MATERIAL ATTENTION: A METHOD TO MEASURE ATTENTION ON DECISION MAKING, USING INFORMATION THEORY AND MARKOV CHAIN MODELS

We present a method to quantify the Regime of Attention (RoA) characteristic of an agent or population that engages with the material world. For this, we derived an information theory-based method to characterize the flow of uncertainty of an agent that synchronizes with a process. Using eye tracking data, we built Gaze Entropy Profiles (GEP) transforming gaze actions patterns into a hierarchical structure, with two levels: (i) At the top level, block entropy captures the global behavior of action sequences, and (ii) the entropy path (EEP) allows us to see actions as a process of discrete states—cycles of exploration and exploitation—that can be modelled as a Markov process. We propose that: (1) such a process may be viewed as a special case of variational free energy minimization; and (2) that the transient information (T), understood as the amount of information an observer needs to extract to synchronize with a process, is a measure of how much information has been efficiently externalized to participate in cognitive processing without being stored in the brain. We demonstrate that the proposed method describes, from the bottom-up perspective, how materiality elicits a characteristic RoA, and from the top-down perspective, how an agent's or population's generative model constrains the RoA. In conclusion, our method proposes a formal way of using information theory to build a quantifiable definition of attention that can be used experimentally and can be extended to other neurophysiological signals. ​