Role of Inhibitory alpha
oscillations during visual
attention
By:
Benjamin Germain
University of Amsterdam Academisch Medisch Centrum10-30-2014
Supervisor: Ali Mazaheri
Co-assessor: Heleen Slagter
Abstract
How the human brain filters out less important visual information is still not fully understood. Since it has been shown that we take in more information then can be analyzed, vision is proposed to rely on attention. Alpha oscillations of 8-15Hz has demonstrated inhibitory properties. Therefore, this experiment has been designed to investigate if the alpha oscillation contributes to inhibition seen during visual attention. An EEG based visual experiment was designed for this test. The computer based test involved participants identifying if two red target letters are identical or not. The spatial distance between these two targets was the variable of interest. Analysis resulted in an increase in the alpha amplitude for the more distant conditions and stronger ERP response to the closer conditions. Concluding that alpha oscillations do play a significant role during visual attention.
Human vision relies on the eyes to convert external, light based information into neurological information that the brain can understand. It is however, under the control of the brain as to how this neurological information is interpreted and perceived in our conscious minds. While the brain is
considered to be a robust information processing powerhouse, it has been shown that the human visual system cannot actively interpret the mass amounts of visual information received by our eyes. Therefore, it is believed that visual processing implements attention driven segmentation of stimuli (Tsotsos et al., 1990; Tsotsos et al., 1995; Itti, Koch, & Niebur, 1998; Lenny et al., 2003; Carrasco, 2011). Visual
attention involves selective facilitation of relevant information and simultaneous suppression of irrelevant information, resulting in a comprehensive inspection of the target stimuli. The visual Selective Tuning (ST) theory states, that we attend to and process relevant information in our visual field, while simultaneously suppressing irrelevant information immediately surrounding the target stimuli, in order to obtain a less obstructed view of the target (Tsotsos et al., 1995; Cutzu & Tsotsos, 2003). This phenomenon has been demonstrated via multimodal studies, including topographical analysis of the visual processing areas resulting in what is known as the “Mexican hat” pattern of activation. This “Mexican hat” pattern demonstrates a central core of strong activation, immediately surrounded by a thin ring of inhibition, followed up by a larger ring of weak-moderate activation (Cutzu & Tsotsos, 2003; Hopf et al., 2006). The precipitations of these patterns have not yet been elucidated as to how they occur, as well as their distinct function in visual processing. Investigation of inhibitory correlates, such as alpha oscillations, may help to demonstrate the complex interplay within our visual system.
Alpha (~ 8-15 Hz) oscillation throughout the brain has been proposed to contribute to attention by gating information flow to relevant sensory regions through inhibition of irrelevant regions (Jensen and Mazaheri, 2010; Klimesch et al., 2007). The alpha-inhibition hypothesis proposes: low alpha activity is indicative of active neuronal processing regions, whereas strong alpha oscillations demonstrate inhibition and disengagement of task-irrelevant cortical areas (Klimesch et al, 2007). Many multi-modal studies support this theory by reporting enhanced alpha range oscillations in brain regions that process unattended information and suppression of alpha wave oscillations in brain regions contributing to attentional processing. (Bauer et al., 2012; Haegens et al., 2012; Jokisch and Jensen, 2007; Medendorp et al., 2007; Rihs et al., 2007; Romei et al., 2008). While the mechanism directing functional inhibition elicited by alpha waves is not fully understood, recent studies demonstrate that alpha oscillations exhibit a strong inhibitory influence on both spike timing and firing rate of neural activity. (Haegens et al., 2011; Mazaheri and Jensen, 2010). Two separate studies by Jensen et al., (2007, 2010) have demonstrated an increase in alpha activity during task-relevant regions and a decrease in alpha activity in task-irrelevant regions. Moreover, a number of studies have established that high pre-stimulus alpha power predicted reduced visual detection performance (van Dijk et al., 2008; Hanslmayr et al., 2007; Babiloni, et al., 2006).
The current EEG study investigates the potential contribution of alpha oscillations on functional inhibition seen during visual processing; based on the ST theory of visual attention. The study attempts to decipher if alpha oscillations functionally inhibit trivial areas of our visual field, in order to achieve a superior focus on salient information. To do so, we composed a visual attention task designed to determine the alpha oscillation fluctuation when observing two related stimuli, based on spatial
separation. In this task, two similar targets are placed among a group of distracters. The participant must identify if these two targets are identical or not. Based on the ST theory, targets directly adjacent to each other should be harder to identify (demonstrate larger alpha synchronization) than targets with 2+ degrees of separation (Cutzu & Tsotsos, 2003). We will be observing the time lock based increase in alpha oscillations in accordance with visual attention stimulation. Elucidation of inhibitory alpha
oscillations results in a more thorough understanding of how the human visual system functions, as well as the faculty of alpha band synchronization.
