Oscillatory Mechanisms of Stimulus Processing and Selection in the Visual and Auditory Systems: State-of-the-Art, Speculations and Suggestions

Abstract

All sensory systems need to continuously prioritize and select incoming stimuli in order to avoid overflow or interference, and provide a structure to the brain's input. However, the characteristics of this input differ across sensory systems; therefore, and as a direct consequence, each sensory system might have developed specialized strategies to cope with the continuous stream of incoming information. Neural oscillations are intimately connected with this selection process, as they can be used by the brain to rhythmically amplify or attenuate input and therefore represent an optimal tool for stimulus selection. In this paper, we focus on oscillatory processes for stimulus selection in the visual and auditory systems. We point out both commonalities and differences between the two systems and develop several hypotheses, inspired by recently published findings: (1) The rhythmic component in its input is crucial for the auditory, but not for the visual system. The alignment between oscillatory phase and rhythmic input (phase entrainment) is therefore an integral part of stimulus selection in the auditory system whereas the visual system merely adjusts its phase to upcoming events, without the need for any rhythmic component. (2) When input is unpredictable, the visual system can maintain its oscillatory sampling, whereas the auditory system switches to a different, potentially internally oriented, "mode" of processing that might be characterized by alpha oscillations. (3) Visual alpha can be divided into a faster occipital alpha (10 Hz) and a slower frontal alpha (7 Hz) that critically depends on attention.

Keywords: alpha; attention; entrainment; oscillation; perception.

Figure 1

Overview of the role of neural oscillations for stimulus selection and processing in vision. (A) Difference in EEG power (color-coded) around target onset between subjects that did not perceive near-threshold visual targets and those that did (reproduced with permission from Hanslmayr et al., 2007). Results indicate that visual detection depends on alpha power, with lower power leading to an improved detection. (B) Detection of a weak visual target also depends on the phase of the alpha band, as measured in the EEG (reproduced with permission from VanRullen et al., , the original data is presented in Busch et al., 2009). The strength of modulation of target detection by the EEG phase in the respective frequency band is color-coded; the significance threshold is marked on the color bar. (C) When a random luminance sequence is presented to human subjects and their EEG is recorded in parallel, a reverberation (“perceptual echo”) of this visual information can be found in the electrophysiological signal for up to 1 s (using cross-correlation between luminance sequence and EEG), but only in the alpha band (reproduced with permission from VanRullen et al., , the original data is presented in VanRullen and Macdonald, 2012). (D) After a visual stimulus cues attention to one visual hemifield, the probability of detecting a succeeding target fluctuates rhythmically, and in counterphase depending on whether the target occurred in the same or opposite hemifield (left; reproduced with permission from Landau and Fries, 2012). This “visual rhythm” fluctuates at 4 Hz per visual hemifield (right), indicating an overall sampling rhythm of 8 Hz, thus lying within the alpha band. Note that some effects (A,C) seem to have a somewhat higher frequency than others (B,D), leading to the distinction between an “occipital alpha” (~10 Hz) and a “frontal alpha” (~7-8 Hz) in this paper (following VanRullen, 2016).

Figure 2

Overview of the role of neural oscillations for stimulus selection and processing in audition. (A) Detection of a near-threshold target is independent of the EEG phase when presented in quiet (reproduced with permission from VanRullen et al., ; the color-code corresponds to that in Figure 1B). (B) It is a widespread phenomenon that oscillations entrain to rhythmic auditory stimulation. Shown is the data from a study in which a train of pure tones, with a repetition rate of 0.5 Hz, has been presented to human subjects, and the EEG was recorded in parallel (reproduced with permission from Zoefel and Heil, 2013). The amplitude of the tones was set to a near-threshold level and subjects had to press a button whenever a tone was detected; the plot shows EEG data, averaged across subjects, in response to three subsequently missed targets (denoted “S”). An oscillatory signal, entrained to the rhythmic stimulation, is apparent—as subjects did not consciously perceive the stimulation, a potential contamination by evoked potentials introduced by the stimulation is minimized. (C) The auditory system seems to be able to switch between a “rhythmic mode.” in which processing is determined by oscillations corresponding to the input rate of the entraining stimulus, and an “alpha mode,” in which alpha oscillations dominate the processing. During rhythmic stimulation, large fluctuations in the amount of phase entrainment (indicated by the amount of phase-locking in moving time windows of 5 s, shown in red) and alpha power (blue) exist (reproduced with permission from Lakatos et al., 2016). Importantly, periods of pronounced entrainment and of high alpha power alternate, suggested by a phase opposition between the two functions. This finding was interpreted as alternating periods of external and internal attention. In this paper, we hypothesize that processing in the “alpha mode” might be generalized to input in which no regular structure can be detected, and this speculation requires further experiments (cf. Box 2). ITC, inter-trial coherence.