Figure 1.
Stimulus spatial organization in the P300 BCI comparing with the oddball paradigm.
(A) Visual oddball paradigm. (B) Matrix (“classical”) P300 BCI layout with stimuli grouped into rows and columns. (C) P300 BCI layout with fixed arbitrary stimulus positions and single-cell presentation mode (without grouping). (D) P300 BCI layout with moving stimulus positions and single-cell presentation mode (a design used in this study). S, standards (non-target stimuli). T, targets (target stimuli). In these examples, a flashing letter B is the target stimulus. Note that the content at the location that should be attended (marked with a red circle) significantly varies with sequential presentations in the oddball paradigm (A) (both targets and standards are presented there), whereas in the P300 BCI (B, C, D), the attended location can be in one of two states only (target stimulus on/off; the standards are presented at other locations).
Figure 2.
Example of the P300 BCI game Billiard Puzzle display.
Each “bouncing ball” contained a fragment of the picture being assembled on the right. The correct order of targets is cued by letters of the Russian alphabet. The current target is marked by a circle in the right panel. The ball with the letter “B” is flashing here. A counter at the bottom right indicates the number of errors (here, four errors). Note that this display follows the scheme presented in Figure 1 D (moving stimulus positions and single-cell presentation mode).
Figure 3.
Online classification results for overall and corrected classification accuracy.
Black diamonds, Single-trial group. Gray squares, Triple-trial group. Means ± standard deviations, in %.
Figure 4.
Interest was self-estimated by the participants using a visual analog scale (VAS) at the end of each session. Black diamonds, Single-trial group. Gray squares, Triple-trial group. Means ± standard deviations, in % (corresponding to mm on a 100 mm-long scale).
Figure 5.
Grand average event-related potentials.
Here, n=5 for each session in the Single-trial group (participant 8 excluded) and n=4 for each session in the Triple-trial group (participants 10 and 12 excluded). O* denotes an average of four occipital channels (PO7, PO8, O1 and O2). Zero time corresponds to the beginning of the stimulus. Waveforms are presented in black for the 1st session and in dark, medium and light gray for sessions 2, 3 and 4, respectively.
Figure 6.
Individual averaged event-related potentials.
For non-target ERPs, only data from representative participants are shown (P04 and P01 had relatively high-amplitude averaged responses to non-targets, whereas P07 and P10 represented those with especially low-amplitude non-target-related averages). O* denotes an average of four occipital channels (PO7, PO8, O1 and O2). Zero time corresponds to the beginning of the stimulus. Waveforms are presented in black for the 1st session and in dark, medium and light gray for sessions 2,3 and 4, respectively.
Figure 7.
Amplitude of event-related potential peaks N1 and P300.
Black diamonds, Single-trial group. Gray squares, Triple-trial group. Means ± standard deviations, in µV.
Figure 8.
Dynamics of the individual index values across the four sessions.
Each circle depicts the index value per session and participant. Lines connect the index values for each separate participant. For classification accuracy, overall accuracy data are shown by empty diamonds connected by dotted lines, and corrected accuracy data are shown by filled diamonds connected by solid lines. P300 amplitude data could not be measured for two participants or during one session for another participant.