Ralph Hertwig, Ralf Kurvers, Pawel Romanczuk

Individuals rarely make decisions in social isolation. In most situations, individuals are subject to social influence. Social influence may be beneficial (e.g., increase decision quality), but it can also be detrimental (e.g., when false information cascades occur). To understand the emergence of collective intelligence—the shared intelligence that emerges from collaborative, collective efforts of individuals—we need to comprehend how individuals integrate personal and social information. A key aspects that has been largely neglected in human collective intelligence research is the dynamic aspect of information exchange. Most studies on human collective decision making assume that individuals simultaneously make decisions, which are then statically exchanged. In reality, however, information exchange is highly dynamic, and the timing of information exchange is linked to (subjective) information quality. Few studies have embraced such an approach; consequently, the dynamics of information flow in human groups remain poorly understood. To fill this important gap, this proposal has four objectives. First, we will investigate how single individuals integrate personal and social information under controlled conditions. Next, we will parameterize a dynamic decision making model to predict information flow and collective dynamics in real-time interacting human groups, and test these predictions. We then further parameterize our model to derive predictions for information flow across different network structures, and test these predictions. Finally, we will bring all these issues together, studying the conditions underlying collective intelligence in collective systems. The analytical system consists of human groups conducting experimental choice tasks. The synthetic component consists of drift diffusion models that address the cognitive processes underlying information integration. We continuously close the loop between analytical and synthetic systems, by using both approaches in concert. As end product, we will develop a versatile set of open-source algorithms (in CRAN R/Python) that can be used to study information integration processes in collectives, as well as for programming robotic swarms to achieve collective intelligence in the face of key challenges, such as speed-accuracy trade-offs6 or optimization at the individual versus collective level. Prior to release, the performance of these algorithms will be extensively tested with genetic algorithms.

Description of the postdoctoral project

A crucial component for achieving collective intelligence, being it in human, animal or robotic groups, is the accurate integration of personal and social information. We, however, know little how individuals in groups dynamically exchange information and update their decisions over time. Here we will investigate this process, using human groups as an analytical example. On the one hand, we will challenge human groups with visual search tasks in which individuals dynamically exchange information over time. On the other hand, we will use agent-based modeling to investigate the generalities of the collective information processing, the results of which will be fed back into the analytical system for testing. Continuously updating these algorithms, the end product will be a set of generic algorithms for studying and harnessing collective intelligence.

Applications should include: motivation letter, curriculum vitae, transcripts of records (for both BSc and MSc + doctoral degree if applicaple), copies of degree certificates (BSc, MSc), abstracts of Bachelor-, Master-thesis, e.g. doctoral thesis, list of publications and one selected manuscript (if applicable), two names of qualified persons who are willing to provide references, and any documents candidates feel may help us assess their competence.

A part-time employment may be possible. The Max Planck Society strives for gender and diversity equality. We welcome applications from all backgrounds. The Max Planck Society is committed to increasing the number of individuals with disabilities in its workforce and therefore encourages applications from such
qualified individuals.

Applicants must hold a PhD (or be close to completion) in Psychology, Biology or related natural sciences and should have proven skills/background in the following topics:

• conducting human collective behaviour experiments
• statistical data analysis of interdependent data (mixed model approaches, conventional statistics, yesian statistics)
• programming skills (for example in MATLAB, Python, R)
• strong preference for candidate with experience in drift diffusion modelling, genetic algorithms and/or models of collective behaviour
• a strong record of publication, reflecting the career stage, in internationally leading journals
• experience in working in collaborative research activities including multidisciplinary teams