Login | Sign up

PhD position in Computational Science: information theory meets causality

Complex adaptive systems typically consist of multiple variables (nodes, agents, particles) which interact in a non-linear manner through a heterogeneous network of interactions, thereby generating a non-trivial systemic emergent behavior which cannot be reduced to the dynamics of a single variable. The main idea behind information processing is that these variables can be interpreted as storing information (or memory); the interactions among variables can be seen as transmitting information from one variable to the other; and the decision of the new state of a variable based on all its interactions can be interpreted as integrating information (or information synergy). The original goal of this framework is to abstract away the mechanistic details of models (since regardless whether the variables represent neurons, birds, or molecules, the framework is solely in the language of ‘bits’) and thereby characterize emergent behaviors (e.g., tipping points, pattern formation, phase transitions) in a domain-free manner. Recently, though, the realization is growing that this concept may be related to a notion of causality. That is, if a causal interaction makes information transmit from A to B, then in what way does this transmitted information represent causal influence?

A different research field altogether is that of causal discovery, causal inference. The most classic types of analyses performed in this field regards at least one of the following: static or equilibrium states, linear interactions, and/or ‘overwhelming’ interventions. Since we are interested in complex adaptive systems this project will focus instead on non-equilibrium dynamics, non-linear interactions, and ‘underwhelming/stochastic’ (nudge) interventions. This is an atypical setting in the causality inference field. Moreover and more importantly, the fields of information processing and causality inference currently hardly overlap nor interact with each other. We see this as a missed opportunity since the two fields can learn from each other, potentially leading to ground-breaking new insights.

Duration/Period4 years
Deadline16/01/2020 0:00
Keywordsinformation theory; causality; complexity; dynamical systems; information flow; networks;

Recover your password

Please, provide the email you used to register. We will send an email to that address with a link that will take you to the reset password page. After resetting your password you will be automatically logged in to the system.

Already a member?

If you had an account on the old website, please provide the email you used to register there. After resetting your password you will be automatically signed in to the system.

Already have an account? Log in