At present, as well as in the past, there is minimal funding for research in areas such as NFB and this is in part because it cannot be easily mass marketed in the same manner as drugs, resulting in large corporate profits.  It is a difficult technology to monetize, as many business experts have commented.  Another difficulty is the nature of research methods when engaging in behavioral sciences.  The behavioral sciences have long struggled to be more positivistic and reductionistic in their research models, even though that paradigm died long ago with the emergence of the Copenhagen interpretation of quantum mechanics (Baggott, 2020; Bohm 1995).  Controlled group experimental designs and the statistical procedures associated with it, emerged out of the USDA agricultural research department in the first half of the 20th century (Parolini, 2015).  Human beings are far more complex than plants and the number of potential confounds involved in defining variables and controlling for potentially hundreds if not thousands of them all influencing each experimental effort is staggering.  It can be well argued at this point in history that most experimental designs are not as well suited for human research as we had hoped in the past. This is just part of why 40% of psychological research and 60% of medical research has been deemed invalid (Ioannidis, 2005).  There is little evidence to support the mistaken belief that case study design is inferior when investigating human beings (Lincoln & Guba, 1985: Soutar, 1996).  As time passes this becomes more self-evident.  Many scientists are profoundly ignorant of these issues and fail to take the time to fully investigate the epistemology of science and the history of their own cherished methods. 

New models of investigation have been emerging.  Combining quantitative and qualitative designs have begun to show fruitful outcomes and more comprehensive analyses. Crowd sourcing with standardized data promises to also be an interesting avenue especially when combined with statistical analysis and case study designs.  New Mind has embraced this more comprehensive approach to investigation.  New Mind Technologies presently provides clinicians with highly standardized and uniform methods of data collection, data analysis and outcome assessment.  Each assessment tool was specifically constructed for use with qEEG data and neurofeedback training as well as for the purpose of using it to assess and treat clinical symptoms.  It also provides these assessment tools to measure and track changes in biological, social and psychological dimensions.  Each assessment tool has been specifically validated according to well establish statistical methods and standards and often involved third party expert validation.  Sample sizes used were typically in the thousands and well above thresholds required for statistical validity.  Samples were draw from hundreds of clinics across the US.  Data was acquired using uniform standard assessment and acquisition methods.  Clinicians were trained in the same standard neurofeedback paradigm and instructed in the same uniform methods and NFB protocols.  The New Mind qEEG maps system statistically analyses each map and derives protocols using a standard statistical algorithm.  The clinicians implement these protocols utilizing the same equipment and identical procedures. It is our argument that this uniformity, although not perfect, provides a much higher quality of resource for the collection of clinical data for analysis than is typically found in clinical work today and especially in psychology and counseling.  Members meet three times a week to review cases to compare findings on different disorders and the protocols utilized for training.  Since the data and methods are all standardized, a very high level of group based analysis of data is ongoing.  System strengths and weaknesses are regularly evaluated and addressed.

The amount of resulting data from over a million brainmaps and hundreds of thousands of cases is daunting and continues to grow. Initial exploration of the data and efforts to generate an appropriate data dictionary are still in the initial stages. Part of what follows is just the first peek at some of the intriguing findings that emerge as we explore the data and organize it for a full scale effort at processing its comprehensive potential.

