A Type-2 Fuzzy Logic Based Explainable Artificial Intelligence System for Developmental Neuroscience

Research in developmental cognitive neuroscience face challenges associated not only with their population (infants and children who might not be too willing to cooperate) but also in relation to the limited choice of neuroimaging techniques that can non-invasively record brain activity. For example, magnetic resonance imaging (MRI) studies are unsuitable for developmental cognitive studies because they require participants to stay still for a long time in a noisy environment. In this regard, functional Near-infrared spectroscopy (fNIRS) is a fast-emerging de-facto neuroimaging standard for recording brain activity of young infants. However, the absence of associated anatomical image, and a standard technical framework for fNIRS data analysis remains a significant impediment to advancement in gaining insights into the workings of developing brains. To this end, this work presents an Explainable Artificial Intelligence (XAI) system for infant's fNIRS data using a multivariate pattern analysis (MVPA) driven by a genetic algorithm (GA) type-2 Fuzzy Logic System (FLS) for classification of infant's brain activity evoked by different stimuli. This work contributes towards laying the foundation for a transparent fNIRS data analysis that holds the potential to enable researchers to map the classification result to the corresponding brain activity pattern which is of paramount significance in understanding how developing human brain functions.