Can a placebo increase learning rates? Evidence from neurofeedback studies

In brain-computer interface (BCI) as well as in neurofeedback (NF) applications, users employ various mental strategies to evoke certain brain activation patterns to control, e.g., external devices or to reach a specific mental or cognitive state [1–3]. However, a large proportion of potential BCI and NF users do not succeed in modulating their brain activation in the desired direction [4]. About 30% [4, 5] to even 50% [6] of BCI or NF users face such difficulties and the reasons for this are still largely unclear. A failure to voluntarily generate certain brain activation patterns during BCI/NF training is associated with poorer training outcomes. For instance, Kober et al. [6] showed that patients with Multiple Sclerosis (MS) who successfully up-regulated their SMR (sensorimotor rhythm, 12-15 Hz) activity in the EEG during NF training also showed significant cognitive improvements after 10 sessions of NF training compared to a pre-test, while patients with MS, who were not able to increase their SMR during NF training (so called non-responders) did not show any cognitive changes. Therefore, a body of research tries to find predictors for the difficulty of modulating one's own brain activation in a desired direction during NF or BCI training. There is evidence that motivation of the user [7], mood [8, 9], control beliefs [10], or the used mental strategies [3, 11] play a role to a certain extent. The resting or baseline EEG seems to be the most stable predictor for successful voluntary neuronal regulation [12, 13]. However, it is still not possible to reliably predict the success of the NF/BCI training performance or to optimize the technical parameters to significantly and consistently reduce the number of users who do not demonstrate successful NF/BCI learning rates.