What is neurofeedback for cognitive enhancement? 

Neurofeedback is a human enhancement technique whose final objective when oriented to the general population, is to optimize brain function to achieve cognitive improvements in psychological variables such as memory, attention, processing speed or executive functions. In this brain training technique, certain brain patterns, for instance, those related to cognitive function, are monitored in real-time using an electroencephalogram (EEG) and fed back to the user in an auditory or visual form, using the computer screen. This way, the user can learn to shape, to a certain degree, some patterns of brain activity related to cognitive function, and consequently to achieve cognitive enhancement.

Figure 1: Closed-loop operation of Neurofeedback. First, the EEG signals are registered and some brain patterns of interest are decoded (extracted) in real-time. These brain patterns are fed back to the user, using, in this particular case, a visual representation on a computer display. The user can thus learn how to module her/his brain patterns in the desired way.

Neurofeedback is a specialization of a broader set of techniques, named biofeedback, which is oriented to obtain a certain degree of control of physiological variables such as heart rate, respiration, skin conductivity, etc. When the physiological variable of interest is “brain wave activity” (EEG), then it is called EEG biofeedback or neurofeedback. Independently of the (neuro)physiological variable, all of these share a common operating principle: a person needs to get feedback about her/his own (hidden) physiological variables in real-time to promote learning/control through operant conditioning (Skinner, 1938). 

How does neurofeedback work? The science behind

The neurofeedback principle or model is simple and its operation is depicted in the previous figure. However, there exist several protocols in the literature according to the selected brain patterns of interest (see Gruzelier 2014 for a review). Below, we will focus on one of the most validated protocols for cognitive enhancement: the up-regulation of the subject-specific alpha band. This means the neurofeedback implementation decodes, in real time, activity in the alpha band, calibrated for each participant, and over posterior locations of the scalp.  It then provides feedback to the user accordingly, who is encouraged to increase her/his levels of activity (i.e., up regulate).

Alpha, cognitive function and neuroplasticity

The brain controls our cognitive functions such as attention, working memory, and executive functions. Brainwave activity recording techniques such as EEG, commonly used in clinical and research environments, has enabled the study and characterization of brain activity patterns related to cognitive functions.

One of the most studied brain patterns to date is alpha activity (Klimesch, 1999). Alpha is the dominant frequency in the human EEG, first recorded in the 1920s (Berger, 1929). This rhythm is characterized by a “peak” in the spectral analysis in the (8-12 Hz) frequency range (Figure 2), and predominates during wakefulness relaxation with closed eyes, best observable over posterior areas of the scalp (Klimesch, 1999). The parieto-occipital alpha rhythm is attenuated by eye-opening, visual stimuli, and by increased attentiveness (Palva and Palva, 2007). It also responds to motor tasks (Pfurtscheller, 1999) and to different cognitive demands such as attention and memory tasks (Klimesch, 1999).

Figure 2: The alpha peak is visible at 10.25 Hz for this individual. According to this, we could define the subject-specific alpha band as the areas colored in different blue tonalities (see Section “Inter- and intra-subject variability of brain patterns”)

Neuroscientific research has drawn hypotheses establishing a causal link between the alpha activity and cognitive performance (Klimesch 1999). One of them suggests that alpha is related to the cognitive performance by actively inhibiting information irrelevant to the task being executed (Klimesch et al. 2007; Jensen and Mazaheri 2010). It has been experimentally explored, and increased alpha activity (by neurofeedback) has shown cognitive improvements in working memory, attention, and visuospatial abilities (Hanslmayr et al. 2005; Zoefel et al. 2011; Nan et al. 2012).

Another hypothesis complementing the previous one is that these techniques might induce neuroplastic changes via learning of self-regulation of the alpha activity. Plasticity is a neural adaptation mechanism that reflects the brain's ability to reorganize throughout life and occurs at different levels (structural, functional, molecular, and cellular).  This has a reflection on the activity and function of the area on the one that produces the change. At least three ways of generating neuroplastic changes relevant to cognitive function are known: with drugs, with electrical stimulation, or through learning a task. Those produced by this last process, such as what neurofeedback approaches promote, are endogenous and natural to humans, so they have the potential to consolidate over time. In addition to this, neurofeedback has not shown any side-effects.