Methods
Participants
Six healthy, normative, non-color blind adults (3 men, mean age 26 years, range 20-30 years)
participated in the study. All participants were right-handed, had no history of neurological disorder and had normal or corrected to normal vision. Participants with corrected vision were asked to wear glasses, as opposed to contacts, for the duration of the experiment. Before the experiment began, participants each gave written informed consent. The experiment was approved by a local ethical committee at the Academisch Medisch Centrum, Amsterdam, NL. Participants were compensated 10 euros for one hour of their time. Twelve subjects were originally tested, while six subjects were selected for analysis due to the quality of the data. The remaining subjects’ data contained unacceptable artifacts/noise and was
therefore rejected for analysis.
Selective Tuning based stimuli
We developed a visual attention based testing paradigm, implementing spatial separation. Methodology and visual stimulation design was derived from previous ST based studies (Hopf et al., 2008, Boehler et al., 2009, Cutzu & Tsotsos, 2003). A central fixation cross was presented for the duration of the test. Initiation of trials was indicated by the onset of a small grey circle (cue, 200 milliseconds). Participants were informed that the cue identifies the location of one of the two target letters, which they must identify. 50ms after the cue disappears, nine stimuli appeared in an isocentric curve in the lower right visual quadrant for 800ms (see fig.1). All of the stimuli were either a T or L. Seven of the stimuli were black (distracters) and two were red (targets). Participants were instructed to focus on the two red stimuli. One target location was identified by the grey cue circle, the second target must be actively located. The
participant’s goal was to identify, via mouse click, if the two red letters were identical or not, as quickly and accurately as possible (left click = identical, right click= different). The cue identified the location of one of the two red target stimuli, thereby giving the participant a reference to being attenuation.
Participant response or time out (+800ms) began the next trial. The test was divided into 10 blocks, with each block containing 100 trials. Resulting in approximately three minute blocks in which the participants were allowed to rest between. The test was designed to measure time locked variation in brain
oscillations (specifically alpha), in relation to the spatial distance between the two target stimuli.
Figure 1: (a) Testing begins with a central fixation cross in the center of the screen. This cross is constant throughout the testing period. (b) A small grey circle appears for 200ms. This is the cue for one of the two target items. (c) The cue disappears and only the fixation cross is displayed for 50ms. (d) Nine stimuli, being either a T or L, are presented in an isocentric circle in the lower right visual field. Seven of the letters are presented in black and two are presented in red (targets). Participants are to determine if the two red stimuli are identical or not via mouse button click.
Conditions
There are five active conditions that are being analyzed for this test. The conditions are based on the spatial separation between the cue (Target 1) and the second target. Condition 1 is when target 1(T1) and target 2 (T2) are directly adjacent. Condition 2 has one distractor separating T1 and T2. Condition 3 has 2 distractors separating T1 and T2. Condition 4 has 3 distractors separating T1 and T2. Condition 5 has 4 distractors separating T1 and T2.
Data Acquisition
Electroencephalogram (EEG) data was acquired using a WaveGuard 64 channel electrode system. The EEG was sampled at 1024 Hz, with an online average reference and then subsequently imported into MATLAB for pre-processing. After pre-processing, Field Trip was implemented for conditional separation and statistical analysis. The electrooculogram was recorded between supra- and infra-orbital sites around the left eye for vertical movement (blinks), and outer lateral locations of the left and right eyes for possible side-eye movements.
Results
Alpha amplitudes
Alpha amplitudes revealed a significant difference for condition 2. A two-tailed t-test analysis confirms rejection of the null hypothesis for a spatially close condition (Condition 2) compared to the other conditions (1, 3, 4 and 5). Resulting in a significant difference in the amount of recorded alpha amplitude for the stimuli with 1 degree of separation compared to the other conditions (p<
0.0472
). Condition 4 displayed a significance of (p< 0.006) when compared with the other conditions. As well as containing the largest alpha amplitude (9.40), while condition 2 exhibited the lowest alpha amplitude (8.56). Alpha amplitude is seen to increase as spatial distance increases (see figure 2).8 8.2 8.4 8.6 8.8 9 9.2 9.4 9.6 1 2 3 4 5 Am p litu d e Condition
Alpha Amplitude
Figure 2. Average alpha amplitude per condition in the visual PO3 and PO5 electrodes. Amount of alpha amplitude increases at spatially distant conditions 3, 4 and 5. Condition 2 shows
significantly weaker alpha amplitude (p<0.006) and Condition 4 shows significantly stronger alpha amplitude when compared to the other conditions (p<
0.0472
)ERP
ERP’s were investigated at 0.15-0.2ms and 0.22-0.27ms post stimuli onset. A paired two sample t-test for means concluded a rejection of the null hypothesis. Resulting in a significant difference in ERP reaction between condition 1 and conditions 2-5 (p< 0.045), collectively for both time sets. Conditions 1 (2.3134) and 2 (2.2824) elicited the strongest response to the stimuli, while condition 4 elicited the smallest response (0.664) (see fig. 3). Moreover, ERP response of condition 4 is significantly different than the remaining conditions (p<.004).
