Neuroimaging workshop February 1st, 2012 Farhan Baluch
Sunday, February 5, 2012

2012 .OUTLINE . February 5.EEG overview -What is the SSVEP? -How is it used to address neuroscience questions? -How to design stimuli to obtain the SSVEP? -What kind of analysis is necessary? -Conclusions Sunday.

E Roy John (1990) Sunday. February 5. 2012 .EEG REPRESENTS THE SUMMED ACTIVITY OF MANY NEURONS Machinery of the Mind .

VISUAL EVOKED POTENTIAL and four slowest individuals on this behavioral mead strongly correlated changes in the rise time of SSVEP Fig.01) and 80.78. The fast switchers reached their peaks of SSVEP ignificantly earlier than the slow switchers. 4). p < . February 5. by 195 ms stimuli and by 269 ms for right-field stimuli (both p correlation of SSVEP peak latency and the time bin of detection performance over the first 600 ms following highly significant for both left (r = 0.01) stimuli. 2012 . p < . Expanded time scale for attend right condition Sunday.

The fast switchers reached their peaks of SSVEP ably throughout amplitude significantly earlier than the slow switchers.LEDs. 4). 1998 ure America Inc.05). February 5. by 195 ms ollowing the cue. rd configuration. ectrode t shown ne) conemporal for the he four ch row. The four STEADY STATE VISUAL EVOKED POTENTIAL n for each row. modulation of the stimulation at a smaller time scale results in o LEDs changed entrainment and a SSVEP both sides with a 00 ms (onset to he SSVEPs were ue in the time er the left or the Muller et al. changed s with a nset to Ps were he time t or the four fastest and four slowest individuals on this behavioral meaEDs on the same sure revealed strongly correlated changes in the rise time of SSVEP argets in that row amplitude (Fig. 2012 < . The correlation of SSVEP peak latency and the time bin of p sites (contralat- .nature. for left-field stimuli and by 269 ms for right-field stimuli (both p ced SSVEP attenSunday. • http://neurosci.

mit. 2012 . February Sunday.csail.1 SLIDE INTRO TO FFT time domain signal FFT frequency domain signal http://groups.