This publication process, then, is a collection of statistical explorations designed to increase understanding of many phenomenon related to qEEG, Neurofeedback and Behavior Assessment.  This is archival analysis and not controlled group designs but the data can be arranged to approximate the quality of CGDs in many cases because of the standardization and uniformity of data collection.  Research is as much about analysis as it is statistics.  With human behavior and biometrics we are dealing with highly complex and non-linear phenomena and we have to avoid thinking in linear progressions.  Much of this data graphs out into nonlinear functions or patterns.  This can be a serious challenge to those not used to thinking in these terms and most clinicians are not academically prepared for this type of thinking.  For instance, the brain is always changing and does so at different rates depending on the general social context and the challenges and levels of stress it presents.  We find that with our measurement tools the average rate of change for those doing Neurofeedback is around 34%, but significantly less than day to day and month to month changes.  This rate may vary and be higher at the outset of training but diminish as time goes by but with several digressions from the trend along the way.  We interpret this as reorganization.  Likewise, the trend screens tend to show movement downward in terms of amplitudes during a training session initially as well as convergence of asymmetry values.  As time goes by, the trends may begin to diverge in normative directions that are reflected in the brain maps in the form of increased general movement toward the norm, or normalization, and decreased movement away from the norm which we interpret as re-organization.  Our goal is to confirm and support or disconfirm these and other observations using archival statistical techniques.  All findings are tentative, like most good science, and will be iterated and updated on a continual basis as we get feedback from other investigators using our data and reviewing our findings.  Emerging controversies will be posted on the New Mind Journal as each of these chapters has been over time.  New Mind invests much of its revenue stream into monitoring the needs of clinicians and providing them with the tools they request to enhance their clinical outcomes utilizing a bio-psycho-social paradigm.  We are as proud to be on the cutting edge of this implementation as we are wary of potential pitfalls and shortcomings that can emerge.  This is however consistent with the history of those people and organizations that take the lead and spearhead novel movements in science or any other endeavor.  We are paving the way with the expectation that even better systems will follow as other seek to imitate our efforts.

Data Analysis Strategies

With so much data one has to be intentional with respect to research questions rather than attempt a data dredging approach. Datamining initially would be useless until we understood more about the types of data we have and how the data varies in quality across different variables.  Additionally, determining what topics to investigate and how to approach those investigations was tantamount in importance. Key areas of investigation drove my initial vectors of analysis. Rather than run all 19 locations on one hypothesis, which highly time consuming and labor intensive, it was more economical to explore relationships between key locations and a variable, then expand the investigations systematically in a theoretical direction to determine if it confirmed predictions based on that theory.  In this first effort I would investigate a relationship between a local area and a symptom and then expand the investigation to see if it extended into a regional or global phenomenon. In the next analysis effort, the findings here will be used to investigate regional and global relationships using all 19 qEEG locations. For instance, is there a relationship between beta and anxiety?  Theory predicts that frontal beta is correlated with anxiety (Engels et al).  Since beta is highly contaminated by EMG, it was important to understand the characteristics of EMG in this dataset if the analysis was to be accurate. The first location to investigate was a relatively EMG free location like Cz.  It was then important to evaluate how free Cz was from EMG by comparing it to T4 which according to the literature was highly contaminated by EMG. Prior research suggests that F3 & F4 should have more of a relationship between beta and anxiety than Cz but it may not be reliable beta in the frontal area if it was very significantly higher in EMG than Cz and similar to T4.  The location T4 could then become the reference for contamination in this particular dataset.  This is similar to reviewing histograms of a dataset inspecting the distribution and descriptives before running a statistical test in order to be sure the data characteristics won’t bias the test.

In selecting the data set we had to take many things into consideration.  For instance several different amplifiers with different characteristics were used to gather data over a period of a decade.  We had 12 location maps and 19 locations maps.  We also had clinics that we knew were very rigorous in their quality control with respect to data collections and others that were less sophisticated.  We had data that was collected serially and data that was collected simultaneously. Was there a difference?  In which neurometric dimensions might they occur?  Some of the data had been collected before we instituted auto-artifacting.  This data was hand artifacted and consequently there would be more variance in the data quality and likely more noise.  The data also had to be formatted in a manner that it was easily manipulated in Excel, which was used for statistical analysis.

When doing the recent update on the New Mind Database, I did considerable exploration of data quality and used extremely high sample sizes to do a comprehensive analysis of Cz to get a feel for the data and how best to run analyses.  Would correlational analysis be useful or would t-tests be more appropriate?  What value would there be in running regression models?  I also had the opportunity to examine the characteristics of all five neurometric dimensions across all locations as a reference to future EEG analysis.  This was helpful because the following analyses are run using raw EEG values rather than standard deviation norms which may or may not be biased.  Raw data tends to be more reliable and this is why most researchers prefer to work with it.

