A BCI Framework using Derivative Similarity Metrics to Characterize EEG Spatiotemporal Dynamics with Applications in Neuromarketing

A real-time brain–computer interface (BCI) framework for social robots to predict impulsive decision-making and hidden emotional responses is proposed. A novel neuromarker is defined as sliding-window Pearson correlation of second derivatives of EEG band power across channels within the early reaction interval (0.5-2s). Validated in a neuromarketing pilot study using Furhat robot, the BCI successfully classified participants' intentions to "like", "dislike" or "purchase" advertised products Results showed strong polarity-dependent neural differentiation (inverse correlations, r < –0.60), most prominently in the theta band (AF3, F3, C3, AF4, Pz, P4). This dynamic analysis was complemented with effect-size statistical evaluations (Cohen’s d). Convergence criteria (|d| ≥ 0.60) revealed robust neuromarkers primarily in frontal-parietal theta-band electrodes, alongside complementary alpha (F3, Pz, T8, T7) and beta band (Cz, Pz) contributions. These findings demonstrate that the proposed derivative-similarity neuromarker provides a computationally efficient and statistically robust approach for robot-mediated BCI decoding of impulsive evaluative decisions.