Signaling is a very basic biological phenomenon. It is present at almost all levels of biological organization and it plays a crucial role in nearly all biological processes. Although these processes comprise a very diverse set of phenomena, there is a common core to their structure: at least one sender interacts with at least one receiver. This allows one to apply the mathematical theory of games, which is an abstract theory of strategic interactions. The goal of this project is to identify stable evolutionary outcomes of signaling interactions by using methods from evolutionary game theory. Although some properties of signaling are well understood, our knowledge of how reliable signaling emerges in a dynamic process is limited, especially in cases where there is substantial conflict of interest between a sender and a receiver. If it is not in the sender's interest to transfer all relevant information, how can reliable signaling be achieved? This question is at the heart of a huge number of problems in biology. It also plays a very prominent role in many economic interactions. This project will provide important new results for a wide array of research programs and will open up unknown paths for experiments and field studies. The project involves the training of undergraduate and graduate students and the development of educational tools.
Signaling can be found at all levels of biological and social complexity. There is between-cell and within-cell signaling. Bacteria signal, as do plants. It is well known that many animals use signaling systems in the presence of dangers or to express need. Humans, of course, have langauge. Moreover, companies use signals in their interactions, which is true also of institiutions, political parties, countries, and so on. Since signaling is ubiquitous, it is of great importance to understand its basic structure, how it emerges, and how it changes over time. In this project we studied signaling at an abstract level with the help of mathematical models. The models we used are rooted in game theory, which has been developed in economics since the middle of last century. Game theory is a general theory of interactive decision making. As such game theory can applied to any situation that involves social interactions, broadly speaking, even if the interactors are not consciously making decisions. We are studying games that are called "signaling games". In signaling games there is a sender (sometimes more than one) and a receiver (also sometimes more than one). The sender has some information that would be valuable for the receiver. The sender can send a "signal" to the receiver -- but without that signal having any meaning as of yet. The question now is: is it possible for senders and receivers to develop a signaling system where signals are used systematically in order so that the receiver gets the sender's information in a reliable way? The answer to this question depends on specific aspects of the signaling interaction. For instance, the interests of senders and receivers may be completely aligned. In this case perfectly reliable information transfer is typically stable. The emergence of such outcomes can be investigated in the context of dynamical systems, that is, systems that describe how the behavior senders and receivers change over time. Reliable information transfer is a fairly robust outcome in this dynamic sense as well. But the interests of senders and recievers need not be aligned. At the other extreme, their interests might be completely opposed. There are many inbetween cases where interests only partially overlap. These cases were the focus of our project. Perfectly reliable information transfer is often not possible in these situations, because it would be unstable or too costly to maintain. A perfectly reliable sender could be exploited by a receiver against the sender's interest, and the other way round. But this does not mean that there is no information transfer. If one analyzes the mathemtaics of the signaling dynamics in detail, a rich variety of possible intermediate outcomes emerges. These outcomes allow general and basic insights into the dynamics of signaling. In particular, they provide us with qualitative predictions and explanations of outcomes in real life. We tested some of these predictions in the lab with human subjects. Our results revealed a fairly close match between theoretical predictions and actual human behavior. The results are significant for a variety of fields, including theoretical biology, animal behavior, economics, philosophy of science, and computer science. This list could probably be continued, since the generality of the models and the ubiquity of signaling allows the results to be useful in many domains -- although concrete applications require further adjustments and specifications, as always. Broader impacts become relevant whenever it is important to understand the basic structure of an interaction that involves signaling in one way or another. This can be the case in negotioans between companies or in many of our day-to-day social interactions. The signaling theory developed in our project helps us classify different types of communciation structures, assess how much honest or reliable communication is possible given the incentives to participants, and suggests ways in which the incentive structure has to be changed in order to make way for more desirable ways of information transfer.
