No surprise - fixed sequence event-related potentials for brain-computer interfaces

Abstract

Abstract-Introduction: In the field of Brain-Computer Interfaces (BCI), the original two-class oddball paradigm has been extended to multiple stimuli with balanced probabilities and random presentation sequences. Exploiting the differences between standard and deviant ERP responses, these multi-class paradigms are suitable for communication and control. Methods: The present study investigates the effect of giving up the randomness of stimulation sequences in favor of a repeated, predictable pattern. Data of healthy subjects (n=10) who performed a single session with a 6-class spatial auditory ERP paradigm were analyzed offline. Their auditory evoked potentials (AEP) resulting from the potentially simpler task (using fixed sequences) are compared with the AEP evoked by pseudo-randomized stimulation sequences. Results: Class-discriminative EEG responses between target and non-target stimuli were observed for both conditions. The binary classification error estimated for standard epochs of was comparable for both conditions (random: 24%, fixed: 25%). Expanding the standard epochs to include pre-stimulus intervals, we found that the regular structure of the fixed sequence can be exploited. Compared to the standard epoch, the MSE improves by 7%, while in the random condition an improvement could not be observed.

Decoding auditory attention to instruments in polyphonic music using single-trial EEG classification

“…These restrictions do not apply to many kinds of real music. However, as stated before, abandoning the random sequences of an oddball paradigm and turning to more musical, and hence more predictable, structures is viable possibility as suggested by recent evidence on BCI classification using fixed sequences [33]. Furthermore, the complexity of the musical piece (by increasing the number of instruments and/or number of patterns per instrument) can probably be further increased.…”

“…The arbitrariness of the deviant patterns within the flow of music to some extent violates the musical structure, which, especially in Pop music, is simple, repetitive and therefore highly predictable. However, recent evidence demonstrates that switching from random sequences to fixed sequences does not hinder performance and that it can, on the contrary, even improve it [33]. Thus, it appears to be possible to make the musical oddball paradigm more musical by abandoning the classical oddball paradigm.…”

Probabilistic simulation framework for EEG-based BCI design

“…A sequence is a burst of trials that present several symbols (the entire alphabet or a carefully selected subset) in RSVP mode. During the presentation of a sequence, currently restricted to not present any symbol more than once, the symbols are ordered randomly for the purpose of strengthening the response for the intended symbol; however, recent evidence from auditory P300 BCIs indicate that randomization may not be crucial for successful operation [33]. The selection of a single symbol in the typing process occurs in an epoch , a period that consists of one or more sequences depending on when sufficient confidence is built based on the language evidence and accumulated multiple EEG evidence for candidate symbols.…”

Improving the performance of a gaze independent P300-BCI by using the expectancy wave

“…Tangermann M studied the effect of fixed stimulus sequence on the repeated and predictable mode of auditory BCI. The results showed that the effect of fixed sequence was even improved compared with random sequence [28].…”