Abstract
In frequency tagging, visual stimulation at a frequency (F) of ∼10 Hz has long been known to generate the highest-amplitude response at F in the frequency domain over the human occipital cortex with electroencephalogram and other high temporal-resolution methods. Brain responses are indeed commonly assessed simply at F (i.e., the first harmonic = 1F), under the assumption that the response is represented at a single frequency, i.e., "steady-state" or approximately sinusoidal in terms of amplitude over time. This condition is met at stimulus presentation frequencies above ∼4-8 Hz in the visual modality; consequently, frequency tagging has often been limited to F above this "floor." Here, we support a less-common perspective, that frequency-tagged responses do not need to be steady-state, such that slower F are valid. In this case, it has been shown that is not appropriate to measure nonsinusoidal responses at only F but that nonsinusoidal responses can still be analyzed simply and advantageously in the frequency domain through baseline-corrected amplitude summation across harmonics (F + 2F + 3F… = F (s)). Critically, we demonstrate that although the highest-amplitude F response occurs at F = 10 Hz, the highest-amplitude F (s) response occurs at approximately F < 4 Hz. We use this example toward understanding the effect of stimulation frequency on response amplitude and discuss its caveats and limitations. We address what defines an "optimal" stimulation frequency (note: it may not always be the F yielding the highest-amplitude response) and reflect on considerations when choosing a stimulation frequency in different contexts.