Information-Theoretic Models of Communication in Biological Systems
Abstract
This thesis aims to find general principles governing the behaviour of biological systems, with a
particular emphasis in the communicational (social) aspect of these systems. Communication
between biological entities plays a major role in their evolution, enabling them to exchange
information about their environment and thereby improving their chances of survival. Communication
also plays a pivotal role in the organisation of populations of organisms, clearly
observed in social insects, but present also at least in bacteria, plants, fungi, animals and
humans. It is also theorised that the genetic code is a by-product of the establishment of an
innovation-sharing protocol between primitive cells [Vetsigian et al., 2006].
This thesis is mainly concerned with identifying necessary conditions for the emergence of
communicational codes, and the problems that arise with their establishment. For this purpose,
we introduce an information-theoretic framework where species maximise their growth
rate by following a Kelly-gambling strategy to bet on environmental conditions. Information
theory provides a powerful tool for abstracting away mechanisms and for focusing on hard
limits of a system's dynamics which cannot be circumvented.
We begin by exploring the relation between information exchange and limited resources.
We show that a transition from cooperation to antagonism in the exchange of environmental
information follows from a change in the availability of resources, from abundant to scarce.
We then assume a non-competitive scenario with abundance of resources, where conflicts in a
population occur only at a communicational (informational) level, rather than on the physical
level, such as competing for (physical) resources. However, traditional Shannon communication
is non-semantic, as opposed to the semantic communication observed in biological
systems, which is necessary for capturing conflicts in communication.
In the traditional use of information theory, it is assumed that every organism knows how
to \interpret" the information offered by other organisms. However, this assumes that one
\knows" which other organisms one observes, and thus which code they use. In our model,
however, we wish to preclude that: namely, we will do away with the assumption that the
identity of the organisms who send the messages and those who receive them is known, and
the resulting usable information is therefore influenced by the universality of the code used
and by which organisms an organism is \listening" to.
We introduce a model which captures semantic communication in information-theoretic
terms, where organisms talk to each other in a communication network. We show that, for
particular population structures, when organisms cannot identify which other organisms they
talk to, the adoption of a universal code emerges as a solution for full interpretation of the
shared information.
However, the evolution and establishment of universal codes for communication introduces
vulnerabilities: organisms can be exploited by parasites. We de ne two types of parasites
whose strategies have different levels of complexity and study the co-evolution of a host (the
population) and a parasite by optimising their respective objective functions in stages. First,
we consider a disruptive parasite (a troll) that inflicts harm in a host by minimising a population's
mutual understanding, and then a more complex parasite, which manipulates the
members of the population via their codes (the puppetmaster). We show emergent characterisations
of both parasites, as well as which host configurations are robust against parasites
and show adaptive properties.
This thesis introduces a framework which allows the study of informational properties
in the host-parasite co-evolution, where the rules of the parasite's habitat, the host, are the
outcome of an evolutionary process, and where these very same rules are those that allow the
parasite to exploit the host.
Publication date
2017-11-10Published version
https://doi.org/10.18745/th.19509https://doi.org/10.18745/th.19509