The aim of this thesis was to create a decision-making domain, in which multiple agents would collectively engage a problem, without being able to communicate with each other. Furthermore, the group decision making was structured such that each agent would always have an influence over the outcome, but could not control the impact of their decisions. Different information conditions simulated information asymmetries, from which potential behavioral changes were to be analyzed. Agents were able to build up expertise in a well-defined learning environment, and later engaged in an ill-defined, metastable and instable decision-making domain, which was either dominated by seemingly deterministic or chaotic feedback. In order to create a problem under uncertainty, “Tower of Hanoi” was chosen as the problem for analysis, which lacks any numerical representations, and thus further avoids subjective or even objective probabilities being built up by human mental models. Different variations of logic, strategies, and feedback were examined in order to derive as much information as possible in this group decision making experiment. The core idea was that this experiment represents reality, where an agent would first gain experience and learn about the systematics of a market, (e.g. by visiting a business school), engaging in well-defined problems. Upon having gained some expertise, which varies amongst the agents, they could then explore the real world, and solve ill-defined problems with their expert knowledge. Real world problems were first simulated as metastable, changing to a more chaotic problem afterwards. Many economic decisions are taken without communicating directly with all shareholders- or stakeholders, and agents collectively solve ill-defined problems, with each agent having different sets of information, and different strategies and ideas about the “hidden rules” of some market or complex decision-making domain. From this idea, five different research questions lead to 10 different hypotheses. The first research question probed, whether public information about environmental change would necessarily lead to a change of behavior, when the new environmental conditions would not have an impact on some strategy’s performance. The routine information group has proven that this was not the case. Agents in the routine information group stuck to their routine strategy from the well-defined problem-solving stages during the metastable condition. The second research questions asked, whether change in behavior was the case if environmental change actually does have an impact on some strategy’s performance; here, individual expertise has proven to be a strong predictor, of whether or not an agent was able to adapt or stick to an effective routine strategy. High expertise lead to less random and volatile behavior in the ill-defined problem-solving stages, and enabled agents to adapt quickly to environmental conditions in well-defined stages. The third research question regarded deviation from routine strategy when different types of information, their contents being truthful and deception-free, were provided. Here, results were not so clear, and individual expertise was certainly a stronger predictor than was public information. The fourth research question can also be answered by focusing on individual expertise, rather than on public information: the higher the individual expertise in the well-defined problem-solving domain, the higher the chances were that participants would maintain an effective routine strategy or adapt their routine if necessary. While public information did not significantly influence the overcoming of parts of a routine strategy, it seems that the dissolution information group deviated the most. Perhaps public information about the individuals being unable to obtain helpful information about the hidden rules discouraged agents, favoring random behavior or absorbed individual motivation to engage in problem-solving with smart heuristics. Further research on the influence of public information that favors a belief of lack of control could shed light on this assumption. The fifth and final research question was partially answered. Individual expertise in the well-defined problem-solving stage showed a strong significant correlation with behavior in the ill-defined stages. While the experiment failed to come to conclusions about group performance, the role and impact of individual expertise was surprising, truly holding more predictive power regarding group decision-making than public information.

All hypotheses were analyzed in detail for gender effects and no convincing differences in behavior between female and male participants led to the assumption that gender effects were found. Just as for NPS performance, where no gender effects were found for any age or country-origins, such as US America, India and Germany, solving ToE in a smart and intuitive way disregarded gender effects. If anything, female participants outperformed in strategy adaption during well-defined stages, which was a crucial part to rank high in individual expertise. Obviously, this result is very favorable for the idea of inclusion in modern workspaces, where NPS performance and smart decision-making under uncertainty will play an ever-growing role.

Expert knowledge could be the key factor for global and interconnected problems, where interpersonal communication is impossible or vastly limited. Identifying the ideal decision-making positions for experts through quick and effective online experiments could lead to less volatile, less chaotic system performance, from which all decision-makers can profit.