Neurofeedback training of the upper alpha frequency band in EEG improves cognitive performance
Research Highlights
►Human EEG oscillations in the upper alpha frequency band are enhanced by neurofeedback. ►EEG-enhancement was accompanied by an improvement of performance in a cognitive task. ►EEG-enhancement due to neurofeedback was restricted to the frequency range that was used for feedback.
Introduction
Recently, the significant role of oscillations for brain functions and behavior as well as for psychiatric diseases became increasingly obvious (Basar & Güntekin, 2008, Basar et al., 2000, Herrmann & Knight, 2001, Herrmann et al., 2004, Strüber & Herrmann, 2002, Uhlhaas & Singer, 2006). Neurofeedback training (NFT) as an operant conditioning method to control oneself's brain activity has been shown to be an appropriate way to control or change these oscillations. In addition to clinical treatment for attention-deficit hyperactivity disorder (Lubar, 2003), epilepsy (Sterman, 2000) and other disorders (Saxby & Peniston, 1995, Gruzelier et al., 1999, Hardt & Kamiya, 1978), NFT is applied to train locked-in patients to communicate (Birbaumer et al., 1999). It even seems to have effects on certain performance measures like semantic working memory (Vernon et al., 2003) and mental rotation ability (Hanslmayr et al., 2005; see Vernon, 2005, for an overview).
But the application of NFT has also received criticism concerning the reliability of its effects. Egner et al. (2004) addressed one main problem that–despite of significant behavioral impact–sometimes no spectral effects can be found after NFT. In that study, when effects with regard to the EEG were observed at all, they often were unreliable or did not meet expectations concerning the frequency spectrum or the topography (see also the tables in Vernon, 2005).
We propose three criteria for the validation of a neurofeedback parameter: There should be spectral effects within the trained frequency band caused by the training (trainability). These spectral changes should not affect other frequency bands (independence). Finally, it is reasonable to choose a frequency band that is associated with certain cognitive functions to increase the probability of reliable behavioral effects as well as applicability (interpretability).
Many of the existing studies on NFT parameters do not satisfy all three criteria. The alpha band (8–12.5 Hz), for example, has been shown to be increased after NFT (Bauer, 1976), but for the frequency range tested there, no cognitive changes could be demonstrated. In a recently published article, Cho et al. (2008) reported an increased amplitude in the same frequency band, but possible effects on cognitive performance were not examined. Other studies concerning the impact of non-clinical NFT of SMR (sensory-motor rhythm, 12–15 Hz) on EEG and performance either found no effects or found effects but at unexpected frequencies or electrodes (e.g., Vernon, 2005, Egner et al., 2004).
Constraining alpha to the individually determined upper alpha band (ranging from the individual alpha frequency, IAF, to IAF+2 Hz) might cause an improvement of trainability (Hanslmayr et al., 2005). However, this was shown only within 1 day of NFT—a possible long-term effect was not examined. There is evidence of at least two independent (lower and upper) alpha subbands (Angelakis & Lubar, 2002, Klimesch et al., 1997, Michels et al., 2008). Finally, upper alpha is widely shown to be correlated with cognitive performance (for a review, see Klimesch, 1999), indicating interpretability. For instance, prestimulus activity in the UA band has been shown to be positively related to performance in mental rotation tasks (Hanslmayr et al., 2005, Klimesch et al., 2003).
Despite these promising results, the usability of upper alpha (UA) for NFT and its spectral as well as cognitive impact for longer periods of time are hardly investigated yet. Thus, in the current study the trainability of the UA band and its impact on cognitive abilities were examined. Our hypotheses were (i) an increase of the UA amplitude, (ii) being independent of other frequency bands, and (iii) being related to improved performance.
Section snippets
Design
For each subject, the experiment consisted of five sessions within the same week from Monday to Friday with one training session each day. Within one session, the structure was the following (Fig. 1A):
At first, a short EOG (electrooculogram) calibration measurement (not depicted) was conducted for subsequent detection of eye-movement artefacts. Then, a base rate of 5 min was recorded. Afterwards, five training blocks of 5 min each, followed by a second base rate measurement, were acquired. On
Trainability
The average IAF was 9.2 ± 0.91 Hz during the first base rate and 9.16 ± 0.76 Hz during the second base rate. No significant changes were observed in the course of the training.
In Fig. 1B, the average UA amplitudes across all responders are shown as a function of the temporal course of the study. Trainability is reflected by the positive slope of the UA amplitudes over the course of time. The gradients of a fitted regression line for each subject (n = 11) are significantly larger than zero (t(10) = 3.18, p
Discussion
In line with our hypotheses, the trainability of the UA band (hypothesis i) has been confirmed: All but three of the fourteen subjects of the NFT group showed significant training success. This is remarkable, since in therapy usually a ten-fold higher number of sessions is used (Lubar, 2003).
In the course of the week, a linear increase of the UA amplitude was visible (Fig. 1B). Also in between sessions the training effects remained present over an extended period of time. Even before training,
Conclusions
Our study revealed promising results for the usage of individually determined UA as a NFT frequency parameter, fulfilling the criteria of trainability, independence, and interpretability, thus inspiring further examinations of the interactions and relations between UA, NFT and cognitive performance.
References (35)
- et al.
A review of brain oscillations in cognitive disorders and the role of neurotransmitters
Brain Res.
(2008) - et al.
Brain oscillations in perception and memory
Int. J. Psychophysiol.
(2000) Short-term memory: EEG alpha correlates and the effect of increased alpha
Behav. Biol.
(1976)- et al.
Intelligence related differences in EEG-bandpower
Neurosci. Lett.
(2005) - et al.
The effects of neurofeedback training on the spectral topography of the electroencephalogramm
Clin. Neurophysiol.
(2004) Event-related desynchronisation in the EEG during emotional and cognitive information processing: differential effects of extraversion
Biol. Psychol.
(2005)- et al.
EEG alpha band dissociation with increasing task demands
Brain Res. Cogn. Brain Res.
(2005) - et al.
Critical validation studies of neurofeedback
Child Adolesc. Psychiatr. Clin. N. Am.
(2005) - et al.
Learned control of slow potential interhemispheric asymmetry in schizophrenia
Int. J. Psychophysiol.
(1999) - et al.
Emotional face expressions are differentiated with brain oscillations
Int. J. Psychophysiol.
(2007)
Mechanism of human attention: event-related potentials and oscillations
Neurosci. Biobehav. Rev.
EEG alpha and theta oscilliations reflect cognitive and memory performance: a review and analysis
Brain Res. Rev.
Event-related desynchronisation in the alpha band and the processing of semantic information
Brain Res. Cogn. Brain Res.
EEG alpha distinguishes between cuneal and precuneal activation in working memory
Neuroimage
MEG alpha activity decrease reflects destabilization of multistable percepts
Cog. Brain Res.
Neural synchrony in brain disorders: relevance for cognitive dysfunctions and pathophysiology
Neuron
The effect of training distinct neurofeedback protocols on aspects of cognitive performance
Int. J. Psychophysiol.
Cited by (364)
Attention-related power and functional connectivity modulation associated with long-term alpha neurofeedback training
2024, Biomedical Signal Processing and ControlDyadic boundary points based empirical wavelet transform for the elimination of eye movement and eye blink-based ocular artifacts from EEG signals
2023, Biomedical Signal Processing and ControlITACA: An open-source framework for Neurofeedback based on Brain–Computer Interfaces
2023, Computers in Biology and Medicine