Brain-based measures of consciousness need to map onto what conscious experiences are actually like. A new class of mathematical measure, based on ‘dynamical complexity’ attempt to achieve exactly this. They are based on the insight, originally outlined by Giulio Tononi and Gerald Edelman (Science, 1998), that conscious scenes are simultaneously differentiated (each conscious scene is one among a vast repertoire of alternative possibilities) and integrated (each conscious scene is experienced ‘all of a piece’, as unified). At the Sackler Centre we are developing new measures of dynamical complexity and applying them both to sophisticated computational models and to experimental data. The key concepts are causal density and integrated information. Causal density uses Granger causality to measure the average level of causal interactivity among the elements of a system – it will be high only when different parts of a system do different things, but when there is also overall coordination among these different parts. Integrated information (based on the work of Tononi) measures the amount of information generated by a system as it transitions from one state to another. At the Sackler Centre, we have developed the first practically applicable versions of these measures which we are now applying to electrophysiological data obtained from people in various conscious levels (awake, asleep, anesthetised, and so on).
People:
Reference:
Seth, A.K., Barrett, A.B., and Barnett. L. (2011). Causal density and integrated information as measures of conscious level. Phil Trans R. Soc. A. 369:3748-3767