The final data we settled on for this analysis was from New Mind 19 channel maps where data was collected simultaneously.  We compared the serial and the simultaneous data and found no significant difference in the values collected.  By the same token, the CEC and ISI values used were raw scores as well for the same reason.  Typically when comparing EEG and CEC scores, I divided the CEC scores into low vs high or low, moderate and high based on the value range and not the normed cut-off values.  The same was done with EEG values as well when necessary.  Ironically I found higher quality findings between the CEC and EEG scores than I did between the ISI and EEG.  This is interesting because I had previously assumed the CEC was less robust because there was considerable shared variance between the variables.  Since there is considerable shared comorbidity between the dimensions of the CEC and since we did not use principle components analysis to determine their orthogonal dimensions, it may have turned out that they more accurately reflect the global physiological state of the brain.  Of course, the CEC measures dimensions known to have a physiological basis and this is far less true of many of the ISI dimensions.  The ISI, however, was constructed to have highly orthogonal dimensions with high internal consistency as measured by Chronbach’s alphas (Soutar, ).  Yet, it may be that social behavior dimensions should be expected to have less than expected orthogonality and physiological symptoms less orthogonality because they are based in a physiological system which we are proxy measuring in terms of EEG.  The CEC dimensions were further cross-validated with highly established clinical instruments, such as the Beck Inventories, Microcog neurocognitive battery, and the Test of Variables of Attention (TOVAS).  The questions had very high face validity as they were modeled on published and highly employed clinical tools.  In addition I utilized nonparametric statistics for ordinal values, which may be better suited for correlating EEG with assessments.  This will be an important consideration for future instrument constructions. In a sense the New Mind System explores what statistical methods are best suited for comparing EEG and Behavior by its very structure.  There is a lot of noise in these comparisons and parametrics may be better suited to handle it. 

Report Organization

I found few variables that correlated with EEG.  I found that dividing the assessment raw values into categories of high, moderate and low and then running t-tests between EEG values and hi assessment scores versus low assessment scores produces more significant and meaningful outcomes consistent with established theory and prior published research.  This encouraged me to pursue this methodology more aggressively as my investigations progressed.  I outsourced my method to those with greater statistical expertise to confirm the validity of it.

This ongoing analysis so far is listed in several major sections. They sought to answer major questions raised by the literature that would be of special value with respect to clinical interventions. One set of investigations deals with The Interactive Self Inventory (ISI) and investigates patterns within the inventory between dimensions that predict anxiety or depression.  The ISI measures 16 dimensions of approach and avoidance psycho-social behavior that covary with anxiety and depression.  Some initially observed relationships began to emerge as we presented the results to hundreds of clients.  This section on the ISI investigates those relationships.  They primarily involved the relationship between anxiety, depression and perfectionism and competitiveness. Finally, in this section, we confirmed earlier cross-validation findings between the ISI and CEC dimensions of anxiety and depression as well as confirming the prior established cut-off points for clinical vs subclinical manifestations.  The findings in this effort enhanced our confidence in the CEC as both a robust measure of cognitive issues but also anxiety and depression.

Another set of investigations deals with the Physiological Checklist.  Of key initial interest was the relationship between alpha frequencies, depression and physical health, especially the impact of thyroid problems on alpha amplitude and alpha dominant frequency. A second area of interest was delta associated with inflammation and its relationship to health symptoms.  These findings help confirm the impact of general physiology on brain function and EEG activity specifically.  It provides additional information to clinicians regarding sources of confounds to training that are physiological in nature and which may require additional treatment beyond NFB to enhance training outcomes.  It also aids clinicians in identifying physiological issues that are in fact reflected in the qEEG headmaps that are a consequence of those issues. Recent research in the microbiome indicates that inflammatory processes at the physiological level can trigger changes in amino acids and neurotransmitter activity as well as generating pro inflammatory responses in the brain resulting in anxiety and depression (O’Mahoney et al, 2015).