Modern vs Classic Neurofeedback for Cognitive Enhancement

The ability to self-regulate brain electrical activity in humans was first demonstrated in the 1960s, targeting alpha oscillations at occipital locations (Kamiya, 1969). Although classical neurofeedback approaches have been used for decades in research and clinical practice, modern scientific and technological improvements have been introduced due to the great advances in the field of Brain-Computer Interfaces. 

These improvements are mainly in the use of high-quality EEG equipment and advanced signal processing algorithms (in terms of artifact filtering and decoding based on AI techniques), which in the specific context of neurofeedback, would allow an adaptation of the online training to a particular subject and moment of usage of the technology (Figure 3).

Figure 3: Closed-loop operation of a modern Neurofeedback. First, a calibration step is performed to adapt the online training to the subject-specific brain patterns (learned using AI techniques). Then, during online training, the EEG signal is filtered from artifacts, and the brain patterns of interest decoded (using the previous calibration settings), which are finally fed back to the user in a monitor display.

1. EEG acquisition technology

There is a wide range of EEG registration systems available, which vary from very reliable research/medical-grade equipment to low-cost wearables with low reliability. It is important that the equipment: 1) presents a high signal quality acquisition to accurately measure brain patterns of interest; and 2) has a sufficient number of sensors, well distributed over the scalp, to allow artifact filtering techniques.

• Classic approach: Low-cost equipment with very few sensors, even one or two (Demos, 2005); which might compromise the quality of the decoded brain patterns. 
• The modern approach: Modern approaches benefit from high-quality EEG acquisition techniques with the right number of sensors distributed over the desired brain areas. Within the modern approach we can distinguish between several types of equipment depending on the needs: dry-EEG equipments, high-quality systems but user-friendly and optimized for real-world applications; or high coverage equipments, such as the EEG caps. Both are equally appropriate for BCI-based neurofeedback. 

Artifact filtering

EEG artifacts are all those electric signals that have a non-neural origin but are still registered by the EEG equipment and blend with the true brain activity. This complicates real-time identification of brain patterns of interest, and might compromise the effectiveness of the training. 

• Classic approach: Some approaches do not perform any filtering, which might compromise the effectiveness; or use simple approaches such as pausing the feedback while an eye artifact is detected, thus reducing the effective training time.
• The modern approach: There are many types of artifacts that interfere with alpha. One of the EEG artifacts with a higher occurrence rate is an eye blink. These artifacts have an stable spatial pattern and can be filtered out using blind source separation techniques (Hyvarinen, 1999), which improves the signal quality and increases the effective feedback time (Figure 4).

Figure 4: EEG signal (7 seconds) recorded from 16 electrodes. Raw signal is depicted in blue color, showing blinking artifacts, most apparent in anterior locations. Filtered signal is superimposed in black color.

Inter- and intra-subject variability of brain patterns

It is well known that EEG brain patterns, and alpha activity among them, have high inter-subject variability (Haegens et al. 2014). This variability is even aggravated when it comes to the clinical population, e.g., the study of EEG has revealed distinct brain patterns in some clinical population such as ADHD (Figure 5).

In addition, alpha activity has high intra-subject variability. It can change among training sessions due to changes in cognitive/emotional state (Klimesch, 1999). In the context of neurofeedback, it can also change due to the self-regulation process which aims at enhancing it among the EEG biofeedback sessions (Figure 6).

Figure 6: EEG power spectrum. This shows the intra-subject variability in alpha rhythm in resting state (blue line) and when performing a cognitive task (red line) for an individual suffering from major depression and one with ADHD in 3 sessions within a training program.

• Classic approach: Some classic approaches do not take into account this variability of alpha activity, considering the alpha band as a fixed interval in the [8-12] Hz frequency band (Vernon, 2005). While there is a consensus in defining alpha that way for the average general population, there is a high individual variability to be considered.
• The modern approach:  Modern approaches adapt the brain patterns to a particular subject, commonly executing a calibration step before online training. In particular, there is a trend to consider the alpha band as a dependent measure on the peak alpha frequency, referred to as the Individual Alpha Frequency (IAF, Klimesch, 1999). Alpha band can be thus determined as the (IAF-2, IAF+2) Hz interval (Figure 2). In addition, the baseline for the online training needs to be recomputed at the beginning of each training session to accommodate the inter-session (intra-subject) variability.

