Alpha oscillations reflect suppression of distractors with increased perceptual load

Abstract

Attention serves an essential role in cognition and behavior allowing us to focus on behaviorally-relevant objects while ignoring distraction. Perceptual load theory states that attentional resources are allocated according to the requirements of the task, i.e., its 'load'. The theory predicts that the resources left to process irrelevant, possibly distracting stimuli, are reduced when the perceptual load is high. However, it remains unclear how this allocation of attentional resources specifically relates to neural excitability and suppression mechanisms. In this magnetoencephalography (MEG) study, we show that brain oscillations in the alpha band (8-13 Hz) implemented the suppression of distracting objects when the perceptual load was high. In parallel, high load increased the neuronal excitability for target objects, as reflected by rapid invisible frequency tagging. We suggest that the allocation of resources in tasks with high perceptual load is implemented by a gain increase for targets, complemented by distractor suppression reflected by alpha-band oscillations closing the 'gate' for interference.

Keywords: Alpha oscillations; Attention; Inhibition; MEG; Perceptual load.

Figure 1. Overview of the cued discrimination task.

A) After fixation, subjects were presented with two face stimuli. A directional cue indicated the target stimulus. After a variable delay, a small eye movement occurred in the face stimuli. Subjects indicated the direction of the eye movement by button press. B) The luminance of the white portions of the stimuli flickered at 63 and 70Hz (rapid frequency tagging; RFT), from stimulus onset until the eye movement change of the stimuli. C) The cued target stimuli were masked with noise or not (respectively high versus low perceptual target load). Likewise, the uncued distractors were masked with noise or not (respectively low versus high salient distractors). Noise levels are increased for illustration purposes and stimulus sizes are not to scale.

Figure 2. Behavioural results (reaction times to targets) of the discrimination task.

(A) When considered per condition, participants responded slower to high (red) compared to lower (blue) target loads. (B) Distractor interference (salient compared to noisy distractors) was significant when the target load was low (M_low = 15.3ms, M_high = -4.07ms, paired samples t-test t = 3.6, p < .001), but not when the target load was high (p > .25). These findings are consistent with perceptual load theory.

Figure 3. Attentional modulation and sensor selection.

A) Alpha band (8-13Hz) modulations were quantified by considering the left versus right directional cues. The modulations in time-frequency representation of power showed a sustained alpha power increase ipsi-laterally to the cued hemifield (left) and a power decrease contralaterally (right). The topographic map demonstrated a power modulation over parieto-occipital areas; the marked location indicates the sensors used in the subsequent analyses. B) Time-courses of the alpha band power relative to pre-cue baseline contralateral to target (green) and distractor (black). We show this with respect to Cue onset (left) and discrimination target onset (right). C) Rapid Frequency Tagging (RFT) power modulations. The time-frequency representations show an increase in power contralateral to the target and a decrease contralateral to the distractor. The target was flickering at 63Hz (left) and 70Hz (right). The topographic plot (centre) shows these modulations to be over central occipital areas. The marked location indicates the sensors used in the subsequent analyses. D) Time-courses of the RFT power relative to pre-cue baseline for target frequencies (green) and distractor frequencies (black) with respect to Cue onset (left) and discrimination target onset (right).

Figure 4. Effects of target load and distractor salience in the alpha band.

A) Average alpha band power per condition, contralateral to target (left) and distractor (right) for high (red) and low (blue) target load conditions with noisy (dashed) or salient (solid) distractors 500 – 1350 ms after cue onset. Error bars denote the standard error of the mean (SEM). B) The effect of target load (high-low target load) for salient and noisy distractors. The mean is denoted by the horizontal line, the median is denoted by the white dot and interquartile range by the grey vertical bar. C) The effect of distractor salience (high-low salience) with high and low target load. D) Distractor-alpha power time courses for high (red) and low (blue) target load locked to cue onset (left) and discrimination target onset (right). Importantly the distractor-alpha was larger for high compared to low target-loads, consistent with the hypothesis that alpha power implements the distractor suppression according to perceptual load theory. E) Source modeling of the distractor-alpha modulation of high versus low target loads (cue-locked, 5 – 1.35 s) collapsed over left and right attention displayed as attention to the left (left hemisphere is contralateral to distractor). Plot shows top and bottom 1% values in the grid-points. F) Correlation between average load modulation index (distractor alpha (high load - low load) / high load + low load), cue-locked, 500 – 1350ms after cue onset) of distractor alpha and behavioural distractor interference (RT_noisy – RT_salient) under low target load. This shows that the larger the difference in distractor alpha power (i.e. lower alpha power with low load), the larger the behavioural distractor interference. G) Distractor alpha power time courses for salient and noisy distractors relative to cue onset (left) and discrimination target onset (right) under high target load.

Figure 5. Load effects in RFT power.

A) Target RFT power per condition. Average RFT power related to the frequencies of the target for high (red) and low (blue) target load conditions with noisy (dashed) or salient (solid) distractors 500 – 1350ms after cue onset. Error bars denote standard error of the mean (SEM). B) Highlight of the significant effect of target load in target RFT power. The effect of target load (high-low target load) for salient and noisy distractors.The mean is denoted by the horizontal line, the median is denoted by the white dot and interquartile range by the grey vertical bar. C) Target RFT power time courses for high (red) and low (blue) target load relative to cue onset (left) and discrimination target onset (right). Importantly the Target RFT power was elevated for high compared to low target-loads which is consistent with an increase in neuronal excitability in the target with an increased perceptual load. D) Source modelling of the relative baseline target-RFT modulation of high versus low target loads (cue-locked, 500 - 1350ms, shown in B) collapsed over left and right attention, displayed as attention to the left (right hemi sphere is contralateral to target). Plot shows top and bottom 1% of the values in the grid-points. E) Correlations between average distractor effect (Target RFT_salient - Target RFT_noisy)/(Target RFT_salient+Target RFT_noisy) cue-locked, (500 – 1350 ms after cue onset) and behavioural distractor interference (RT_noisy – RT_salient) of target RFT with respect to low (left) and high (right) target load.

Figure 6. Effects in distractor RFT.

A) Average distractor RFT power related to the target for high (red) and low (blue) target load conditions with noisy (dashed) or salient (solid) distractors 500 – 1350ms after cue onset. Error bars denote standard error of the mean (SEM). B) Effect of target load (high-low target load) for salient and noisy distractors.The mean is denoted by the horizontal line, the median is denoted by the white dot and interquartile range by the grey vertical bar. C) The effect of distractor salience (high-low salience) for high and low target load. D) Source modelling of the salient minus noisy distractor difference under high load 0.5-1.1s after cue onset. Cooler colours denote the decrease in RFT power, while warmer colours show the alpha power increase. Contrasts are collapsed over left and right attention and displayed as attention to the left (left hemisphere is contralateral to distractor). Plot shows top and bottom 1% of the values in the gridpoints. E) Noisy distractors-RFT power time courses with high and low target load relative to cue onset (left) and discrimination target onset (right) under high target load. F) Distractor RFT power time courses for salient and noisy distractors relative to cue onset (left) and discrimination target onset (right) under high target load.