01) stimuli.78. The fast switchers reached theirincrements peaks of SSVEP bly throughout amplitude significantly earlier than the slow switchers.1998 Nature America Inc. 1998 e a continuous g. The correlation of SSVEP peak latency and the time bin of asymptotic detection performance over the first 600 ms following the cue was highly significant for both left (r = 0. as determined by the direction of the cue.05). detections during the second 144-ms interval for left-field stimuli and by 269 ms(for stimuli (bothinterval p d SSVEP attenative to the first interval p <right-field .nature. The fast switchers reached their peaks of SSVEP amplitude significantly earlier than the slow switchers. (b) Time course of SSVEP amplitude in the frequency domain obtained from the waveform shown in (a) by a moving-window Fast Fourier Transform (FFT) at the stimulus frequency. by 195 ms for left-field stimuli and by 269 ms for right-field stimuli (both p < .0-second interval of continuous flickering following the cue. successive window steps were 4 ms.01) andlatency during the fourth interval asymptotic detection the first 600 ms(432–576 following ms) onwards.01) and right (r = 0. The analysis windows for the FFT (480 ms in duration for left stimulus. Fromover the fourth interval the cue was highly significant for both left (r = 0. Mainte) connance of central eye fixation was verified by recordings of the elecmporal trooculogram. nature neuroscience • volume 1 no 7 • november 1998 Muller et al. 4). Individual right (r = 0.01) stimuli. 2012 632 . 2. and trials contaminated with eye movements.01) and subjects difquency domain behavioral detection rates remained stable. The of SSVEP peak and the time bin of relative to sites (contralatto correlation the second (p < . The SSVEP amplitudes to the left and right stimuli were measured in the frequency domain by fast-Fourier transform (FFTs) at the respective flicker frequencies.02). MainteFebruary 5. flicker frequenfered considerably. 4). blinks or the e four or other artifacts were rejected from analysis. The midpoints of these intervals are indicated by vertical lines in Fig. VEP amplitudes the third (p performance < . were The SSVEP elicited by the attended flickering row showed a sharp time or the increase in amplitude from baseline starting at around 100–200 ms after the cue. Representative time. Note that the last 500 ms were not analyzed because the moving window reaches the end of the epoch. p < . (a) Averaged time-domain waveform following the cue to attend right and time-locked to the right flickering stimulus. however. 3a) and for the right stimulus (Fig. • http://neurosci. As in our previous study18. during the third relative < .Sunday. by pressing a button.78. 1). Bold tracing is attended tion effects were found at occipito-temporal scalp sites (contralateral to the visual field of stimulation. the most pronounced SSVEP attene America Inc. Target-detection rates also showed a steep increase as a funcfour fastest and four slowest individuals on this behavioral meaDs on the same tionstrongly of timecorrelated after the cue (Fig.80. 1998 for the remainder of the trial.01). The SSVEP to the attended stimulus remained significantly greater than to the unattended stimulus Muller et al. the SSVEP curves are superimposed for the attended and set to unattended conditions. This increase became significantly greater than baseline during the interval 240–288 ms after the cue for the right row and 288–336 ms for the left row (both p < 0. SSVEP activity to this attended flicker can be seen at the expanded time scale. 3.80. • http://neurosci. 3. The time course of SSVEP amplitude changes following the cue ration.nature four fastest and four slowest individuals on SSVEP this behavioral measubject to shift attention toENHANCES the flickering row of LEDs on the same ATTENTION THE POWER OF THE sure revealed strongly correlated changes in the rise time of SSVEP side and to report occurrences of color-change targets in that row amplitude (Fig.the bar graphs).and frequency-domain waveforms from a single subject.05). p < . to shift attention was averaged across all nine subjects for the left anged stimulus ( Fig. Fig. For each with a stimulus. 2a) to produce a continuous hown function of SSVEP amplitude change over time (Fig. in how rapidly their detection perforduration for left mance reached asymptotic levels. p < . 2b). 360 ms for right stimulus) were moved progressively along trode the averaged SSVEP waveform ( Fig. thin tracing shows unattended waveform elicited by the same stimulus. both the left and sure revealed changes in rise time ofFor SSVEP gets in that row right there were significant in correct target amplitude (Fig. byfollowing 195 ms the cue relowing the cue. the SSVEP to the same stimuli when unattended showed no significant deviations from baseline. Dividing the subjects into the gressively along Expanded time scale for attend right condition Expanded time scale for attend right condition a b Cue Fig.stimuli. Targets occurred unpredictably throughout the 3. Thin horizontal line is drawn through pre-cue baseline. 3b). row. In contrast.05). p< . 2b).

