Project No. 2360

STANDARD PROJECT

Primary Supervisor

Dr Ediz Sohoglu – University of Sussex

Co-Supervisor(s)

Prof Howard Bowman – University of Kent

Summary

How are we able to perceive the world despite the ubiquitous ambiguity of sensory information?

This question remains one of the great unsolved mysteries in neuroscience and has puzzled generations of thinkers. Many believe that the answer to this question is provided by a theory known as predictive processing. According to this theory, the brain is constantly predicting what is going to happen next and uses those predictions to fundamentally shape what is perceived.

The predictive processing theory is highly influential yet the underlying neural mechanisms remain unknown. A further key question is the relationship between prediction and other cognitive processes such as attention. Addressing these issues is challenging: How can we infer underlying mechanisms from non-invasive recordings of human brain activity that integrate over millions of neurons?

Advances in neuroimaging analysis mean that it is now possible to probe neural processing with unprecedented resolution. These methods, which include multivariate analysis and other machine learning tools, exploit subtle spatial or temporal patterns in neuroimaging signals that traditional methods cannot detect. This opens new avenues for investigation that this project will pursue in the context of speech processing.

Using electroencephalography (EEG) recordings of brain activity and multivariate analysis, the student will measure neural signatures of predictive processing by correlating moment-by-moment changes in speech predictability with brain responses. The focus of this project is to establish the relationship between predictive processing and attention: Do they operate independently of each other, in separable brain systems or with different time delays? Alternatively, are they fundamentally intertwined, interacting to determine perceptual outcomes? This research will advance our understanding of arguably the most influential theory in neuroscience. The student will also gain skills in machine learning and statistics that are highly sought after not only in neuroscience but also the data science industry.