Javascript must be enabled for the correct page display

A brain-computer interface based on pupillometry using the rapid serial visual presentation paradigm

A.K. Kleppe, Ko (2018) A brain-computer interface based on pupillometry using the rapid serial visual presentation paradigm. Master's Thesis / Essay, Human-Machine Communication.


Download (1MB) | Preview
[img] Text
Restricted to Registered users only

Download (100kB)


Brain-computer interfaces (BCI) use brain signals to interact with external devices (computers, prosthetics). A brain-computer interface can provide a means of communication for severely disabled people. We created a BCI based on the pupillometry method using a paradigm whereby stimuli are presented in rapid temporal succession, this paradigm is known as rapid serial visual presentation (RSVP). The pupil constricts and dilates in response to brightness. Pupil size is not solely dependent on retinal input but is guided by cognitive influences such as attention. It is well documented that attention and the pupil light response interact in such way that the pupil dilates when you attend a dark object, we have coined this phenomenon as the interaction effect. In our first experiment, in which the participant had to pay attention to letters, we investigated under what conditions this interaction effect would manifest itself. Participants were instructed to pay attention to a prespecified letter while a RSVP stream of twenty-six letters was presented to them. Our hypothesis was that the mean pupil size area would differ across RSVP streams with different target brightness. The interaction effect was not found in our first experiment presumably due to the absence of the pupil response associated with attention. In our second experiment we built a BCI which inferred which target the participant was paying attention to in real time. Rather than letters, participants had to select, by paying attention, either a triangle or a diamond. This method is an instance of the method known as two-forced alternative choice (2AFC). The data of the second experiment displayed a strong learning effect and the average perfor- mance of 59.11% was significantly above-chance level. Moreover, our BCI achieved an information- transfer rate of 0.06 bits/min. In conclusion, the RSVP paradigm in combination with the pupil- lometry method does show some potential for creating an effective BCI.

Item Type: Thesis (Master's Thesis / Essay)
Supervisor name: Mathot, S. and Borst, J.P.
Degree programme: Human-Machine Communication
Thesis type: Master's Thesis / Essay
Language: English
Date Deposited: 05 Oct 2018
Last Modified: 08 Oct 2018 12:57

Actions (login required)

View Item View Item