Brain-computer interface neurofeedback at work

This video shows how a modern neurofeedback approach works in real-time. First, the raw EEG activity is recorded: the upper left figure media shows the EEG recorded in five electrical sensors located over the parieto-occipital area. This activity is transformed into the frequency domain using a short-term Fourier transform, specifically on the last EEG second (gray color). This frequency information is displayed as the power spectra in the [0-30] Hz, where the gray area represents the upper alpha power (upper center media). The topographical distribution of the instantaneous upper alpha power values in all the scalp is also displayed (upper right media). Time-instantaneous upper alpha values (averaged across the parieto-occipital locations) can then be mapped to a color scale using the settings obtained in an initial calibration step that characterizes the inter- and intrasubject variability. The center media shows the time-course feedback value, which directly translates to a color scale, whereas the bottom media shows the final visual feedback that receives the participant.

Use cases

Below we report some scientific results of a modern neurofeedback approach in mental health aimed at cognitive enhancement, comprising the general population, individuals suffering from major depressive disorder, and children with ADHD. They can provide the reader with an overview of the results that can be expected at an electrophysiological and cognitive level after the application of such technology. See the original publications in peer-reviewed journals and conferences for an in-depth view.

Patients with Major Depressive Disorder (Escolano 2014a)

Controlled study. The experimental group (n=40) completed 8 neurofeedback sessions and was compared against a non-interventional control group (n=20). At an electrophysiological level, only the experimental group showed a significant increase (pre/post study) in the subject-specific alpha activity, with a 25% average increase. The main result of the cognitive evaluation was a significant increase for the experimental group only in the PASAT test: 24% decrease of errors, and 15% decrease in the execution time. This suggests that the working memory and processing speed improved for patients suffering from depression, thus alleviating the cognitive symptoms. 

Figure 7: Neuroplastic changes in patients with Major Depressive Disorder. The alpha activity is shown through the 8 training sessions in the EEG screenings (black dots, pre-post session) and during the training trials (grey dots). A positive increase can be shown, thus suggesting a learning of the self-regulation process.

Children suffering from attention deficit hyperactivity disorder ADHD (Escolano 2014b)

Exploratory, uncontrolled study. Children with ADHD (n=20) performed 18 neurofeedback sessions. At an electrophysiological level, a significant increase (pre/post study) was found in the subject-specific alpha activity, with a 13% average increase. Furthermore, the cross-sessions trend in different frequency bands was analyzed, obtaining an increase in alpha and a decrease in low-frequency activity (note that ADHD children usually show an excess of low-frequency activity). 

The cognitive evaluation showed an increase of 16% and 10% of correct answers in the Letter–Number Sequencing test of WISC-IV, suggesting an enhancement of working memory and sustained attention. Finally, parents reported a clinical improvement of their children regarding the attention deficit and hyperactivity/impulsivity (approx... 9 points in both) measured with the Conners’ Parent Rating Scales (CPRS-R). 

Figure 8: Neuroplastic changes in children with ADHD. Frequency-sensors maps for absolute and relative activity are shown. The frequency range is expressed relative to the value of individualized alpha, i.e., the upper part of alpha corresponds to the [0-2] interval. A significant increase in the absolute and relative activity of the alpha rhythm is observed, most prominent in the central and parieto-occipital areas.  

General population (Escolano, 2011)

Controlled study. The experimental group (n=6) completed 5 neurofeedback sessions (consecutive days) and was compared to a non-interventional control group (n=6). The sample consisted of university students with an average age of 25 and 27 years, respectively. No EEG recordings were performed on the control group.  At an electrophysiological level, the experimental group showed a significant increase (pre/post study) in the subject-specific alpha activity, with a 65% average increase. The experimental group showed a significant increase (12%) in the number of words remembered in the conceptual span test, thus suggesting an improvement in working memory.

Figure 9: Neuroplastic changes. The alpha activity across the 8 training sessions is shown in the EEG screenings (black dots, pre-post session) and training trials (gray dots). A positive trend is visible, suggesting a learning of the self-regulation process. 

General population, sham-controlled single session study (Escolano 2014c)

Double-blind, sham-controlled study. The experimental group (n=10) received only one training session and was compared to a placebo control group (n=9). The sample consisted of university students with an average age of 26 and 24 years, respectively. At an electrophysiological level, only the experimental group showed an increase in the subject-specific alpha activity (13% on average) as well as a desynchronization increase during the execution of a cognitive task (phasic activity, 16%). At a behavioral level, there was no significant difference between the two groups, suggesting that a single training session is not enough to achieve a cognitive enhancement (although electrophysiological changes were already apparent).