Figure 3: Collective ERP response amplitude at 0.15-0.2ms and 0.22-0.27ms post stimuli onset per condition. Condition 1 and 2 demonstrate the strongest ERP based response to the stimuli, while the more distant 3, 4 and 5 express weaker ERP response. Condition 4 shows a significantly low amount of response power (p<.004).
Topographical analysis
Topographical maps display areas of alpha oscillation fluctuation for each conditions. All conditions display a relative overall decrease in the alpha oscillation (see figure 4). However, time
frequency response (TFR) charts display an increase in the alpha oscillation immediately following stimuli onset (0-0.2) followed up by a larger decrease (0.2-1.25) (see figure 5). The initial alpha increase could be contributed to the ST model, where alpha oscillations inhibit the adjacent area of a target stimuli. After
0 0.5 1 1.5 2 2.5 1 2 3 4 5 ERP amp litu d e Condtiions
ERP response
the initial unimportant information has been inhibited and the target has been attended to, the alpha waves then being to decrease.
Figure 4: Topographical maps for each conditions with power ranging from -14.5 to 14.5. All conditions display overall decrease in alpha oscillation amplitude over the visual area.
Figure 5: Time frequency response of alpha (8-15Hz) oscillations during condition 2. Time 0 (solid black line) indicates stimuli onset. Alpha increase can be seen immediately following stimuli onset, followed up by a decrease in alpha oscillation around 250ms lasting until about 1250ms. Demonstrating a small increase in alpha amplitude followed by a larger decrease.
Discussion
The current study investigated the role of inhibitory alpha oscillations during a visual attention task. Participants were asked to detect if two target items were identical or not. One of the target locations was indicated to the participant via a cue circle prior to stimuli onset. The second target would then
appear at a specific spatial distance from the cued target. It was hypothesized that the adjacent condition would be the most difficult to detect due to an increase in alpha inhibition, based on the selective tuning model. This study also investigated if the ring of inhibition seen in ST and Mexican hat models can be related to the inhibitory alpha oscillation. Ultimately, probing into the overall role of alpha oscillations in relation to visual attention tasks.
The alpha amplitude analysis concluded unexpected results in terms of the initial hypothesis. The hypothesis stated that the adjacent condition should show high levels of alpha and the more distant conditions should display lower alpha amplitudes. However, the highest alpha levels were seen during the more distant conditions 3-5 with 4 being the highest (figure 2). It appears that alpha amplitudes do
contribute to attention based visual processing. Larger alpha amplitudes for conditions 3, 4 and 5 could indicate a few possibilities. One explanation is that alpha waves contribute to inhibiting more than just the adjacent distractors. When the participant shifts attention from the cued target to a more distant target (2+ degrees of separation) they must now filter out additional distractors in order to focus properly on the target. Therefore, the alpha amplitude increase observed around condition 4 could be contributed to an increase in the amount distractors that needed to be filtered out. The increase in alpha power would contribute to inhibition of the new distractors, resulting in a better view of the target 2. Another explanation is that the target itself at location 4 is being inhibited by the alpha wave. When observing the cue stimuli, the alpha wave inhibits the more distant stimuli (not explicitly the adjacent distractors) in order to observe the cued target properly. This would result in increased alpha amplitude for the further items.
For the ERP analysis, we expected that condition 1 would exhibit a weaker ERP response (Hopf et al., 2006) based on ST theory. The latter states that the adjacent target will show weak ERP response as it is spatially close to the point of focus and thus is being inhibited. However, the results of this experiment do not suggest this same conclusion. The directly adjacent condition 1 did not display inhibited ERPs, but in fact resulted in the strongest response. Condition 4 demonstrated significantly weaker ERP responses (p<.004) when compared to the other conditions. Alpha amplitudes aggrandize with the ERP analysis, resulting in the lowest ERP response for conditions with the strongest alpha oscillation. Demonstrating that the inhibitory alpha wave can affect the response strength during a visual attention task. It should be noted that condition 4 displayed the weakest ERP as well as the strongest alpha amplitude. Additionally, figure 2 (alpha amplitudes) and figure 3 (ERP response) appear to be inverse of each other. Establishing that alpha oscillations can have an inhibitory effect on visually attentive based response power.
