Decoding of semantic categories of imagined concepts of animals and tools in fNIRS

Abstract Objective. Semantic decoding refers to the identification of semantic concepts from recordings an individual’s brain activity. It has been previously reported in functional magnetic resonance imaging and electroencephalography. We investigate whether is possible with near-infrared spectroscopy (fNIRS). Specifically, we attempt differentiate between categories animals tools. also identify suitable mental tasks for potential brain–computer interface (BCI) applications. Approach. explore feasibility a silent naming task, first time fNIRS, propose three novel intuitive based on imagining using sensory modalities: visual, auditory, tactile. Participants are asked visualize object their minds, imagine sounds made by object, feeling touching object. A general linear model used extract hemodynamic responses that then classified via logistic regression univariate multivariate manner. Main results. successfully classify all mean accuracies 76.2% 80.9% visual imagery 72.8% auditory 70.4% tactile task. Furthermore, show consistent neural representations exist applying classifiers across tasks. Significance. These findings fNIRS. The study step toward use BCI applications communication.