Björn Lindström

Human cultural progress is powered by collective innovation - the process of gradual exploration and refinement of innovations across individuals and generations. However, the exploration and social learning strategies fueling this collective innovation process remain poorly understood, hindering efforts to enhance its efficiency. In this project, our goal is twofold: (i) to understand the exploration and social learning strategies humans use for collective innovation, and (ii) to enhance collective innovation by partnering human participants with Artificial Intelligence (AI) algorithms. We will use computational modeling and interactive collective innovation experiments to address the first goal. The second goal will be met by harnessing AI algorithms' unique abilities to explore and learn from others in ways distinct from humans. We aim to design and test algorithms that complement human collective innovation, enhancing its efficiency while mitigating human biases. Our results will both provide new insights into the strategies humans use for innovation and social learning while uncovering how AI can facilitate these processes to improve collective innovation. In an era where AI is poised to shape the course of human existence, studying the interplay between human and AI capabilities is essential. Through this project, we aim to shed light on what makes us human while harnessing the collaborative potential of humans and AI to drive collective innovation forward.

Social learning is essential for human success, enabling individuals to acquire skills and knowledge - traits - from others. While theories of "cultural evolution" suggest that success-based imitation is crucial for the spread of adaptive traits, recent studies show that it is much rarer than expected. Our project addresses this mismatch by recognizing that traits are hierarchically structured: learning advanced skills requires mastering more basic ones, like learning guitar chords before solos. Through three studies using computational models and experiments, we will investigate how such structure influences social learning. Study 1 will use simulations to quantify the performance of social learning strategies in structured trait environments. Study 2 employs experiments to test these predictions, and explore how individuals adapt their strategies based on trait structure. Study 3 will use findings from Studies 1 & 2 to test predictions about how traits spread in groups with large-scale experiments.This research addresses key puzzles regarding the underuse of success-based social learning, variability in learning strategies, and the mechanisms allowing their adjustment with experience. By integrating formal models with empirical testing, we aim to understand how trait structure impacts social learning and its role in cultural evolution. This project will not only illuminate the nature of social learning but also inform interventions for efficient knowledge dissemination

Human cultural progress is powered by collective innovation - the process of gradual exploration and refinement of innovations across individuals and generations. However, the exploration and social learning strategies fueling this collective innovation process remain poorly understood, hindering efforts to enhance its efficiency. In this project, our goal is twofold: (i) to understand the exploration and social learning strategies humans use for collective innovation, and (ii) to enhance collective innovation by partnering human participants with Artificial Intelligence (AI) algorithms. We will use computational modeling and interactive collective innovation experiments to address the first goal. The second goal will be met by harnessing AI algorithms' unique abilities to explore and learn from others in ways distinct from humans. We aim to design and test algorithms that complement human collective innovation, enhancing its efficiency while mitigating human biases. Our results will both provide new insights into the strategies humans use for innovation and social learning while uncovering how AI can facilitate these processes to improve collective innovation. In an era where AI is poised to shape the course of human existence, studying the interplay between human and AI capabilities is essential. Through this project, we aim to shed light on what makes us human while harnessing the collaborative potential of humans and AI to drive collective innovation forward.

Social learning – learning from the behaviors of others – is key for our success as a species. Ideas and behaviors are transmitted between individuals and across generations via social learning, which gives rise to the evolution of human cultures. Social learning might also fuel modern threats, such as the growing social contagion of misinformation in social networks. Despite the crucial role of social learning across a myriad of human behaviors, surprisingly little is known about the (i) psychological, neural and computational mechanisms of social learning, and (ii) how these mechanisms create societal-level phenomena. The SOLAR project takes on the challenge of developing a new theoretical framework that addresses these issues. The framework will be formalized in novel computational models. The same coherent set of models will be tested on (i) the level of the brain using brain-imaging, (ii) the level of individual behavior, and (iii) the level of the population, using multi-agent simulation and analysis of real-world interaction in social networks. SOLAR involves three programs that address three fundamental, but as-of-yet unanswered questions about human social learning: (1) What are the mechanisms that produce social learning? (2) Does social learning drive social contagion? (3) How does social learning promote cultural evolution? In contrast to previous accounts, I predict that these seemingly disparate questions can be unified and understood by a coherent set of simpler social reinforcement learning mechanisms. I will test this hypothesis with a unique multi-method approach that bridges the “nano-level” of the brain, via the micro-level of human behavior, to the macro-level of the social group. SOLAR will shed new light on both fundamental scientific questions about the nature of human social learning, and on pressing social issues, such as how to explain and reduce collective risks due to social contagion (e.g. spreading of misinformation).