An open-source approach for closed loop cognitive studies

Spanish Society for Experimental Psychology (SEPEX)

Almeria. España
Congreso. 2024

The study of cognition and perception involves establishing meaningful correlations and causal relationships with the environment; a black box approach entered by strictly controlled inputs that generate a measurable output. Here, we present MILPS (Multisensory Information Labeled Processing System): an open-source project that, on the one hand, enables performing perceptual experiments with an inexpensive setup; on the other hand, it will offer a novel Input Inference Label Approach that frees the experimenter from a traditional black box approach. Under the active inference framework, the dynamics of the Bayesian brain and the external world are understood as action perception loops that self-evidence by prediction error minimization. MILPS will offer an improvement to the classical approach to directly infer perception through the action that a subject is performing, adding less constraints to input control, in such a way that we can observe these adaptive systems, these action-perception loops, in the most ecological way tagging actions as events used by the subject to minimize prediction error. Finally, MILBPS offers an open-source repository for perceptual experiments that synchronize data acquisition and data analysis from eye tracking, Emotibit and EEG recordings. Further work will enable the addition of gesture-tracking data and other devices. We have already replicated studies to validate our set up and we will continue moving forward for further improvements.