Introduction: Alpha/Theta neurofeedback treatment (A/T NFT) has been administered to adults with anxiety disorders since the late 1960s, yet the efficacy of this treatment remains unclear. The present, single-blind study, for the first time, uses an active placebo NFT control group to test the A/T NFT protocol for trait anxiety on prodromal and clinical adult female participants. The effects this treatment has on activation and arousal states, self-perceived anxiety levels, neural oscillations, and other parameters were assessed.
Methods: Twenty-seven women ranging in age from 19 through 69 who had scored higher than the 66th percentile in the STAI trait anxiety sub-scale (75% of whom had previously been diagnosed with an anxiety disorder) were randomly assigned to either the experimental (EG) or the control group (CG). The EG (n = 14) received ten sessions of A/T NFT in which alpha and theta EEG amplitudes were uptrained at Pz. The CG (n = 13) received ten sessions of active placebo NFT at Pz. During successive sessions beta- (15–19 Hz) and high beta amplitudes (20- 24 Hz) were uptrained or downtrained. Growth curve modeling (GCM) and traditional 2x5 repeated measures ANOVA were performed on the NFT sessions data to model individual and average group learning curves. Cognitive variables, such as treatment outcome expectancy, personal attribution styles, use, types, and efficacy of cognitive strategies in NFT, and correlations between NFT learning performance, time of day the NFT sessions were held, and a participant’s best or worst time to learn, were also investigated.
Results: The analysis of individual learning curves, GCM, and ANOVA all confirmed that the majority of participants of the EG up-regulated absolute and relative A+T amplitudes within a NFT session, but so did the participants of the CG. However, a non-significant trend for the EG to have steeper learning curves was observed. Participants of both the EG and the CG felt significantly more deactivated by the end of a NFT session and reduced their self-perceived anxiety on all anxiety measures (STAI, BAI, GAD-7) by the end of the NFT trial. Although a trend could be observed that the EG reduced anxiety scores more than the CG, these differences did not rise to statistical significance. Lastly, no significant changes in the pre-post trial QEEG were found, although a trend of higher combined relative A+T power at the end of the trial was observed in the EG. In the EG the use of mental strategies was correlated with lower T/A ratio difference scores between the beginning and the end of the NFT trial but not with increased relative and absolute T+A amplitudes. The Time-of-day participants prefer or avoid learning did not correlate significantly with alpha or theta NFT amplitudes, i.e., NFT sessions being held during sub-optimal times of day were not associated with poorer learning performance.
Conclusions: For both EG and CG absolute and relative T+A amplitudes increased within sessions and absolute and relative alpha increased across sessions although the CG protocol had not included an uptraining of alpha or theta amplitudes, nor low beta amplitudes (below 15 Hz) which may have represented upper alpha peak frequency in some of the younger participants. Thus, upregulation of beta and upper beta in NFT may be associated with alpha frequency uptraining due to functional coupling of alpha and beta EEG frequencies or it may be due to placebo and other non-specific effects such as EEG frequency drifts, alpha’s idling mode and inhibitory role during task performance, or perhaps simply that some frequency bands (alpha) are more susceptible to change and easier to train. Especially the inhibition of flanking bands in the NFT protocol, i.e., beta bands in A+T training, to prevent frequency drifts, will be necessary along with detailed GCM modeling of all frequency bands to see if and how the bands change over time and how those processes relate to NFT learning curves.
Keywords: neurofeedback, EEG biofeedback, quantitative EEG, trait / state anxiety, anxiety disorders, active placebo control, alpha/theta protocol, growth curve modeling.