Figure 2.
An example of a two-dimensional test question (again, meant to demonstrate format rather than actual questions as they will appear in the BCI). In a graded model, the maximum reward is associated with a correct and confident answer and a penalty is associated with incorrect and confident answers.