Detecting cognitive traits and occupational skills using EEG and machine learning algorithms

A new paper has been published by faculty members in the Q1 journal Scientific Reports.

Machine learning is widely used for classification tasks: for example, to detect various cognitive conditions or neurological diseases. Data are most often obtained using non-invasive methods, such as electroencephalograms (EEG). However, successfully detecting specific cognitive skills, such as maths skills, remains a challenge. 

A new study conducted by scientists from the Department of Biology and Biotechnology at the National Research University Higher School of Economics, along with IHNA & NPh RAS (a partner institute of the Cognitive Neurobiology educational programme), compares several machine learning algorithms to classify individual EEGs recorded in volunteers during verbal and mathematical tasks. Participants were divided into two groups based on their education and occupation: maths and humanities professionals. 

Three different machine learning algorithms demonstrated high accuracy in identifying EEG recordings belonging to individuals from either the ‘maths’ or ‘humanities’ group. Co-author of the paper O.V. Martynova, academic head of the Cognitive Neurobiology Department at the National Research University Higher School of Economics, comments: “Our results show that machine learning methods can be used to recognize individual cognitive traits from bioelectrical brain activity.”