Signal and noise in P300 recordings to visual stimuli

Abstract

The P300 of the event-related potential is typically obtained for infrequent target stimuli that are embedded in a sequence of frequent irrelevant stimuli. The P300 has been suggested as a marker of high-level cognitive processing and might be useful in ophthalmology to confirm the diagnosis of a functional disorder. However, typical P300 measurements require relatively lengthy recording sessions. It would therefore be desirable to minimize the required time and to maximize the signal-to-noise ratio by finding the optimal balance between parameters such as stimulus probability and the number of target trials or the recording time. This is different from previous studies, which assessed the amplitude only. We recorded event-related potentials to visual stimuli using standard oddball paradigms with various target frequencies ranging from 2:1 (target majority) to 1:16 (massive non-target majority). We compared the signal-to-noise ratios for a fixed number of target trials as well as for a fixed total recording time and assessed effects of the immediate stimulus history. As expected, P300 amplitudes depend strongly on target infrequency. This did not reach saturation within the range tested. For a given number of target trials, the signal-to-noise ratio also increases with target infrequency. For a given recording duration, the signal-to-noise ratio is optimal around 1:8. In the 1:4 condition, the signal-to-noise ratio can be improved by excluding trials that were preceded by a target trial.

Closing the loop in cortically-coupled computer vision: a brain–computer interface for searching image databases

“…Interestingly, the performance gains in the interest detector are less consistent when target prevalence increases beyond 10-15%. This is reminiscent of P300 oddball studies that have found the strength of the P300 response increases for more infrequent target appearances [45][46][47][48][49], and which commonly employ target probabilities of 20% or less. While the interdependence between the HV and CV modules could theoretically lead to situations in which each module waited for the other to first achieve high performance, the closed-loop aspect of this system allowed the modules to gradually reinforce each other.…”

Robust Single Trial Identification of Conscious Percepts Triggered by Sensory Events of Variable Saliency

“…1 B), series of frequent (Standard) stimuli were interspersed with rare stimuli (Target), with a target frequency ratio of 1∶16 (target-to-target frequency ratio that elicits larger P300 amplitudes over midline electrodes) and associated occurrence probability of 6% [26].…”

Increasing BCI communication rates with dynamic stopping towards more practical use: an ALS study

“…The standard approach to increase the P300 SNR within noisy EEG data is to average data from multiple trials for improved accuracy in classification performance (16). This means each potential target character is sampled multiple times before the classifier determines the desired character, increasing the time needed to make a selection and decreasing the communication rate.…”