Enhancing transient motion-onset visual evoked potentials via stochastic resonance: Unimodal and cross-modal noise effects

Background

Motion-onset visual evoked potential (mVEP) are transient brain responses triggered by sudden motion stimuli and are widely used in brain-computer interface (BCI) systems. However, the inherently weak nature of mVEP signals poses a significant challenge to achieving reliable and accurate BCI performance. Enhancing the signal quality of mVEP responses is therefore critical for improving system robustness and usability.

New method

This study introduces a novel approach based on stochastic resonance (SR) theory, where appropriate levels of noise can enhance the performance of nonlinear systems such as the brain. By applying auditory and visual noise of varying intensities alongside mVEP stimuli, both unimodal SR and cross-modal SR effects were investigated. The method examines the effects of these noise conditions on brain activation and classification performance in mVEP-BCI.

Results

The results show that moderate levels of auditory or visual noise significantly enhance the P2 component amplitude of mVEP and improve classification accuracy in BCI tasks. In contrast, excessive noise leads to suppression of neural responses, forming an inverted U-shaped relationship between noise intensity and mVEP amplitude.

Comparison with existing methods

Conventional mVEP enhancement techniques typically rely on signal processing methods such as spatial filtering or feature extraction. In comparison, the proposed noise modulation strategy directly enhances neural responses, offering a biologically inspired and computationally simple alternative that complements existing approaches.

Conclusions

Both unimodal and cross-modal SR effectively enhance mVEP responses and BCI performance. This strategy provides new insights into SR mechanisms and supports the development of more robust mVEP-BCI systems.