The next set of investigations that followed looked ta the relationship between EEG and Anxiety and what component bands and locations were related to anxiety.  Some unexpected  findings appeared here.  Traditionally, in the field of neurofeedback,  anxiety is associated with elevated beta and there is some good research to support this idea (Engels et al, 2007).  However, research in the last decade or so sheds some doubt on these findings as well as the manner in which clinicians interpret beta activity.  These researchers, using very sophisticated methods, have found the majority of beta activity is EMG activity related to bracing and the complex variance and subharmonic contamination in these frequencies makes statistical comparisons difficult and unreliable (Whitham et al, 2007). Asymmetry appears to be more reliable but can suffer from the same contamination with respect to beta and anxiety.  Fortunately our findings indicate low theta is an excellent correlate of increased anxiety. 

The same type of investigation was done with EEG and Depression with similar surprises.  Fortunately alpha is not contaminated by artifact and yet low alpha is not as good an indicator of anxiety and high alpha appears in clients with both anxiety and depression.  Consequently alpha asymmetry still prevails as the best physiological indicator of depression with respect to qEEG.

Finally I looked at EEG and Cognition, focusing on the dimensions measured in the CEC.  What EEG frequencies and locations seemed to be related to attention, memory and impulsivity?

I strategically investigated those locations suggested by the research literature as well as anatomically based expectations a la the Brodmann MRI research or large scale network findings (Laird et al, 2012; Menon, 2015).

Additionally we looked at midline correlates to these cognitive dimensions in search of a faster method of basic assessment than a full qEEG.


Baggot, Jim (2020). Quantum reality. The quest for the real meaning of quantum mechanics; a game of theories. Oxford. Oxford University Press.

Bohm, David (1995). Wholeness and the implicate order. New York. Routledge.

Engels, A. S., Heller, W., Mohanty, A., Herrington J. D., Banich, M. T., Webb, A. G. & Miller, G. A. (2007).  Specificity of regional brain activity in anxiety types during emotional processing.  Psychophysiology, 44, 352-363

Ioannidis, J.P.A. Why most published research findings are false. PLoS Medicine 2005; 2(8): e124.

Laird, A.R.,  Fox, P.M., Eickhoff , S.B., Turner, J.A., Ray, K.L., McKay, D.R., Glahn, D.C., Beckmann, C.F., Smith, S.M., Fox, P.T. (2012). Behavioral interpretations of intrinsic connectivity networks.  Journal of Cognitive Neuroscience, 1–16. In Press.

Lincoln, Y., Guba, E. (1985).  Naturalistic Inquiry. New York. Sage Publications.

Menon V. (2015) Large-scale functional brain organization. In: Arthur W. Toga, (ed). Brain mapping: An encyclopedic reference, 2, 449-459. Academic Press: Elsevier.

O’Mahoney, S. M., Clark, G., Borre, Y. E., Dinan, T. G., Crvan, J. F. (2015). Serotonin, tryptophan metabolism and the brain-gut-microbiome axis. Behavior Brain Research, 277:32-48. doi: 10.1016/j.bbr.2014.07.027.

Parolini G. (2015). The emergence of modern statistics in agricultural science: analysis of variance, experimental design and the reshaping of research at Rothamsted Experimental Station, 1919-1933. Journal of the history of biology, 48(2), 301–335.

Soutar, Richard (1996).  Methods in the social sciences:  A critique of positivism. Doctoral Dissertation.  Oklahoma State University, Department of Social Psychology. 

Soutar, R. (2008).  The interactive self inventory. Atlanta: New Mind Publications.

Whitham, E.M., Pope, K.J., Fitzgibbon, S.P., Lewis, T., Clark, C.R., Loveless, S., Broberg, M., Wallace, A., DeLosAngeles, D., Lillie, P., Hardy, A., Fronsko, R., Pulbrook, A., Willoughby, J.O. Scalp electrical recording during paralysis: Quantitative evidence that EEG frequencies above 20 Hz are contaminated by EMG. Clinical Neurophysiology 2007; 118: 1877–1888.

Anxiety vs Depression: Finding 1

Anxiety & Competitiveness: Findings 2

Alpha Dominant Frequency & Depression: Findings 3