The clinical efficacy of neurofeedback is still a matter of debate. This paper analyzes the factors that should be taken into account in a transdisciplinary approach to evaluate the use of EEG NFB as a therapeutic tool in psychiatry. Neurofeedback is a neurocognitive therapy based on human–computer interaction that enables subjects to train voluntarily and modify functional biomarkers that are related to a defined mental disorder. We investigate three kinds of factors related to this definition of neurofeedback. We focus this article on EEG NFB. The first part of the paper investigates neurophysiological factors underlying the brain mechanisms driving NFB training and learning to modify a functional biomarker voluntarily. Two kinds of neuroplasticity involved in neurofeedback are analyzed: Hebbian neuroplasticity, i.e. long-term modification of neural membrane excitability and/or synaptic potentiation, and homeostatic neuroplasticity, i.e. homeostasis attempts to stabilize network activity. The second part investigates psychophysiological factors related to the targeted biomarker. It is demonstrated that neurofeedback involves clearly defining which kind of relationship between EEG biomarkers and clinical dimensions (symptoms or cognitive processes) is to be targeted. A nomenclature of accurate EEG biomarkers is proposed in the form of a short EEG encyclopedia (EEGcopia). The third part investigates human–computer interaction factors for optimizing NFB training and learning during the closed loop interaction. A model is proposed to summarize the different features that should be controlled to optimize learning. The need for accurate and reliable metrics of training and learning in line with human–computer interaction is also emphasized, including targeted biomarkers and neuroplasticity. All these factors related to neurofeedback show that it can be considered as a fertile ground for innovative research in psychiatry.
Introduction
Neurofeedback (NFB) is a neurocognitive therapy based on human–computer interaction. The objective of NFB is to enable subjects to voluntarily train and modify functional biomarkers that are specific to mental disorders, in order to improve symptoms or cognitive processes. In psychiatry, a biomarker is usually a psychophysiological variable that is objectively measured and evaluated as an indicator of pathogenic processes or therapeutic responses [1]. However, most of the current electroencephalographic (EEG) NFB protocols are not based on the modulation of disorder-specific biomarkers but on the modulation of a few spontaneous brain rhythms, mainly defined by the frequency of their oscillation [2], [3], [4]. This strategy is prevalent since spontaneous brain rhythms demonstrate a high signal-to-noise ratio in EEG recordings, and because they can be disrupted in some mental disorders, e.g. increased theta and reduced beta power in patients with Attentional Deficit and Hyperactivity Disorder (ADHD) when compared to healthy controls [5]. However, the clinical efficacy of this approach remains a controversial and delicate issue even for well-investigated applications, such as the therapeutic use of EEG NFB in ADHD [6], [7]. Indeed, the effectiveness of neurofeedback is largely debated [8], [9], [10], [11]. In this paper, we propose that several factors related to the concept of biomarker may be responsible for the conflicting results in the EEG NFB literature:
•
limited understanding of the brain mechanisms driving NFB learning to modify a functional biomarker voluntarily, i.e. neurophysiological factors [8];
•
the inconsistent relationship between EEG biomarkers and clinical dimensions (symptoms or cognitive processes), potentially due to the symptom-based classification of psychiatric disorders and the heterogeneity of diagnostic categories, i.e. psychophysiological factors [12];
•
superficial knowledge of how best to measure and optimize NFB learning during the closed loop interaction, i.e. human–computer interaction factors [13].
This paper investigates these factors (neurophysiological, psychophysiological and human–computer interaction) in a critical review of the existing literature on EEG NFB. The objective is to integrate these interdependent issues into a general NFB framework in order to demonstrate that EEG NFB can be considered as fertile scientific ground for psychiatry and to provide a roadmap for future research in this field.
Section snippets
From electroencephalographic oscillations to neurofeedback
The EEG may be recorded via non-invasive electrodes placed on the scalp as a result of intracranial fluctuations of electromagnetic field potentials, which are generated by ionic exchanges at cell membranes and synapses during neuronal activity. When neuronal activities occur in a circumscribed region and become temporally synchronized, their local field potentials (LFPs) are then spatially summated, giving rise to large fluctuations of the EEG signal [14]. Hence, changes in EEG oscillation
Dimensional approach for neurofeedback in psychiatry
Because the psychiatric nosology has weak biological grounds, on the one hand, and because the link between biomarkers (electrophysiologic biomarkers in particular with EEG or metabolic biomarkers with functional neuroimagery) and cognitive processes remain mostly unraveled, on the other hand, it is impossible to confirm the functional specificity of current NFB EEG biomarkers. In fact, contemporary psychiatry is undergoing a taxonomic crisis that is characterized by the poor diagnostic power
A human–computer interaction model for neurofeedback
To globally improve NFB efficacy in patients, it is necessary to understand and then reduce its variability. To this end, Sitaram et al. and Gaume et al. have reviewed the neurophysiological [17] and neuropsychological [12] mechanisms underlying NFB training procedures. In addition, Enriquez-Geppert et al. have proposed a tutorial explaining how to design rigorous NFB training protocols [71]. While Sitaram et al. and Gaume et al. adopted a standpoint purely centered on “human learning” (i.e.
