To assess attention, psychologists use different tests. Two of them are the Trail Making Test (TMT parts A and B) and the Perception of Differences Test (PDT). The results are usually based on time employed and right answers. Eye-tracking technology may significantly improve these assessments gathering indirect latent information about the subjects’ performance. However, raw eye-tracking data interpretation still poses challenges, since most of the underlying processes are not well understood.
In this work we propose a novel analysis, based on statistical complexity measures, for time series obtained from an eye-tracking version of these tests. A total of 86 participants performed the PDT, and 65 performed the TMT. From each time series we calculated two probability distributions: position, related to where the subject is looking, and directional patterns [1], related to the directions followed by the gaze. For each distribution we calculated the Jensen-Shannon entropy and the statistical complexity [2]. The results were placed in an entropy-complexity plane, which displays typical specific features associated with different complex dynamics, showing that PDT behaves similar to the Logistic Equation (chaotic antipersistent behavior) while TMT behaves similar to fractional Brownian motion (chaotic persistent behavior).
[1] F. Avlia, C. Delrieux & G. Gasaneo. Eur. Phys. J. B 92, 273 (2019)
[2] M Zanin, L. Zunino, O. A. Rosso & D.Papo. Entropy 14, 1553 (2012)