AbstractsComputer Science

Groups of primates form complex spatial and social structures. Often, dominance rank is reflected in an individual's spatial position within the group, and individuals maintain individualized reciprocal relationships with affiliates, which reflect earlier interactions. The hypothesized underlying mechanisms and the cognitive capacities thought necessary differ in their complexity. While primates have been shown to possess advanced socio-cognitive abilities, computer models have proven that complex patterns may be generated by simple interactions between simple entities through self-organization. In this thesis, we equipped the simple entities of previous models with more primate-like abilities. Based on an extensive array of empirical data and some widely accepted (partial) hypotheses, we constructed a simulation model in which behaviors and cognitive capacities were implemented that have been shown (or hypothesized) to be employed by primates, such as aggressor avoidance, social vigilance, emotional regulation of behavior and emotional bookkeeping. Comparing these more elaborate models to simpler models allowed us to explore the implications of more advanced behavior or cognition for spatial and social group-level phenomena. We identified individual variation in several movement or perception properties, such as avoidance, social vigilance (i.e. monitoring and scanning) and even velocity, as important factors potentially structuring socio-spatial group properties. Moreover, we show that affiliative partner preferences, based on fixed properties or emotionally memorized interaction history (emotional bookkeeping), result in reciprocal affiliative relationships. However, only emotional bookkeeping allows for individualized, partner-specific relationships, given that partner-selectivity is strong and the timeframe of emotional bookkeeping is not too short and not too long.