Conclusion
This paper investigated the neurophysiological, psychophysiological and human–computer interaction foundations of neurofeedback. A transdisciplinary approach is now needed to evaluate rigorously the use of EEG NFB as a therapeutic tool in psychiatry (Fig. 4). Notwithstanding the debate on the efficacy of NFB for treating mental disorders, this field of research remains fertile ground for innovative research in psychiatry. Neurophysiology, psychophysiology and human–computer interaction
Disclosure of interest
The authors declare that they have no competing interest.
Acknowledgments
We thank Anatole Lecuyer for his participation to the second French congress on neurofeedback organized by NExT.
This work was supported by the French National Research Agency within the REBEL project (grant ANR-15-CE23-0013-01), the European Research Council with the BrainConquest project (grant ERC-2016-STG-714567), the Inria Project-Lab BCI-Lift and the EPFL/Inria International Lab.
Clinical application of brain imaging for the diagnosis of mood disorders: the current state of play
Mol Psychiatry
(2013)
M. Arns et al.
A decade of EEG theta/beta ratio research in ADHD: a meta-analysis
J Atten Disord
(2013)
J. Micoulaud-Franchi et al.
EEG neurofeedback treatments in children with ADHD: an updated meta-analysis of randomized controlled trials
Front Hum Neurosci
(2014)
T. Fovet et al.
On assessing neurofeedback effects: should double-blind replace neurophysiological mechanisms?
Brain
(2017)
J. Vion Dury et al.
Modalités d’acquisition et d’analyse du signal EEG
A.K. Engel et al.
Intrinsic coupling modes: multiscale interactions in ongoing brain activity
Neuron
(2016)
K.L. Coburn et al.
The value of quantitative electroencephalography in clinical psychiatry: a report by the Committee on Research of the American Neuropsychiatric Association
J Neuropsychiatry Clin Neurosci
(2006)
R. Sitaram et al.
Closed loop brain training: the science of neurofeedback
Nat Rev Neurosci
(2016)
F. Lotte et al.
Defining and quantifying users’ mental imagery-based BCI skills: a first step
J Neural Eng
(2018)
T. Ros et al.
Tuning pathological brain oscillations with neurofeedback: a systems neuroscience framework
Front Hum Neurosci
(2014)
S. Halder et al.
Prediction of brain–computer interface aptitude from individual brain structure
Front Hum Neurosci
(2013)
R.C. Kluetsch et al.
Plastic modulation of PTSD resting-state networks and subjective well-being by EEG neurofeedback
Acta Psychiatr Scand
(2014)
J. Kamiya
Biofeedback training in voluntary control of EEG alpha rhythms
Calif Med
(1971)
M.B. Sterman et al.
Facilitation of spindle-burst sleep by conditioning of electroencephalographic activity while awake
Science
(1970)
T. Ros et al.
Tuning pathological brain oscillations with neurofeedback: a systems neuroscience framework
Front Hum Neurosci
(2016)
T. Ros et al.
Endogenous control of waking brain rhythms induces neuroplasticity in humans
Eur J Neurosci
(2010)
C.M. Butefisch et al.
Mechanisms of use-dependent plasticity in the human motor cortex
Proc Natl Acad Sci U S A
(2000)
S.R. Heilbronner et al.
Dorsal anterior cingulate cortex: a bottom-up view
Annu Rev Neurosci
(2016)
M. Papoutsi et al.
Stimulating neural plasticity with real-time fMRI neurofeedback in Huntington's disease: a proof of concept study
Hum Brain Mapp
(2018)
C. Zich et al.
High-intensity chronic stroke motor imagery neurofeedback training at home: three case reports
Clin EEG Neurosci
(2017)
J. Ghaziri et al.
Neurofeedback training induces changes in white and gray matter
Clin EEG Neurosci
(2013)
B.N. Cuthbert
The RDoC framework: facilitating transition from ICD/DSM to dimensional approaches that integrate neuroscience and psychopathology
World Psychiatry
(2014)
T. Insel et al.
Research domain criteria (RDoC): toward a new classification framework for research on mental disorders
Am J Psychiatry
(2010)
A. Buttfield et al.
Towards a robust BCI: error potentials and online learning
EEE Trans Neural Syst Rehabil Eng
(2006)
A. Brandmeyer et al.
Decoding of single-trial auditory mismatch responses for online perceptual monitoring and neurofeedback
