**4. Discussion**

The present study provided initial evidence for (i) cortical oscillatory responsivity to speed modulation in videos either representing a naturalistic scene (a video clip of ripples on the beach) or RDK (with no directional information), and (ii) dissociation between perceptual and cortical responses, consisting in the presence of cortical responses with the perceptually invisible lowest stimulus strength.

#### *4.1. Shape and Distribution of Cortical Responses*

In some cases, SSVEPs presented rather smooth, sinusoidal-like signal variations, while in other cases more abrupt changes were observed. In this preliminary study, we took an agnostic position as to the actual shape of the cortical response and used sinusoidal fitting as a simple method to quantify the periodicity of scalp potentials, at the cost of losing the details of the full cortical response. We remark, however, that the choice of using sinusoidal fitting was by no means meant to imply that in our conditions SSVEPs had sinusoidal-like waveforms, as is the case with traditional SSVEP protocols where higher stimulus frequencies are used [12]. Given the relatively low frequency of our stimuli, an alternative way to analyze the data would be to use the traditional event-related approach (e.g., finding the peaks and latencies of the various components), which would require, however, a more thorough and systematic consideration of the actual appearance of this particular speed-related cortical potential, including identifying the relevant components. Future work will allow full characterization of this cortical response.

A somewhat unexpected finding was an apparent lack of clear regional differentiation in the amplitude of cortical responses. Luminance-evoked VEPs, as well as cortical responses to motion onset [13–17], involve mainly posterior cortical areas. Yet, in our study, the coefficients for the random effect of a channel were largely overlapping, with only a single electrode (F4) showing a contribution larger than the other ones. One possible explanation is that the oscillatory rhythm we impressed on video clips results in global cortical entrainment, with several components at play (e.g., sensory, attentional, motor, imaginative). Indeed, the instruction to mentally focus on visual pulsation may have favored such multi-component entrainment, making these responses quite different from typical motion-evoked cortical responses. Strictly speaking, our RDK stimuli were not even motion stimuli but speed stimuli (no luminance or directional information), and rhythm/pulsation was the main distinguishing phenomenal characteristic of both RDK and naturalistic stimuli used in the present study. For this reason, they are likely to induce high-level resonance that may go beyond sensory stimulation (see, e.g., [18] for a similar consideration in the auditory domain).

Alternatively, our approach might simply not have been fine-grained enough to detect small regional differences. It is possible that with a more in-depth approach (e.g., high-density electroencephalography with event-related component analyses and source reconstruction) and a larger sample size, some differences would emerge.

#### *4.2. Attentive Subliminal Resonance*

We found diffuse cortical responses with the weakest video speed modulation (10%), a condition that was associated with null perceptual sensitivity. This phenomenon calls for subliminal speed entrainment, indicative of an automatic resonance process, which may qualify as a form of attentive subliminal processing [19], or at least at the fringe of awareness. By further noting the disparity between the steep increase in perceptual sensitivity at increasing speed modulation amplitudes and the corresponding very moderate increase in cortical responsiveness, it is tempting to speculate that, at least for certain phenomena, global cortical activation to visual stimuli may reflect automatic processing without necessarily being the signature of perceptual awareness.

However, especially when null results are obtained under low statistical power, it should be borne in mind that "absence of evidence" rather than "evidence of absence" is the proper underlying notion. Furthermore, when the goal is to measure perceptual awareness, null results can be problematic on their own, as it is often difficult to clearly tease apart null from fringe perception [20]. Perceptual ratings may help in this regard, as they involve more gradual responses rather than gross dichotomous responses [21]. Exploiting all possible response criteria behind perceptual ratings through the ROC curve afforded a reliable index of perceptual awareness [6]. Thus, pending further investigation with a larger sample size (see below), we believe that most stimuli with 10% speed modulation were closer to a condition of null than fringe speed modulation perception.
