**1. Introduction**

A crucial set of experiments by Michael Posner and colleagues [1] buttressed the theory that minds are equipped with a covert visual orienting system to enhance detection of targets even as they lie out of the foveal center of the participant's overt visual attention. This is the notorious spotlight that had been postulated by James in 1890 [2], and evidenced by Helmoltz [3]—see also the reviews and controversies [4–8] and overviews of neuroanatomical bases [9–11] that lent credit to a view of multiple independent networks and processes for attention [9].

In a separate stream of research founded by Hans Berger in the 1920s [12,13], a rhythmic brain wave at about 10 Hz was shown to transpire from the scalp of human participants and reacted to such events as eye opening, involuntary attention to sudden startle from gunshot sound, other auditory, visual, olfactive, tactile, and pain stimuli, voluntary concentration, anesthesia, medications, and a variety of clinical conditions [12–17]. This brain wave is a dominant activity in human waking electroencephalogram (EEG) [18]. After the controversy of its cerebral origin was finally settled [15], what came to be known as the alpha wave [14] rose as one of the most studied electrophysiological phenomenon, with firmly established correlation to attentive processes [19–22].

Here, we restrict the name alpha to the 10 Hz phenomenon peaking in parieto-occipital regions when eyes are closed or attention and vigilance reduced. Alpha's anticorrelation with attentive behavior was noted from the start [12–14] and continued to be observed [23] after warning signals of target occurrence [24] and in relation with many tasks derived from Posner's cueing paradigm, consensually with a contralateral organization, that is, increase power opposite the stimulus side or decrease power ipsilateral to it [25–38] after target onset as well as during the cue period [26,39] and

even in the absence of ultimate target [40]. Alpha is also modulated by temporal expectations [33] and abides to reverse causal inference that finds more omissions when background alpha is intrinsically larger [28,41], externally entrained [42] or perturbed [11].

The functional complexity of 10 Hz oscillations in posterior regions has also been noted [43,44]. First, a so-called paradoxical alpha response (increasing alpha during some exemplars of attentive behavior in violation of the otherwise strong record of alpha suppression during attention, see above) has led to a fervent debate [19,45–48] that appears to have been resolved with a distinction between endogenous and exogenous attention [21,46]: Internally-generated attentional processes are purported to increase alpha to actively inhibit sensory information. This hypothesis retained the idea of a unitary alpha that equally suppresses exogenous and endogenous distractors, but firm evidence of complete anatomical and dynamical equivalence remains unfulfilled. Second, studies of oscillatory power that paid close attention to the timing of oscillatory processes have suggested a complex temporal organization of posterior 10 Hz activities with di fferent sub-bands of alpha playing a role at di fferent moments [18,24,49–51], see also [31]. These studies beg for finer-grained models for the relation between local rhythmic activity and attentional processes. Finally, several studies of functional connectivity also sugges<sup>t</sup> a complex spatiotemporal organization [11,27,52–54].

In the following, I introduce a case study from a small-scale pilot experiment that asked subjects to sustain the dissociation between fixation and covert attention ("looking from the corner of the eyes"). During this task, a 10 Hz neural activity was uncovered that is clearly distinct from alpha (Figure 1b), and I used its existence to develop the hypothesis on a multifaceted model of 10 Hz rhythms' contribution to attentional processes. I based this proposed model on a discrete view of neural oscillations that are spatially specific (e.g., [55,56] and Figure 1a), appearing intermittently over time [57]. This view was gained from a large improvement in spectral and spatial resolution of EEG analysis, and I outline the methodological requirements that might allow to further study the interplay of 10 Hz rhythms, alpha, and others, in attentional processes.

