Moving Past the Minimum Information Partition: How To Quickly and Accurately Calculate Integrated Information

An outstanding challenge with the Integrated Information Theory of Consciousness (IIT) is to find a way of rapidly and accurately calculating integrated information from neural data. A number of measures of integrated information based on time series data have been proposed, but most measures require finding the “Minimum Information Partition” of a network, which is computationally expensive and not practical for real brain data. Here, we introduce a novel partition, the Maximum Modularity Partition, across which to quickly calculate integrated information. We also introduce a novel detection task on simulated data to evaluate the performance of integrated information measures across different partitions. We show that integrated information can be reliably and quickly calculated across the Maximum Modularity Partition, as well as the previously proposed atomic partition, even in relatively large networks, which constitutes an advance in researchers’ ability to empirically test the predictions of IIT in real brain data.