A comparison of two information theory approaches to animal communication systems

Abstract

Information theory has been used to analyze and explore animal communication systems using two different approaches. The first approach—which is the original application outlined by Shannon and Weaver (1949)—considers the temporal distribution (i.e., probabilities of occurrence) of the components of a communication system to quantify the diversity and complexity of a system’s (repertoires) organization. This quantification of the communication complexity is based on the information content, not just at the repertoire level, but also at the sequential (i.e., Markovian) organizational levels of the signals constituting the communication system. The second approach, which has been much more widely applied, with mixed results over several decades, considers a communication system’s ability to essentially ‘‘encode’’ signaling between individuals in order to measure how much information has been ‘‘transmitted.’’ We will outline and contrast these two approaches, pointing out that until the communication complexity of the entire signaling system is quantified, there is little hope of correctly interpreting information exchange between individuals using that system of communication [C. Shannon and W. Weaver, The Mathematical Theory of Communication (University of Illinois Press, 1949)].