**Figure 1.** The 10 Hz frequency band carries a number of regionally and functionally specific neuromarkers: (**a**) An overview of their peak scalp locations is provided over a colorimetrically-encoded electrode map (recalling that the inverse problem prevents direct cortical localization); (**b**) xi is found in the upper 10 Hz band, with maxima at left centroparietal scalp locations as shown by the green color of its peak inherited from the colorimetric mapping; **(c)** an exemplar pair of Rolandic mu (here denoted mu central left and right) illustrates the limited spatial overlap that left mu has with xi, though both share the upper 10 Hz band; **(d)** an exemplar of alpha rhythm shows discrepancy both in spatial and spectral organization, with alpha having a slower peak frequency (population mode robustly at 10 Hz [57,58]) and a spatial distance to xi of several centimeters, as manifested with the distinct color (inherited from spatial location and not randomly assigned, see (**a**) for legend). Spectra from (**b**), (**c**), and (**d**) are sampled from di fferent subjects and tasks.

#### **2. Materials and Methods**

A subject participated in multiple sessions of an EEG recording (protocol approved by Florida Atlantic University's institutional review board; and written informed consent obtained prior to the experiment). The subject was a right-handed male, healthy young adult (early 20s), with no history of neurological disease, and with normal vision and audition. The sessions were aimed at developing a brain–computer interface [59] and consisted of various ideomotor activities such as executing or imagining movements of the mouth, knees, feet, hands, and fingers by self or other, and lifting of small objects. To control for the potential confound of spatially shifted attention, especially in the case of imagined and executed lower limb movements, a small task was added to the protocol asking the subject to preserve fixation on a crosshair in the center of the computer screen while looking from the corner of the eyes at an unmarked location at the top or bottom of the screen (alternations of 15 s epochs looking from the corner of the eyes and 5 s epochs releasing attention to the central fixation, cumulative duration, 120 s for each direction). This task was created under the rationale that subjects would comply to fixation (validation with EOG) ye<sup>t</sup> might covertly orient their attention to the spatial locus where the targeted body part lay. Since there were no explicit state variables to detect and control for such covert occurrence, we reasoned that sample neuromarkers for this covert activity were important to collect and characterize spatiotemporally. The experiments were conducted in a sound-proof electromagnetically shielded chamber. EEG was recorded using a 60-channel EEG cap with Ag–AgCl electrodes (Falk Minow Services, Herrsching, Germany) arranged according to the 10 percent system [60] (including midline and rows 1 to 8). Electrodes were laid on standard elastic caps whose positioning emphasized the accuracy of vertex electrodes (midway between nasion and inion) and the adequacy of the midline to improve interpretations of lateral symmetries. Electrode impedances were maintained below 10 k Ω, and special attention was paid to the reference electrodes, a pair of digitally linked mastoids (subjected to removal of lipidic film with alcohol swab, double abrasion with hair brush and then gel nuprep, careful adhesive, and compressive securing of the electrodes with tape and elastic cap), leading to their impedance to be low and matched [61]. The ground electrode was located at electrode FPz. The signals were fed to an amplifier (Synamp2, Neuroscan, Texas). The signals were analog-filtered (Butterworth, bandpass from 0.05 Hz (−12 dB/octave) to 100 Hz (−24 dB/octave-)), amplified (gain of 2010), and digitized at 1000 Hz with a 24-bit ADC in the range 0–900 microV (vertical resolution of 0.11 nV). Electro-oculographic (EOG) traces were obtained from two pairs of electrodes placed above and below the right eye (vertical EOG) and on the canthus of each eye (horizontal EOG) to ascertain compliance with instructions not to move the eyes during the tasks.

Three EEG analysis strategies are succinctly presented below, which have been described elsewhere. Multielectrode spectra are obtained via the fast Fourier transform on epochs prepared for enhanced spectral resolution, that is, with the sampling of a longer time interval (e.g., 8.192 or 16.384 s at sampling rate = 1 k Hz) that provides a bin size of 0.06 to 0.12 Hz, much more detailed than a usual 1 Hz resolution and therefore amenable to detecting small discrepancies in frequency of the brain's many 10 Hz activities (Figure 1, see also [55]; strategies to achieve same resolution with smaller epochs can be found in [56]). The spatiotemporal analysis of band-passed filtered EEG uses gently tuned Butterworth filters chosen for their flat passband and applied in both time-positive and time-negative directions to prevent phase distortion [56]. Envelopes are used to scrutinize instantaneous power changes. They were obtained after mean-removed bandpass-filtered signals were rectified and smoothed with a moving average of 100 milliseconds.
