An Agent-Based Model of Leader Emergence and Leadership Perception within a Collective

Figure 1

A schematic illustration of our agent-based model. A leaderless group consisting of n agents trying to find the maximum point of the true utility function displayed in the top-left corner, which is not directly accessible from any agent. Agents have limited knowledge about the true utility function. Prior to discussion, each agent initially samples a certain number of opinions from the true utility function as its inherent opinions, whose quantity and quality depend on its intelligence. Agents reconstruct models of the utility function as their individual utility functions from those memorized opinions. Meanwhile, each agent also picks one plan with highest utility in its memory as its current best plan. The individual utility function and the best plan will be updated as new opinions are added to memory. In the planning process, the group members either keep silent and think about the problem by themselves or discuss about the problem openly. When silent, each agent studies a new opinion around its best plan by itself, then updates its individual utility function. Also, it changes its best plan if the new opinion is better. The speaking process goes as follows. First, a speaker proposes a suggestion of modifying one aspect of the group plan. Second, other agents (listeners) evaluate the speaker’s perceived leadership according to the quality of its suggestion and the performance of its presentation. Finally, the listeners decide whether to adopt the speaker’s suggestion or not at individual and group levels. If the new plan is supported by more than half of the listeners, the group plan will be changed according to the speaker’s suggestion, otherwise the group plan remains unchanged. This cycle repeats for a certain number of iterations.