Figure 10: Neuroplastic changes. The alpha activity is displayed for the experimental and control group (left and right figures, respectively) in the EEG screenings (blue color, pre-post session) and training trials (black color). A positive trend is observable for the experimental group only.

General population, effects in cognitive performance and mindfulness (Navarro-Gil, 2018)

Controlled study. This one focused on the evaluation of the effects in cognitive performance and mindfulness scale. The experimental group (n=27) performed 6 training sessions and was compared to a non-interventional control group (n=23). Average age was 37 and 35 years, respectively. At an electrophysiological level, only the experimental group showed a significant increase (pre/post study) in the subject-specific alpha activity, with a 31% average increase. The main result of the cognitive evaluation was a significant improvement for the experimental group only in the PASAT test: 55%  decrease in wrong answers and 5% in execution time. Additionally, the experimental group showed an increase in the mindfulness scale: 12 points increase in the overall FMQ variable and 6 points in the MAAS.

Figure 11: Neuroplastic changes: The increase in alpha rhythm (pre/post study) is shown for the experimental group (black dot) and control group (white dot), normalized with respect to the first measurement.

General population, out of the laboratory results (Escolano, 2019a; 2019b)

Evaluation of the technique out of the laboratory (no control group). 59 participants underwent five training sessions, with pre- and post- cognitive evaluation sessions on the first and last days. Analysis of the pre-post enhancement in the trained parameter revealed a significant increase with an average increase of 40.2%. The cognitive evaluation showed a significant increase in the PASAT test (55.6% increase in the number of recalled words, and an 8% decrease in the execution time), thus suggesting an improvement in working memory.

Figure 12: Neuroplastic changes. The alpha activity across the 5 training sessions is shown in the EEG screenings (black dots, pre-post session) and training trials (gray dots). A positive trend is visible, suggesting a learning of the self-regulation process. Upper left box shows the intra-session alpha activity (averaged across sessions), showing a positive trend as well.

Conclusions

The use of modern neurofeedback based on brain-computer interface technology (BCI) can provide, on one hand, a higher understanding of the effects of the neurofeedback at electrophysiological level. The aforementioned results consistently show an increase on the subject-specific brain patterns of interest, which were found in neuroscientific literature to correlate with cognitive performance due to inhibitory mechanisms of the brain.

On the other hand, due to this scientific-technological approach, where all the training parameters are computed based on computer algorithms according to the subject-specific brain patterns in the moment of use of the technology, one might expect a higher reliability in the cognitive outcomes. Although clinical trials with larger population samples and more rigorous control conditions would be warranted, general cognitive improvements were consistently obtained, specifically in working memory, attention, and processing speed.

Our Elevvo technology implements modern BCI-based neurofeedback for cognitive enhancement, with the latter study reporting the results corresponding to its application by customers of the technology (Escolano, 2019a; 2019b). These positive results indicate that this innovative modern EEG biofeedback therapy is a powerful complementary cognitive treatment for those populations who wish to maintain and enhance their cognitive functions. BCI-based neurofeedback therapies have demonstrated to be more effective and the induced neuroplastic changes are maintained for a longer time. 