2008 Sunday.SSVEP FOR FEATURE TAGGING Anderesen et al. 2012 . February 5.

Artifact-free epochs were averaged separately for the 18 combinations of hemifield and stimulus pairs to obtain ssVEPs containing both driving frequencies. subjects viewed the 72 different pictures again in a randomized order and were asked to rate the respective picture on the dimensions of affective valence and arousal on the 9-point Self-Assessment Manikin (SAM. the mean rejection rate across all conditions was 36%. Epochs of 600 msec prestimulus and 3600 msec poststimulus onset tion of statistical parameters of the EEG epochs extracted (absolute value. Due to the long epochs and these stringent rejection criteria. angry– happy. and on-line band-pass filtered between 0. referenced to Cz. Off-line. in which the fixation cross was presented against a gray background. digitized at a rate of 250 Hz. the Fast Fourier Transformation on this ssVEP. This resulted in a total of 216 trials (24 trials × 9 conditions). ion ent. For interpolation and all subsequent analyses. a low-pass filter of 40 Hz was applied. data were arithmetically transformed to the average reference. Bradley & Lang.dure alidation study (Goeleven. Note that scales used oth groups are different. 2011 . resulting in 24 trials per condition. neither hemifield presentation nor flickering was used. happy– happy. neutral–happy. Such strict rejection criteria also allowed us to exclude trials contaminated by vertical and horizontal eye movements. Wieser et al. OR. The ssVEP amplitude for each condition was extracted by means of complex demodulation. All expressions were combined and were fully crossed over visual hemifields. happy–neutral. wn separately for high ally anxious (HSA) and socially anxious (LSA) cipants. re 4. neutral. angry–neutral. maximum of the differences) across time points. resulting in nine conditions (angry–angry.1 and 50 Hz. for each channel. participants viewed each picture for 6 sec before the SAM scale was presented on the screen for rating. in which the two pictures were always taken from the same actor. and vice versa. which extracts a modulating signal from a carrier signal (Regan. After the EEG recordings. The number of remaining trials did not differ between experimental conditions and groups. collapsed over the e competitor conditions both hemifields. The mean scalp topographies of both frequencies show clear medial posterior activity over visual cortical areas. neutral–angry. happy–angry. The raw ssVEP for a representative electrode (POz).5 Hz). as recommended for the Electrical Geodesics high-impedance amplifiers. spherical spline interpolated values. USA). had durations between 2500 and 3500 msec. neutral–neutral). 2008). February and neutral 5. Grand ns are averaged across a window between 100 3000 msec after stimulus t. the ssVEP by motor potentials and transient responses to the task stimulus. standard deviation. 2012 Volume 23. EEG Recording and Data Analysis The EEG was continuously recorded from 129 electrodes using an Electrical Geodesics System (EGI. nd wo e ely. In one half of the trials of each condition. and—in a subsequent step—across channels. happy facial expressions. Grand mean steady-state visual evoked potential (ssVEP) averaged across all participants and conditions. 1989) using in-house written MATLAB scripts (for a more detailed description. Mean time course of P amplitudes elicited by y. recorded from an occipital electrode (approximately corresponding to POz of the extended 10–20 system). showing faces nd neutral expressions in 12 female SSVEP COG NEURO EXAMPLE Figure 2. Electrode impedances were kept below 50 kΩ. 72 pictures were selected irected emotional faces (KDEF) datakt. and the spatial topography of the two driving frequencies averaged across all subjects and conditions are shown in Figure 2. Number 8 Twenty-four stimulus pairs were created per condition. The maximum number of approximated channels in a given trial was set to 20. 1998). De Raedt. Eugene. as shown by the frequency domain representation of the same signal (Fast Fourier Transformation of the ssVEP in a time segment between 100 and 3000 msec) in the upper panel. 1994). the left picture was presented at 14 Hz. The order of trials was pseudorandomized. Hz. happy. Note that the ssVEP in the present study contains a superposition of two driving frequencies (14 and 17. In this last block. re. (A) Grand mean gnitive Neuroscience graphical distribution e ssVEP amplitudes in onse to angry. & Öhman.Sunday. Sensors contaminated with artifacts were replaced by statistically weighted. Intertrial intervals.

01s 99.01. 2012 .2Hz 7 6 5 4 3 0..6/7 =14.IMPLEMENTATION-STIMULUS refresh rate -stimulus frequencies must be chosen as a multiple of the monitor refresh frequency 99.6/4 =24.6 = 0. February 5.9Hz or 1/(0.9hz 2 1 99.01s Sunday.6Hz frame interval 1/99.*4)=24.

now = getSecs(). 2012 . currentInterval = 0. end Sunday.window). while(now < start + displayTime. window). loop to display flicker displayTime = 3. isi = 1/f. end vbl = Screen(‘Flip’.window). getting inter-stimulus interval f = 14. prevVbl = vbl.2. start = getSecs(). currentInterval = currentInterval + round((vbl-prevVbl)/ifi). prevVbl = Screen(‘Flip’. February 5.STIMULUS-PSYCHTOOLBOX FUNCTIONS getting inter-frame interval of monitor ifi = Screen('GetFlipInterval'.window.texture).) if(currentInterval >= isi) Screen(‘DrawTextures’.

4 0.7 ANALYSIS time domain signal averaging single frequency presentation in all quadrants 11.3 0.45Hz 8. 2012 .5579 0.07Hz 14.3 Y: 0.a o e e e 0. February 5.1 0 Sunday.3Hz 9.96Hz matlab FFT 0.2Hz 12.5 power uV2 0.6 X: 8.2 frequency domain signal 5 10 15 frequency/Hz 20 25 0.

45Hz 8. February 5.3 0.2Hz 12. 2012 .3Hz 9.5579 0.07Hz 14.96Hz 0.2 0.1 0 5 10 15 frequency/Hz 20 25 Sunday.6 X: 8.SEPARATE FREQUENCY RESPONSES single frequency presentation in all quadrants 0.5 power uV2 0.4 0.3 Y: 0.7 11.

2012 .CONCLUSIONS • SSVEP is a powerful technique specially useful for probing attention related aspects • provides high temporal resolution signals for analysis • analysis is fairly simple • reliable and robust response if you get the stimulus and setup right. February 5. Sunday.

Sign up to vote on this title
UsefulNot useful