Improving the signal-to-noise ratio of the visual P300

Abstract

The P3b (“P300”) [1] of the event-related potential is considered to reflect cognitive —or even conscious [2]— stimulus processing. P300 recording sessions employ an “oddball paradigm”: target stimuli appear only infrequently among non-target stimuli. Thus they are quite lengthy and difficult to use with patients. We tested different strategies for minimizing the time needed to record a reliable visual P300. Targets were gratings of different orientation for target and non-target stimuli. Since previous studies only assessed absolute amplitude [e.g., 3, 4], we first determined the optimal signal-to-noise ratio for different target infrequencies. We found target-to-non-target ratios of 1:4 to 1:8 to be optimal. We next assessed whether shrinking the interval between stimulus onsets from 1000 ms to 214 ms would improve the signal-to-noise ratio when the recording time is kept constant, despite a temporal overlap of responses. This was indeed the case by a margin of up to 60%, but depended on the presentation duration of the stimuli. Finally, we assessed whether it is feasible to record a P300 in a manner similar to a steady-state visual evoked potential with a fixed number of intervening non-targets. We found a high inter-individual variability, but could reliably identify responses in all subjects with a multi-harmonic frequency-space analysis. Compared to conventional P300 recordings, less recording time was needed to acquire significant responses. In conclusion, the signal-to-noise ratio and the efficiency of visual P300 recordings can be improved by choosing optimal target infrequencies and by employing rapid stimulation schemes. [1] Linden, D. E. J. (2005). Neuroscientist 11:563-576. [2] Kotchoubey, B. (2005). Progr Brain Res 150:427-444. [3] Johnson, Jr. R. (1986). Psychophysiology 23:367-384. [4] Polich, J., & Margala, C. (1997). Int J Psychophysiol 25:169-176.