TFR topographical maps displayed provocative results as well. While the overall alpha wave appears to be suppressed throughout all conditions, time lock analysis demonstrates a more complex interworking. An increase in alpha can be observed immediately following stimuli onset (figure 5, 0 seconds). However, at about 200ms post stimuli onset, the alpha increase turns into a larger alpha decrease lasting until approximately 1250ms seconds post stimuli onset. This increase in alpha could be
contributed to the finding of the initial target. When the participant first looks for a target (0 seconds) they must focus on the target and ignore the distractors. The initial focusing on the target is accompanied by the alpha based inhibition of the background objects. Once the target has been located and identified (after about .25 seconds) a decrease in alpha wave is seen. The reduction in alpha is seen to end shortly after the trial is completed.
Conclusion
The current EEG study demonstrates that the inhibitory alpha oscillations contribute to information gating during a visual attention based task. This study did not display the typical ERP and alpha amplitude response expected from ST theory. Instead, we observed an increase in alpha amplitude coinciding with decrease in ERP response for the more distant conditions. Since condition 4 contained the highest alpha amplitude as well as the weakest ERP response, we conclude that alpha oscillations have an effect on visual attention. The timing and extent of the alpha oscillations contribution to visual attention is not yet fully elucidated and can be the basis of future research.
References
Bauer, M., Kennett, S., & Driver, J. (2012). Attentional selection of location and modality in vision and touch modulates low-frequency activity in associated sensory cortices. Journal of
neurophysiology, 107(9), 2342-2351.
Carrasco, M. (2011). Visual attention: The past 25 years.
Vision research,51(13), 1484-1525.
Cutzu, F., & Tsotsos, J. K. (2003). The selective tuning model of attention: psychophysical evidence for a
suppressive annulus around an attended item.Vision research,
43(2), 205-219.
Hopf, J. M., Boehler, C. N., Luck, S. J., Tsotsos, J. K., Heinze, H. J., & Schoenfeld, M. A. (2006). Direct
neurophysiological evidence for spatial suppression surrounding the focus of attention in
vision. Proceedings of the National Academy of Sciences of the United States of America, 103(4),
1053-1058.
Haegens, S., Handel, B.F., Jensen, O., 2011. Top–down controlled alpha band activity in somatosensory
areas determines behavioral performance in a discrimination task. J. Neurosci. 31, 5197–5204.
Haegens, S., Luther, L., Jensen, O., 2012. Somatosensory anticipatory alpha activity increases to suppress
distracting input. J. Cogn. Neurosci. 24, 677–685.
Itti, L., Koch, C., & Niebur, E. (1998). A model of saliency-based visual attention for rapid scene
analysis. IEEE Transactions on pattern analysis and machine intelligence, 20(11), 1254-1259.
Jensen, O., Mazaheri, A., 2010. Shaping functional architecture by oscillatory alpha activity: gating by
inhibition. Front. Hum. Neurosci. 4, 186.
Jokisch, D., Jensen, O., 2007. Modulation of gamma and alpha activity during a working memory task
engaging the dorsal or ventral stream. J. Neurosci. 27, 3244–3251.
Klimesch, W., Sauseng, P., Hanslmayr, S., 2007. EEG alpha oscillations: the inhibitiontiming hypothesis.
Brain Res. Rev. 53, 63–88.
Medendorp, W.P., Kramer, G.F., Jensen, O., Oostenveld, R., Schoffelen, J.M., Fries, P., 2007. Oscillatory
activity in human parietal and occipital cortex shows hemispheric lateralization and memory
effects in a delayed double-step saccade task. Cereb. Cortex 17, 2364–2374.
Rihs, T.A., Michel, C.M., Thut, G., 2007. Mechanisms of selective inhibition in visual spatial attention
are indexed by alpha-band EEG synchronization. Eur. J. Neurosci. 25, 603–610.
Romei, V., Brodbeck, V.,Michel, C., Amedi, A., Pascual-Leone, A., Thut, G., 2008a. Spontaneous
fluctuations in posterior alpha-band EEG activity reflect variability in excitability of human
visual areas. Cereb. Cortex 18, 2010–2018.
Thut, G., Theoret,H., Pfennig,A., Ives, J., Kampmann, F., Northoff,G., Pascual-Leone, A., 2003.
Differential effects of low-frequency rTMS at the occipital pole on visual-induced alpha
desynchronization and visual-evoked potentials. Neuroimaging 18, 334–347.
Thut, G., Nietzel, A., Brandt, S.A., Pascual-Leone, A., 2006. Alpha-band electroencephalographic
activity over occipital cortex indexes visuospatial attention bias and predicts visual target
detection. J. Neurosci. 26, 9494–9502
Tsotsos, J. K. (1990). Analyzing vision at the complexity level.
Behavioral and brain sciences, 13(03),
423-445.
Tsotsos, J. K., Culhane, S. M., Kei Wai, W. Y., Lai, Y., Davis, N., & Nuflo, F. (1995). Modeling visual attention via selective tuning. Artificial intelligence,78(1), 507-545.
