The social dynamics of cultural behaviour: transmission biases and adaptive social learning strategies in wild great tits.

All animals need to make the most of new opportunities or deal with changing environmental conditions. These changes may be short-term such as seasonal change, or long-term shifts such as climate change, and often impact the availability of food resources and, potentially, survival. Broadly, two different strategies might be used to increase access to resources in a changing environment. Animals might develop new solutions to problems (innovation), and thus find new resources themselves, or they might observe others and copy successful solutions. The latter, called social learning, is expected to be much more frequent than innovation, allows new behaviours to spread rapidly between individuals and is thought to be fundamental in forming traditions. Social learning has long fascinated biologists and anthropologists; understanding how behaviours spread and traditions are maintained in animals can shed light on the factors promoting complex culture in humans.

An important determinant of social learning is the social organisation of the population in which learning occurs. It was long thought that only humans could exhibit highly developed cultural transmission due to their capacity for communication and learning that is facilitated by long-term social bonds. However recent research has found locally maintained cultural behaviour in a wide range of animals. Further, social network analysis in both human and other animal populations has allowed population structure to be accurately measured. Thus, using social networks to map the spread of new behaviours provides an exciting opportunity to understand this important learning process.

In this study, we will study the spread of novel information in wild populations of a common bird, the great tit. All individuals in our large study population are tagged with microchips allowing us to track them automatically; our pilot data shows that they will learn socially. We will develop devices where one of two simple solutions provides access to food, and train an individual to solve one solution on the task in captivity before releasing it back into the woodland where we will place a number of these devices. Using this approach, we will be able to track who has learnt, from whom they learnt, and which of the two solutions they learnt. Using the social network of this population, we will track the spread of the new behaviours, and determine what characteristics made some individuals more important in spreading them. By training different individuals on the two different solutions, we will also see how local traditions develop and are maintained.

Not all traditions or behaviours are advantageous. For example, in humans it has been shown that obesity can spread through friendship groups. In the second phase of this project we will alter the reward to different solutions of the task by replacing the popular solution with a low reward (peanut granules instead of a worm), maintaining the high reward on the less popular solution. This will test whether bad traditions are maintained through social reinforcement where individuals blindly copy the majority even when better solutions exist.

Finally, we will develop some new technology that will predict what solution new individuals should be learning based upon the behavior of the group they belong to. By changing the behaviour at the device in response, we will then test in detail what elements of the behaviour observed in others is used when social learning. This will be the first time that anyone has used an active device to directly manipulate the behaviour of wild animals in this way. This will itself advance scientists' abilities to understand what rules individuals follow when making decisions such as who to copy and when. Such knowledge will be widely applicable across disciplines, for example in providing new opportunities for active conservation of threatened species by introducing behaviours that improve survival.

Animal culture plays a potentially important role in shaping adaptive behavioural plasticity, and understanding the evolution of cultural behaviour in animals may shed led light on the development of cumulative culture in humans. However cultural transmission takes place within a specific social context: group dynamics and social networks will determine the formation and persistence of traditions. Further, population structure may interact with individual variation and social learning strategies, with consequences for whether such transmission biases result in group-level phenomena such as conformity.

The lack of large-scale experimental data from wild populations is major gap in the study of animal culture. Using recently developed electronic equipment enabling automated data collection, and controlled presentation of stimuli in the field, we propose to disentangle the factors promoting the spread and establishment of traditions in a large wild bird population. Using a two-action-and-control experimental design, we will train demonstrators to perform one of two actions to solve automated puzzle-boxes. We will then track the spread of this action through replicated PIT-tagged subpopulations, collecting social network data to test the relationship between social structure and diffusion dynamics. Behavioural assays will further allow us to assess the interaction between social learning and individual variation. We will build on this baseline data with a series of targeted manipulative experiments: (i) altering pay-offs in order to test whether adaptive learning strategies will prevent the emergence of maladaptive traditions; (ii) testing the impact of producer-scrounger polymorphisms on diffusion dynamics, and; (iii) manipulating quality and reliability of information to determine whether frequency-dependent learning biases resulting in conformity. In doing so, we will be able to study the social context of social learning at an unparalleled scale.

The primary justification for the work proposed here is to advance our fundamental understanding of social learning in animals. Social learning is thought to underpin a range of mechanisms that allow individuals to modify their behaviour and adapt to environmental or social changes. This work will test existing theory, and help establish the next generation of work in this area. This project will break new ground in terms of both the scale of the experiments but also through its scope via the incorporation of social networks and individual traits such as personality.

We anticipate the work having particular impact outside the relevant academic fields in the following four areas:

First, we will develop novel technology for integrating computational modelling and experimental biology in the wild. Using our pilot and first phase data, we will fit predictive Bayesian models from collective decision-making and run simulations to determine parameter boundaries where conformity is predicted to occur. This model will then be programmed into the experimental device in order to manipulate the reliability of the information available to targeted individuals. This approach, though ambitious, has the potential to make significant advances in both our knowledge of conformist transmission (the academic goal of this work), but also much more broadly to provide a test case where we directly manipulate behavioural mechanisms. We expect that this will be applicable in a wide range of fields. To this effect, we have budgeted to run a workshop in Oxford entitled "Advances in the use of technology for wild cognition research" where we will invite researchers in both cognition research and broader more applied fields.

Second, we will develop a detailed understanding of the social dynamics associated with the spread of information through natural populations. This knowledge will be influential in cognition and social evolution research, but can also be expected to be influential more widely across conservation and animal sciences. We will demonstrate how the technology we plan to develop can be used to disseminate novel behaviours in wild populations, which has potential applications for the conservation of threatened species, particularly those under threat from human induced rapid environmental change. These findings will be equally applicable for the management of livestock, where technology could be used to provide stimulating new behaviours for improving welfare.

Third, this research will contribute to the public understanding of science. Research into animal cognition and social behaviour, especially in familiar garden species, creates tremendous public interest. One famous case is milk-bottle opening in blue-tits, a classic example of social learning and diffusion of innovation in a related species. We expect that our research outputs will attract considerable media attention, and we will assist this in producing high quality HD video footage of birds 'in action'.

Finally, the proposed work will have a large impact on fostering the scientific careers of the two research co-investigators. This will be achieved through the production of leading research, both through the development of novel technology and in the scale of the research being conducted in the wild. The PDRAs will develop national and international collaborations both within the project team (with the co-investigator and project collaborator) and externally (with the laboratory of Dr. Gonzalo de Polavieja). Together, the components of proposed research cross disciplinary boundaries between cognition, evolutionary biology and computational biology, which will provide a depth of experience for the researchers involved.