While much BCI work has traditionally focused on providing an alternate input mechanism for the (physically) disabled, we have worked on projects exploring the feasibility of reducing the cost of using such technologies so that we can enable more research in the space, but also ultimately deliver utility to able-bodied end users.

To this end, we have shown in a series of studies that we can use the electroencephalograph, or EEG, to classify both general task activity but also cognitive or working memory load with high accuracies (up to the 90% range for 3 task classifications). This is true even when using small temporal windows (2-4 seconds), very little training data (order of 5-10 minutes), simple EEG equipment (2-4 channels), and relatively simple computational techniques. We have also shown that some of the models we build generalize across task variants and users, which bodes well for real-world use.

In addition to exploring the feasibility of BCIs for user state evaluation, we have also started to develop novel applications that use BCIs as an information source. Human-Aided Computing is an approach that uses brain sensing technologies to measure the presence and outcomes of implicit cognitive processing, processing that users perform automatically and may not even be aware of. In this work, we demonstrate a proof-of-concept system that uses cognitive data measured with an electroencephalograph, or EEG, to classify images (e.g. into faces, animate objects, and inanimate objects) shown to a user, even when the user is not told to do the task.