Gamma and beta neural activity evoked during a sensory gating paradigm: effects of auditory, somatosensory and cross-modal stimulation

Abstract

Objective: Stimulus-driven salience is determined involuntarily, and by the physical properties of a stimulus. It has recently been theorized that neural coding of this variable involves oscillatory activity within cortical neuron populations at beta frequencies. This was tested here through experimental manipulation of inter-stimulus interval (ISI).

Methods: Non-invasive neurophysiological measures of event-related gamma (30-50 Hz) and beta (12-20 Hz) activity were estimated from scalp-recorded evoked potentials. Stimuli were presented in a standard "paired-stimulus" sensory gating paradigm, where the S1 (conditioning) stimulus was conceptualized as long-ISI, or "high salience", and the S2 (test) stimulus as short-ISI, or "low salience". Three separate studies were conducted: auditory stimuli only (N = 20 participants), somatosensory stimuli only (N = 20), and a cross-modal study for which auditory and somatosensory stimuli were mixed (N = 40).

Results: Early (20-150 ms) stimulus-evoked beta activity was more sensitive to ISI than temporally-overlapping gamma-band activity, and this effect was seen in both auditory and somatosensory studies. In the cross-modal study, beta activity was significantly modulated by the similarity (or dissimilarity) of stimuli separated by a short ISI (0.5 s); a significant cross-modal gating effect was nevertheless detected.

Conclusions: With regard to the early sensory-evoked response recorded from the scalp, the interval between identical stimuli especially modulates beta oscillatory activity.

Significance: This is consistent with developing theories regarding the different roles of temporally-overlapping oscillatory activity within cortical neuron populations at gamma and beta frequencies, particularly the claim that the latter is related to stimulus-driven salience.

Fig. 1

Example waveforms used for analysis of early, stimulus-locked activity. The broadband response, at left, is an average evoked potential (electrode Cz referenced to nose) elicited by a long-ISI auditory stimulus for one individual. This waveform was transformed through complex demodulation and band-pass filtering into two separate frequency bands, producing the time-varying estimates of stimulus-locked activity within gamma (30-50 Hz) and beta (12-20 Hz) bands shown at right. Note that these functions represent the absolute value of response amplitude, whereas a band-passed ERP waveform would exhibit both positive and negative deflections. The response was quantified for each frequency band as the latency and amplitude of the peak response between 20 and 150 ms post-stimulus.

Fig. 2

Auditory-only condition: Inter-stimulus interval (ISI) between auditory clicks modulated peak response amplitude. The grand-average (N=20) evoked amplitude for both long (9 s) and short (0.5 s) ISI stimuli is shown separately for gamma (top left) and beta (top right) bands. Note especially the strong response reduction in the beta band when ISI was shortened. The beta response peaked significantly later than the gamma response (bottom left), and the sensitivity of the beta response to ISI was greater than the sensitivity of the gamma response (bottom right). Symbols represent mean values; vertical lines represent standard errors.

Fig. 3

Somatosensory-only condition: Inter-stimulus interval (ISI) modulated peak response amplitude. In the grand-average waveforms (N=20) note the strong response reduction in the beta band when ISI was shortened (top right). The sensitivity of the beta response to ISI was greater than the sensitivity of the gamma response (bottom right). Also note the expected latency differences between the responses in the separate frequency bands, where the gamma response peaks before the beta response regardless of ISI (bottom left).

Fig. 4

Combined auditory and somatosensory condition: Only the beta band response elicited by the short-ISI stimulus was sensitive to the modality of the preceding stimulus. Shown are the mean response amplitudes (+/-standard errors) to the short-ISI stimuli in either the auditory or somatosensory domains. These responses are sorted according to whether the preceding stimulus was auditory or somatosensory. For example, at left in the BETA plot (bottom), it can be seen that the response to an auditory stimulus following an auditory stimulus was significantly smaller than the response to a somatosensory stimulus following an auditory stimulus. Following the diagonal line from bottom left to top right in this same BETA plot, the auditory-following-somatosensory response was significantly larger than the auditory-following-auditory response. The same general pattern applies to responses to the somatosensory stimuli presented at short-ISI. A similar separation of response amplitudes was not detectable for the gamma band response (top).

Fig. 5

The beta response exhibited a cross-modal sensory gating effect. Grand-averaged (N=40) peak beta activity in response to a short ISI cross-modal stimulation was larger than short-ISI intra-modal stimulation, but smaller than long-ISI stimulation for both auditory (top) and somatosensory (bottom) stimuli. Thus, although the beta response dishabituated when the stimulus modality changed, it did not recover to the original long-ISI response amplitude.

Fig. 6

Grand average stimulus-induced (non phase-locked) activity for the auditory-only condition. Based on previous studies (Haenschel et al., 2000; Tallon-Baudry and Bertrand, 1999), the planned analysis window for this activity was between 200 and 400 ms. However, as illustrated here, stimulus-induced gamma activity in that latency range was not substantially different than random noise, and stimulus-induced beta activity showed an apparent suppression (i.e., reduction below pre-stimulus levels). The earlier (20-150 ms) stimulus-induced beta activity temporally overlapped the stimulus-evoked beta activity described above, and amplitude of this relatively small response was related to ISI. Note that the vertical-axis scale was preserved from previous waveform plots to allow comparison between the relative amplitude of stimulus-evoked and stimulus-induced oscillations.

Fig. 7

Stimulus-locked gamma (30-50 Hz) and beta 1 (12-20 Hz) activity contribute to ERP components commonly used to study cortical processing of stimulus-driven salience. Some of these components are indicated for grand averaged waveforms computed from the 20 individuals used for the auditory-only study presented here (left): Pa (“ P30”), P1 (“ P50” , “Pb”), and N1 (“ N100”). Illustrated are the responses to the long (top) and short-ISI (bottom) stimuli. In the sensory gating literature this corresponds to S1 (“ conditioning”) and S2 (“ test”) responses, respectively. Bandpass filtering of these waveforms separately for gamma and beta bands (both filters with 48 dB/oct roll-offs on both corners; filters applied forward and reverse to avoid distortion of phase) demonstrates that component Pa primarily reflects the early gamma response, P1 receives strong contributions from both frequency bands, and of these frequency bands only the beta band contributes to N1 amplitude. The majority of amplitude reduction for component P1 from long- to short-ISI stimulation (compare top left to bottom left) can be seen to result from reduced stimulus-evoked beta activity (compare top right to bottom right). The reduction in the beta band also contributes to reduced N1 amplitude evoked by the short-ISI stimulus. Note difference in scale between the broadband and bandpassed waveforms. The bandpass filtered waveforms shown here are for the purpose of comparison with previous literature; this transformation is non-identical to the complex demodulation analysis employed for the purpose of the present study.