We are conducting an SSVEP experiment with stimuli flickering at 3Hz and 10Hz. However, FFT (attached) shows larger peaks at 6Hz and 20Hz. This does not seem to be a result of preprocessing decisions, as this can also be observed on non-preprocessed data.
We used a photodiode to ensure that the flicker frequency is correct. Our stimulus was first a flickering sine-grating of black and white. After we observed this issue, we changed the stimulus to a solid dot alternating black and white (black-to-white-to-black per cycle) but the issue remains.
This phenomenon was recently reported in Lui et al. (2018, IJP) but there was no clear explanation about this phenomenon.
Any idea of what could cause this issue?
Thanks
Hello, I noticed a similar thread that you may find useful: https://www.researchgate.net/post/In_SSVEP_why_sub-harmonic_of_SSVEP_become_the_largest_peaks_in_FFT most likely is a phenomenon determined by the stimulus pattern. If not the case, then I reckon it would be difficult to explain it on the basis of the stimulus information content (very basic compared with a human face)
Short answer - so called "frequency dooubling". This is a well known phenomena of the brain responses to symmetrical periodic stimuli. For details see fig. 3 and corresponding section in:Article The steady-state visual evoked potential in vision research: A review
Actually, I would worry more about the presence of the 1st harmonic in the response spectra. For the perfect symmetrical stimuli there should be none. If you provide more details about your stimuli (full screen screenshots with time diagrams of luminance change), we might be able to figure out why there is a response at 1st harmonic.
Leon, I am not sure I understand what you mean by "the phase reversals might underlie the first harmonic" and "this (VV: this what?) doesn't explain the lack of the base frequency at 3 Hz". Did you read the paper I linked?
Leon, the harmonic sturcure of the SSVEP response depends on the type of the stimuli.
For the *prefectly symmetric * stimuli, e.g. contrast reversing sinusoidal gratings, where every screen pixel flips its luminance two times per cycle of the fundamental frequency F, the responses *must* contain only even harmonics of F (2F, 4F, 6F, ...), but no fundamental itself and no odd harmonics (3f, 5F, ...). If there are statistically significant responses at F and its odd harmonics, it can only mean the stimuli were not perfectly symmetrical, and indicate some problem with the stimulus generation algorithms, presentation timing or luminance calibration.
If the stimuli are not symmetric, e.g. pattern on/off, the responses do contain a fundamental and its odd harmonics.
In the papers that "shown SSVEP at the expected frequency (i.e., not just harmonics)", were the stimuli symmetric or not?
Thank you Vladimir for the answer and the paper. I am not sure I follow the link between your first and your last answer. For both 3Hz and 10Hz flickering stimuli we see a clear 2F harmonics (and a smaller one at 4F), which fits with the symmetric stimuli we used. Could you tell us more about the issue with the 1st harmonic?
Thanks
Thomas Hinault, Leon Gmeindl
Let's summarize. In your original post you described the stimuli as a "flickering sine-grating of black and white" and show the spectra. My take on it: The stimuli were symmetric, the responses should contain even harmonics, but no fundamental and no odd harmonics. Looking at the spectra I see mostly even harmonics, which is expected, and in my first response I refer to the paper explaining this. However, at least in the 10Hz picture it seems there is a response at 10Hz, which should not be there, and I point to it (Thomas, this is "the issue with the 1st harmonic").Later Leon mentioned "Other papers have obviously shown SSVEP at the expected frequency (i.e., not just harmonics)", which led to my second post, where I tried to emphasize, that expected harmonic structure depends on the type of stimuli, therefore comparing your results to a paper it is important to make sure the stimuli were of the same type. Otherwise, comparing e.g. responses to contrast-reversing sinusoidal gratings with responses to appearing-disappearing white dot is like comparing apples to oranges.
I would be glad to answer your questions, but let's be precise in describing the stimuli and findings. This is important, otherwise I cannot reply anything to "we found the same results for contrast reversing sinusoidal gratings as well as solid black/white stimuli". What results? What are solid black/white stimuli - appearing-disappearing white dot? Dot size? Screen background - black, grey? If black, this is a "pattern appearing-disappearing" asymmetrical stimuli with odd harmonics expected in the response.
If you would like to discuss the topic in more detai, I think it will be more effective to take the conversaion off RG. You may look up my e-mail at Stanford Psychology dept. Best regards.
Hi all, in our experiments we've seen increases to stimulus harmonics that are dependent on both flicker structure (black-white vs black-grey-white for example), and the allocation of attention. If you have not seen it already, I recommend checking this paper (and associated citations) for more info : Article Differential Roles of Frequency-following and Frequency-doub...
.I'd like to echo VV's suggestion above, more information on the stimulus presentation/configuration would be helpful. It sounds as though you might expect an increased response from light-following neural populations at 3 or 10 Hz, but since these stimuli were symmetric, overall (i.e. for any-change in polarity) you can expect a 6 or 20 Hz response. Is your updated stimulus still a sine-wave, or now a square wave? It also seems that you are averaging over a variety of occipital and parietal sources in the plots above. The topography of different SSVEP harmonics can vary, so first plotting the log-power (or SNR) is worthwhile, before combining across electrodes.
- Badges
- Science topic