Development of a Binary fMRI-BCI for Alzheimer Patients: A Semantic Conditioning Paradigm Using Affective Unconditioned Stimuli

Abstract

With the aim of developing a brain-computer interface for the communication of basic mental states, a classical conditioning paradigm with affective stimuli was used, assessing the possibility to discriminate between affirmative and negative thinking in an fMRI-BCI setting. 6 Alzheimer patients and 7 healthy control subjects participated to the study. Congruent and incongruent word-pairs were respectively associated to pleasant (baby laughter) and unpleasant (scream) affective stimuli. A Support Vector Machine classifier focusing on insula, amygdala and anterior cingulate cortex was used to discriminate between the activations relative to congruent and incongruent word-pairs (eliciting respectively affirmative and negative thinking), following the conditioning process. Classification accuracy was on average 71% for Alzheimer patients, reaching 85%, and on average 69% for control subjects, reaching 83%. This study shows that it is possible to extract information on individuals' mental states by exploiting affective responses, overcoming the typical obstacles of traditional BCIs, which generally require time-consuming trainings and intact cognition.