Brain Signal Complexity (RITMO)
Information Dynamics Based Analysis of Auditory Perceptual Experiments
To facilitate perception and detection of new events, our brain relies on the knowledge of statistical structures in the environment. It is thought that it relies on such, in order to build an internal model of the immediate environment. This model constantly makes predictions, comparing its output to real world input (such as the sound of a tone). By doings so, the brain tries to minimize the prediction error between prediction and stimuli. Once the internal model shows a good fit between input and prediction, the brain is able to perceive named input. Accordingly, it is thought that the brain perceives through prediction or explaining away stimuli input. Unexpected events result in a higher prediction error or surprise compared to an event that perfectly fits into a chain of sequences (such as a musical scale).
Following this idea, which is named the theory of Predictive Coding (Clark 13), the measurable brains activity resulting from such an unexpected event should be qualitatively higher in comparison to an expected one (Sitt et al 2014). One way to estimate this activity is to use the idea of information content. Given the brain’s event response in the form of an electrical signal (i.e. measured through the method of EEG), it is possible to estimate the information content of it. That estimate can be then compared within a set of other event responses. Hence, it is possible to examine the hypothesis whether it is actually true if the information content, and thus the brain activity, following an unexpected event is higher in comparison to a highly expectable event.
This project will focus on exploring the application of the notion of information content to EEG recordings. Examples of such estimates are the measure of entropy, Kolmogorov complexity or Lempel-Ziv complexity (Sitt et al 2014, Rosas 2019). These measures can be then used to examine an available EEG data set such that it is possible to perform an qualitative comparison between different estimates or to further deploy on such apt measure to investigate the introduced perceptual theory of Predictive Coding.
Tasks for the project include:
- Doing a literature study of related approaches to the subject
- Implement information content / complexity measures for EEG signals
- Perform comparisons between the measures and analyse the results
- Writing the thesis report