BRAIN-COMPUTER INTERFACE INTEGRATED NEUROPSYCHOLOGICAL ASSESSMENT FOR QUASI-PSYCHOTIC SYMPTOMS

Abstract

In clinical psychology/psychiatry, quasi-psychotic symptoms refer to experiences such as suspiciousness or paranoia, brief hallucination-like experiences, ideas of reference, odd beliefs that are not fixed or delusional, experiences that resolve quickly, and insight usually retained. Studies using structured tools show that subclinical psychotic-like experiences (PLEs) occur in a portion of people in India. The present paper explores the potential of BCI-derived neural markers as objective indicators of quasi-psychotic states. BCI systems enable real-time acquisition and interpretation of neural signals, particularly electroencephalographic (EEG) patterns associated with perception, cognition, and self-referential processing. Alterations in neural oscillations, functional connectivity, and event-related potentials captured through BCI paradigms may reflect disruptions in reality testing, perceptual integration, and cognitive control.

In quasi-psychosis, BCI would monitor Neural Oscillatory Dysregulation (including increased theta activity, reduced alpha coherence, and aberrant gamma activity), Event-Related Potential (ERP) alterations (including reduced P300 amplitude and mismatch negativity changes), and Functional Connectivity Shifts (including alterations in prefrontal regions and temporal-parietal regions). BCI systems can track moment-to-moment neural instability, detect onset, intensity, and resolution of quasi-psychotic states, differentiate quasi-psychosis from anxiety or dissociation, full psychosis, and normative stress responses. This is particularly valuable in disorders like borderline personality disorder, trauma-related conditions, and mood disorders, where quasi-psychosis is episodic.

The conventional diagnosis using neuroimaging tools such as fMRI and CT scans is time-consuming, and error prone. Hence, Machine Learning models in BCI like Supervised Learning models (Support Vector Machines, Random Forests, Logistic Regression), Deep Learning models (Convolutional Neural Networks, Recurrent Neural Networks, Hybrid CNN-LSTM models), and Unsupervised Learning models (K-Means/Hierarchical clustering, Autoencoders) can be used to generate risk scores, and understand temporal trajectories. On a concluding note, this paper offers an interdisciplinary view of how BCIs can be helpful in assessing quasi-psychotic symptoms.