Wavelet-based image decomposition affects SSVEP signal amplitude for face identification

Previous studies have predominantly examined N170 sensitivity in a binary manner, focusing on the presence or absence of distinct facial features, either independently or within a facial context (e.g., Parkington & Itier, 2018). However, recent work has suggested that the N170 operates more like a continuum, with amplitude increasing as diagnostic information accumulates (Audette et al., 2023). In parallel to the study of ERPs, the method of steady state visual evoked potentials (SSVEP) has been instrumental in exploring neuronal responses to oscillating visual stimuli, shedding light on the brain's capacity to synchronize with and process visual information across various frequencies. Seeking to replicate the amplitude continuum observed in ERPs, we utilized SSVEP, incorporating wavelets into our stimuli to enhance decomposition while preserving low-level information. Presenting modified faces at five decomposition levels (0 to 20%) and three flickering frequencies (4, 5, or 6 Hz) to 11 observers, we implemented an oddball paradigm featuring identity changes (AAAAAB). Participants completed 45 trials of 53 stimulation cycles, encompassing three trials for each of the 15 conditions. Our results suggested no effect of stimulus presentation frequency (F(10) = 0.45, p = 0.630) but high responsiveness to the level of decomposition in presented faces (F(10) = 8.65, p < .001). In essence, as the diagnostic information in faces decreased, neural activity synchronization to identity diminished. In other words, the less diagnostic information was available in faces, the less the participant’s neural activity synchronized to the change in identity. These findings not only advance our comprehension of cognitive processes in face recognition but also hold promise for optimizing facial feature extraction in real-world applications.