References

• Berger, H. (1929). Über das elektrenkephalogramm des menschen. European Archives of Psychiatry and Clinical Neuroscience, 87(1):527–570.
• Demos, J. N. (2005). Getting started with neurofeedback. WW Norton & Company.
• Escolano, C., Navarro-Gil, M., Garcia-Campayo, J., Congedo, M., De Ridder, D., & Minguez, J. (2014a). A controlled study on the cognitive effect of alpha neurofeedback training in patients with major depressive disorder. Frontiers in Behavioral Neuroscience, 8(296). https://www.frontiersin.org/articles/10.3389/fnbeh.2014.00296/full
• Escolano, C., Navarro-Gil, M., Garcia-Campayo, J., Congedo, M., & Minguez, J. (2014b). The effects of individual upper alpha neurofeedback in ADHD: An open-label pilot study. Applied Psychophysiology and Biofeedback, 39(3-4), 193-202. https://link.springer.com/article/10.1007/s10484-014-9257-6
• Escolano, C., Navarro-Gil, M., Garcia-Campayo, J., & Minguez, J. (2014c). The effects of a single session of upper alpha neurofeedback for cognitive enhancement: A sham-controlled study. Applied Psychophysiology and Biofeedback, 39(3-4), 227-236. https://link.springer.com/article/10.1007/s10484-014-9262-9
• Escolano, C., Aguilar, M., & Minguez, J. (2011). EEG-based upper alpha neurofeedback training improves working memory performance. In International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) (p. 2327-2330). Boston (USA). http://ieeexplore.ieee.org/document/6090651/
• Escolano, C., Montesano, L. & Minguez, J. (2019a). On Modern Neurofeedback Solutions based on Brain-Computer Interfaces in Uncontrolled Real-World Settings. In IEEE International Conference on Systems, Man, and Cybernetics (SMC). Bari (Italy).
• Escolano, C., Montesano, L. & Minguez, J. (2019b). A  Business Proof-of-Concept of a Brain-Computer Interface for Cognitive Enhancement. In IEEE International Conference on Systems, Man, and Cybernetics (SMC). Bari (Italy).
• Gruzelier, J. H. (2014). EEG-neurofeedback for optimising performance. I: a review of cognitive and affective outcome in healthy participants. Neuroscience & Biobehavioral Reviews, 44, 124-141.
• Haegens, S., Cousijn, H., Wallis, G., Harrison, P. J., and Nobre, A. C. (2014). Inter-and intra-individual variability in alpha peak frequency. NeuroImage, 92:46–55.
• Hanslmayr, S., Sauseng, P., Doppelmayr, M., Schabus, M., and Klimesch, W. (2005). Increasing individual upper alpha power by neurofeedback improves cognitive performance in human subjects. Applied Psychophysiology and Biofeedback, 30:1–10.
• Hyvarinen, A. (1999). Fast and robust fixed-point algorithms for independent component analysis. Neural Networks, IEEE Transactions on, 10(3):626–634.
• Jensen, O., & Mazaheri, A. (2010). Shaping functional architecture by oscillatory alpha activity: gating by inhibition. Frontiers in human neuroscience, 4, 186.
• Kamiya, J. (1969). Operant control of the EEG alpha rhythm and some of its reported effects on consciousness. Altered states of consciousness. New York: Wiley, 1069.
• Klimesch, W. (1999). EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis. Brain Research Reviews, 29(2-3):169 – 195.
• Klimesch, W., Sauseng, P., and Hanslmayr, S. (2007). EEG alpha oscillations: the inhibition-timing hypothesis. Brain research reviews, 53(1):63–88.
• Palva, S. and Palva, J. M. (2007). New vistas for alpha-frequency band oscillations. Trends in neurosciences, 30(4):150–8.
• Pfurtscheller, G. and Lopes da Silva, F. H. (1999). Event-related EEG/MEG synchronization and desynchronization: basic principles. Clinical Neurophysiology, 110(11):1842–1857.
• Nan, W., Rodrigues, J. P., Ma, J., Qu, X., Wan, F., Mak, P. I., Mak, P. U., Vai, M. I., and Rosa, A. (2012). Individual alpha neurofeedback training effect on short term memory. International Journal of Psychophysiology, 86(1):83–87.
• Navarro-Gil, M., Escolano, C., Montero-Marín, J., Minguez, J., Shonin, E., & Garcia-Campayo, J. (2018). Efficacy of neurofeedback on the increase of mindfulness-related capacities in healthy individuals: a controlled trial. Mindfulness, 9, 303-311. https://link.springer.com/article/10.1007/s12671-017-0775-1
• Skinner, B. F. (1938). The behavior of organisms: An experimental analysis. Oxford, England: Appleton-Century.
• Vernon, D. J. (2005). Can neurofeedback training enhance performance? An evaluation of the evidence with implications for future research. App. Psychophysiology and Biofeedback, 30:347–364.
• Zoefel, B., Huster, R. J., and Herrmann, C. S. (2011). Neurofeedback training of the upper alpha frequency band in EEG improves cognitive performance. NeuroImage, 54(2):1427 – 1431.

You might also be interested in: 

• Bringing BCIs to the user’s home for neurorehabilitation and assistive applications
• EEG-based Neurotechnology for Human Enhancement and Rehabilitation 
• Epilepsy and EEG seizure-detection
• All about EEG artifacts and filtering tools
• What is QEEG brain mapping and normative databases?
• How Deep Learning is changing machine learning AI in EEG signal processing
• All about EEG artifacts and filtering tools
• Main features of the EEG sensor layer explained
• Main features of the EEG amplifier explained
• How to select a dry-EEG headset for your research application
• An introduction to brain-computer interface using EEG signals