Nikulin V, Linkenkaer-Hansen K, Nolte G, Lemm S, Müller KR, Ilmoniemi RJ. A novel mechanism for evoked responses in the human brain. Eur J Neurosci 2007;25:3146–54. Nikulin V, Linkenkaer-Hansen K, Nolte G, Curio G. Non-zero mean and asymmetry of neuronal oscillations have different implications for evoked responses. Clin Neurophysiol 2009. doi:10.1016/j.clinph.2009.09.028 . van Dijk H, van der Werf J, Mazaheri A, WP Medendorp, Jensen O. Modulations in oscillatory activity with amplitude asymmetry can produce cognitively relevant event-related responses. Proc Natl Acad Sci USA 2009. doi:10.1073/ pnas.0908821107 . Ole Jensen a,* Hanneke van Dijk a,b Ali Mazaheri a a Donders Institute for Brain, Cognition, and Behaviour, Radboud University Nijmegen, NL-6500 HE Nijmegen, The Netherlands b Institute for Clinical Neuroscience and Medical Psychology, Heinrich-Heine University, D-40225 Düsseldorf, Germany * Tel.: +31 243610884. E-mail address: ole.jensen@donders.ru.nl (O. Jensen) 1388-2457/$36.00 Ó 2010 International Federation of Clinical Neurophysiology. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.clinph.2010.02.012 Non-zero mean of oscillations as a mechanism for the generation of evoked responses Reply to ‘‘Amplitude asymmetry as a mechanism for the gener- ation of slow evoked responses” We thank Jensen et al. (2010), this issue, for their commentary on our recent paper (Nikulin et al., 2010), which presents several statements that are useful in considering the relative importance of the ‘‘amplitude asymmetry” and ‘‘non-zero mean” of EEG/MEG oscillations in the generation of evoked responses. Jensen et al. write: ‘‘The notion of ‘non-zero mean’ pertains to an observation rather than a mechanism”. We emphasize here that any ‘‘mecha- nism” should link cause and effect, not merely correlate with an ef- fect. For the following discussion it is necessary to distinguish between (1) physiology, (2) generic mathematical models, and (3) analytical tools. Nikulin et al. (2007, 2010) and Mazaheri and Jensen (2008) agree that unidirectional primary currents in pyramidal cells are the likely neurophysiological cause for the mean of EEG/MEG oscil- lations not to be zero. Further, both research teams use a generic mathematical model of oscillations with a non-zero term. Thus, there is room for discussion about analysis techniques to estimate this non-zero term. To investigate non-zero mean oscillations, Nikulin et al. (2007) developed the baseline-shift index (BSI), which shows how base- line shifts are related to the amplitude envelope of oscillations. Oscillations with a non-zero mean may exhibit amplitude asym- metry, i.e., have peaks and troughs of different magnitudes. Maza- heri and Jensen (2008) used this observation for developing the Amplitude Fluctuation Asymmetry index (AFA index ), which is in- tended to indicate whether evoked responses are generated through amplitude modulation of oscillations. Remarkably, Jensen et al. propose that because the AFA index may occasionally be corre- lated with evoked responses and because it is an index of ampli- tude asymmetry, the amplitude asymmetry per se should be viewed as ‘‘a mechanism”. Furthermore, Jensen et al. maintain that non-zero mean of oscillations ‘‘...can best be explained by amplitude asymme- try, in which peaks are modulated stronger than troughs (Fig. 1B)”. Although their Fig. 1 was meant to emphasize the asym- metry of the oscillations, its most important feature is the demon- stration that averaging of such non-zero mean oscillations leads to an evoked response. The averaging acts as a low-pass filter and re- sults in a slowly varying baseline shift. This in turn reflects the fact that the oscillations contain a slowly varying DC-like compo- nent, which scales proportionally with the changes in the ampli- tude of oscillations. Without these baseline shifts, no evoked responses can be created through amplitude modulation of ongo- Fig. 1. (A) Conventionally neuronal oscillations are thought to be symmetric around zero; i.e. the mean of the signal will be unbiased with respect to the magnitude of the signal. (B) Recent work has demonstrated that the mean of the oscillatory signal in fact is biased with respect to the magnitude of the signal (Nikulin et al., 2007; Mazaheri and Jensen, 2008). This could be explained by the notion of amplitude asymmetry: i.e. only the peaks, but not the troughs, of the signal are modulated over time. (C) Systematic modulations in oscillatory activity in response to stimuli (t = 0 s) can produce evoked response due to amplitude asymmetry. Each of the three upper traces represents an EEG/MEG trial and the corresponding task modulation. When the trials are averaged a slow evoked responses is produced (last trace). This phenomenon has been empirically demonstrated by Mazaheri and Jensen (2008) and van Dijk et al. (2009). Letters to the Editor / Clinical Neurophysiology 121 (2010) 1148–1150 1149