Sommersemester 2017

Structural discrimination appears to be a persistent phenomenon in social systems. We here outline the hypothesis that it can result from the evolutionary dynamics of the social system itself. We study the evolutionary dynamics of agents with neutral badges in a simple social game and find that the badges are readily discriminated by the system although not being tied to the payoff matrix of the game. The sole property of being distinguishable leads to the subsequent discrimination, therefore providing a model for the emergence and freezing of social prejudice.

Every organism needs to gather information about its noisy environment and build models from that data to support future decisions. Information processing, however, comes at a thermodynamic cost. Stochastic thermodynamics is a powerful framework to analyse this interplay between information processing and dissipation in small, fluctuating systems far from equilibrium.

In this talk, we will first give an introduction to stochastic thermodynamics. In the second part, we focus on neural networks and show that the total entropy production of a network bounds the information it can infer from data or learn from a teacher. We will discuss a number of examples and give some directions for further research.

When choosing between multiple alternatives, people usually do not have ready-made preferences in their mind but rather construct them on the go. The resulting preferences are stochastic and highly dependent on the context, i.e., on the alternatives in the choice set and on any external reference points. Three context effects have particularly challenged traditional models of preferential choice: Similarity, attraction, and compromise effects. The simple choice tree model is able to produce these effects by means of a preference construction process and additionally accounts for the positive correlation between the attraction and compromise effects and the negative correlation between these two and the similarity effect. The preference construction process is based on attribute weights, differences between attribute values, and noise. A so-called focus value determines whether decisions are based on eliminations of unwanted alternatives or on direct choice of preferred options.

The lecture focuses on decision making processes in investment clubs, groups of retail investors who decide together where to invest their common money. These kinds of decisions are made under conditions of fundamental uncertainty, because financial markets trade expectations about future outcomes which goes along with uncertainty and complexity. Every investor has to deal with it, but making a decision in a small group influences the decision making in specific ways. How the decision is influenced by the group level is subject of this lecture and I refer to my ethnography of investment clubs in Germany. The influence can be differentiated in three dimensions; the informational, social and time dimension. In the time dimension, investment clubs are bound to a restricted time budget which frames decisions. Besides that they are compelled to make a decision at some points. On the informational dimension investment clubs offers on the one hand the opportunity to incorporate heterogeneous information in the process. Looking at the conversational level on the other hand, it attracts attention that most times reluctant information aren’t yielded in the process of decision making. I will argue, that this orientation to reach a common understanding of the situation in the social dimension has different, and sometimes irritating impacts on the group. Group Think can be a result of it. That maybe lowers the “quality” of the decision, but heightens the cohesion of the group. At the end of my presentation I will discuss what implications studying decision making in investment clubs has for our understanding of decision making under conditions of uncertainty and concepts of rationality.

For fast and effective behavior in a natural environment, the brain must restrict itself to the processing of relevant information for decision-making. It also needs to be able to switch its focus of processing flexibly in dependence of changes in the environment or a change of the motivational state, such as appetite or thirst. These requirements and restrictions imply a need for a mechanism, which selects and routes currently relevant information along the visual processing pathways, while it suppresses irrelevant information. Furthermore, this mechanism needs to be sufficiently fast and flexible in order to cope with changing demands on behavior. However, the timescale at which network functions need to change is too fast for changes of anatomical connections between neurons. Hence, it has been hypothesized that the selective processing of relevant information is based on fast changes of the correlation between neuronal activity patterns. This hypothesis is supported by the observation of attention-dependent modulations of the synchronization between neuronal populations encoding and processing the currently relevant information. However, so far there has been no demonstration of a causal relation between the modulation of synchronization and changes in selective signal transmission. Therefore, we applied weak signals by intracortical microsimulation within the visual system of macaque monkeys performing an attention task. We found that the impact of these additional signals, which were delivered to neurons in area V2, on downstream area V4 and on behavior strongly, depended on the phase of γ-oscillatory activity in V4. These findings strongly support the idea of temporal synchronization and desynchronization as a mechanism for selective information routing and processing.