A brain-computer interface (BCI) gives those suffering from neuromuscular impairments a means to interact and communicate with their surrounding environment. A BCI translates physiological signals, typically electrical, detected from the brain to control an output device. A significant problem with current BCIs is the lengthy training periods involved for proficient usage, which can often lead to frustration and anxiety on the part of the user and may even lead to abandonment of the device. A more suitable and usable interface is needed to measure cognitive function more directly. In order to do this, new measurement modalities, signal acquisition and processing, and translation algorithms need to be addressed. This work implements a novel approach to BCI design, using noninvasive near-infrared spectroscopic (NIRS) techniques to develop a userfriendly optical BCI. NIRS is a practical non-invasive optical technique that can detect characteristic haemodynamic responses relating to neural activity. This thesis describes the use of NIRS to develop an accessible BCI system requiring very little user training. In harnessing the optical signal for BCI control an assessment of NIRS signal characteristics is carried out and detectable physiological effects are identified for BCI development. The investigations into various mental tasks for controlling the BCI show that motor imagery functions can be detected using NIRS. The optical BCI (OBCI) system operates in realtime characterising the occurrence of motor imagery functions, allowing users to control a switch - a “Mindswitch”. This work demonstrates the great potential of optical imaging methods for BCI development and brings to light an innovative approach to this field of research.