Error-driven learning of communication systems in biological life forms

Life forms on Earth use a vast array of methods of communication, within which there is a wide range of complexity and adaptivity. As far as is currently known, the systems most capable of adapting to new input and new communicative needs are human languages. This seems to result from the high degree of reliance on learning, relative to genetic encoding, in human linguistic systems. We propose that the relative contribution of learning versus genetic encoding will have a similar effects on complexity and adaptivity in extraterrestrial communication systems. We discuss this issue from the perspective of discriminative learning, a learning theory that has accounted for many learning effects in animals, including in human language. We investigate two aspects of the human communicative signal that are likely to differ in extraterrestrial life forms compared to human speech - atmospheric effects and biological effects - and how variation in these characteristics on other planets might impact the signal. Our analyses show that even subtle differences in these attributes have a substantial impact on signal properties.