge, putting knowledge, learning, and cooperation at center-stage. This theory, developed in Evolutionary Anthropology, studies culture as an evolutionary system. It specifies micro-foundational mechanisms for inheritance, selection and variation. Culture is defined as information that is acquired from other individuals via social learning mechanisms, such as imitation and teaching. Information includes beliefs, norms, knowledge, skills and artifacts/technology. In this second section, we develop a cultural evolution model that illuminates the evolution of pre-modern productive organizations, such as, partnerships and guilds. Specifically, we introduce productive organizations in a workhorse cultural evolution model, widely used to explore the conditions that make social learning fitness enhancing. If organizations are exclusive and facilitate social learning, they stop the negative externality generated by the replication of social learners. The basic insight provided by the model is that productive organizations evolved because they favoured the conditions that sustain the process of cumulative culture. Productive organizations make social learning –and therefore culture– useful to society, playing an important role on the adaptive success of the human species. Our model has predictions regarding the benefits of organizations for society that are at odds with standard models of firms in economics and management based on transaction costs. For example, while in transaction costs theories the firm-like organizations is more valuable when uncertainty is high, in our model the firm-like organization is more valuable when it is low. These differences allow for empirical comparison of the theories. We test our theory using data from the Ethnographic Atlas and the Standard Cross Cultural Sample. We measure the presence of technologies in pre-modern societies (e.g., weaving, metalworking, pottery) and whether they were used throughout the society or mainly by a small group of people, that is, within a productive organization. Across several tests and robustness checks, we find consistent evidence for the propositions and comparative statics of our model. The third section, titled “The evolution and Impact of Cooperation in Large Groups: Evidence from Administrative Data and a Field Experiment”, I zoom in on the informal structure of firms (or “culture”) by studying the drivers of cooperation in large groups. As groups grow large, it is increasingly hard for workers to accept to pay a cost in order to provide a benefit to colleagues or the group at large. There is a vast theoretical literature in the fields of evolutionary biology and evolutionary anthropology regarding the conditions and mechanisms that favour the evolution of cooperation in large populations (i.e., increase in frequency). For cooperation to evolve, a mechanism is required that allows favouring cooperators over defectors. This mechanism is an interaction structure that specifies who interacts with whom in the population (i.e., random v/s structured v/s flexible) and how the agents interact in order to receive payoffs (e.g., what is known by whom, degree of repetition, order of play, details of payoff functions, enforcement technology). On the former element of ‘how’, the main mechanisms are spatial/network selection and group selection; on the latter, direct reciprocity and indirect reciprocity. Using this theory, we collaborate with three organizations to study a workplace safety practice that is based on voluntary cooperation by workers. In this practice, an initial core of cooperators strives to expand cooperation within the implemented site (e.g., plant or store). The methodology leverages cooperation: training and counselling is costly to observers while the benefits of improved safety flow mostly to the observed workers. Moreover, the strive for expansion provides a unique setting to study the evolution of cooperation (i.e., its increase in frequency). Using a detailed administrative dataset, we first show that the methodology reduces accidents and improves culture, documenting the power of cooperation. However, the dataset also demonstrates that, in line with theories of cooperation breakdown in large groups, this positive impact decreases very quickly as the number of observers expands. Then, we examine the idea of interaction structures, by analysing the impact of direct reciprocity in two ways. First, using the administrative dataset, we document traces of the positive impact of direct reciprocity in the adoption and impact of the practice. Second, we executed a field experiment in four sites where we intervened the established safety methodology with a baseline “direct reciprocity” treatment, plus two additional interaction treatments, aimed at solving the breakdown problem. We show that: i) the effort of the additional observers is restored when the expansion of observers is structured around small groups (1st treatment – “Direct Reciprocity”), ii) lifting the anonymity of the observed workers is detrimental to observers’ effort, eliminating the benefits of direct reciprocity (2nd treatment – “Identity”), and iii) public display of effort is mute (3rd treatment – “Indirect Reciprocity”), but interacts with the ‘private enforcement’ –measured with administrative data– in subtle ways. Further, we find that these treatment effects on effort translate into the speed of diffusion (i.e., the likelihood of becoming observer) and into safety outcomes (i.e., safe behaviour and accidents of the workforce): both increase with treatment 1 but decrease with treatment 2. Overall, the third section provides unique field evidence of cooperation breakdown when groups grow large, as well as of “structured growth” (sustained by direct reciprocity) as a crucial mechanism that allows for its recovery and evolution.