The thalamic involvement in the generation of the alpha rhythm is being under-valued when looking at the LORETA images of alpha current source generators. The alpha power may come from the sources that LORETA identifies, but the thalamus is intimately involved in alpha rhythm generation, and this is not part of the LORETA image of the sources.

The polarization within the thalamus sets the base frequency of the alpha, but the cortical rhythm requires a complex multi-layer feedback loop from the thalamus to the cortex, and back to the thalamus. Without the cortex, there is a total disruption of the normal spatio-temporal distribution of the alpha wave’s spike trains within the thalamus, and cortical damage often disturbs coherence due to this mechanism.

The thalamus distributes the alpha posteriorly via specific sensory relays, which have a simple return circuit. Like the white matter relay from the lateral geniculate of the thalamus to the occipital lobe’s primary visual areas, and directly back. This thalamo-cortical-thalamic loop is relatively faster than the loop seen frontally. The frontal return circuitry is not simple, but the descending routes are complex and somewhat circuitous, taking more time, and thus it is common for the frontal lobe’s alpha to be at the slower end of the individual’s alpha frequency range. The frontal lobe has a return path through the striatum.

The five divisions of the frontal-striatal pathways are the motor circuit, the oculomotor circuit (from the frontal eye fields), the dorsolateral prefrontal circuit (cortical gating), lateral orbito-frontal circuit (emotive), and the anterior cingulate circuit (emotional and cognitive flexibility). The striatal-thalamic pathways are divided into two descending pathways which both start from the cortex to the head of the caudate and then the putamen, and then this pathway divides between the globus pallidus and substantia nigra, and then these both go to the thalamus. The thalamo-cortical completion of the circuit projects to both the premotor and motor cortex directly.

Not all circuits are simple thalamus-to-cortex-to-thalamus “echoic” returns to the original source…

A cortico-thalamo-cortical projection system exists which originates from the primary visual cortex, relayed by the lateral posterior nucleus of the thalamus, projecting to the suprasylvian visual area (which is involved in highest levels of visual integration and comprehension). This finding suggests that the thalamus modulates transmission of cortical signals from one cortical area to another… the coherence or “connectivity” of the cortex is not cortical-cortical, but cortical-thalamo-cortical.

With maturation, the cortex provides a stimulatory effect on the alpha frequency, raising it to a slightly faster frequency tuning through feedback to the thalamus, but the basic frequencies of alpha are generated by the reticular nucleus of the thalamus providing acetylcholine to the thalamic nuclei, and by the underlying polarization within the thalamus, which is effected by the NE levels from the brainstem, and by fluctuating DC field strength levels in the brain. The other effects are the thalamo cortical transmission times, and an effect of the cortical-thalamic processing time for any given pathway…. Longer time needed for frontal than posterior circuits.

Crudely stated: The frequencies of alpha are set in the thalamus, and the spatial and temporal distribution of alpha are controlled by the cortex, with rhythmic “initiation” (phase reset) done by the DC system’s “modulatory” influence on the AC rhythms of the EEG.

The thalamus can provide rhythms in the range from 3 to 16, with the common range of 8-12 representing an adult group’s “average”. Hyperpolarization of the thalamus slows the alpha, and hypopolarization speeds it up until it desynchronizes at about 16 Hz, and becomes a low voltage fast EEG.

The addition of some GABA (an inhibitory neurotransmitter) easily acquired with the addition of some alcohol will slow the alpha back into a rhythmic pattern. This basic mechanism is the reason alpha-theta training works so well on the low voltage fast EEGs seen so commonly in alcohol addiction.

LORETA may show a generator in the precuneus/cuneus area for the occipital alpha component, and the posterior cingulate for the parietal component (when alpha modulators are identified with ICA analysis and then source localized)… but these localizations miss the full beauty of the real mechanism’s complexity and especially the primary importance of the thalamus.

The thalamus gates our perceptions into “perceptual packets”, with the “thalamic gate” being open during the negative half-wave (up-side of the waveform), and less open during the positive half wave (the downward going half). Two stimuli presented within 75 to 100 milliseconds of each other will be “perceptually synchronized”, or though of as being instantaneously simultaneous.

The alpha frequency is the perceptual sampling rate… how many perceptual packets are evaluated per unit time, with a better semantic or declarative memory function seen with faster alpha frequencies. This is from the work on IAF (individual alpha frequency) from Professor Dr. Wolfgang Klimesch’s lab in Salzburg Austria, with significant contributions from Drs Michael Doppelmayr and Simon Hanselmayr.

The databases have difficulty characterizing alpha frequency tuning issues, with many identifying too much power at a slower frequency (like 7 Hz)… although the power values would be healthy and normal if alpha were only faster (like 9 Hz)… the databases seldom tell you it is merely 2 Hz slow. The normal alpha coherence values, if the alpha is slowed, are seen as hypercoherence, although they are perfectly normal for alpha. Databases that rely on predetermined band’s peak frequency may miss a shift if it exceeds their defined band, and this will miss the mean frequency if the peak is good but the band width has less faster content than slower content.

Faster alpha may cause similar issues (too much 12-15 Hz power or 12-15 Hz hypercoherence) when it is not “too much” of either that is really wrong, just alpha being too fast.

Thus when there are tuning issues, databases often have difficulty characterizing the core issue of tuning. When a tuning issue is noted, the coherence and power values may be “off” according to the database, when the real values are not really abnormal, just that they are too slow or too fast.

Theoretically, these issues may prove to be an area where Z-score training may have difficulty, flagging red herrings of power and coherence… though this is an empirical question that will be answered with time and experience

There recently has been some discussion regarding the use of low resolution brain electromagnetic tomography or LORETA, sLORETA and eLORETA. I felt compelled to make a few comments regarding this since there may be some confusion of how LORETA works and the usage of LORETA as an inverse solution specifically the licensing agreements of the KEY Institute for Brain-Mind Research at the University Hospital of Psychiatry, Zurich.

My intention is to very briefly explain the license agreement so that the end user can be informed. I’ll do so in an informal way by telling the story of the implementation of these methods from my perspective. For a more formal description of the use of LORETA families and some examples you can see a recently written chapter 4 by myself (Sherlin, 2009) in the latest edition of the book Introduction to Quantitative EEG and Neurofeedback edited Budzynski, Budzynski, Evans & Arbarbanel.

In 2000 I had the great privilege to visit with Roberto Pascual-Marqui PhD, the developer of the LORETA family, with my colleague and fellow student Marco Congedo. At this time the LORETA-Key software (Pascual-Marqui, 1994, 1999), had not been widely distributed and utilized in the United States. Marco had significant interest in using LORETA for visualizing brain activity and for exploring newer methods for neurofeedback and had many questions for Roberto. So upon the invitation of Roberto, Marco found funding to travel to Zurich and learn the details from the creator and I happen to be standing in the right spot at the right time. Roberto Pascual-Marqui trained us extensively on how to use his software, named LORETA-Key, which had been already released as free academic software. The LORETA-Key software is a collection of independent modules that the user must run in sequence in order to get from raw EEG to LORETA images.

Upon our return after learning the mechanics of performing an analysis I wanted us to develop a mini-program that would perform the extensive steps necessary in the LORETA-Key software all at once taking care of all possible options in a clear and understandable manner. The analysis was very labor intensive with many batch steps running many different modules. The room for error was great and the user had to be tedious and detailed and still the process would take significant time. With some already existing programming skills Marco wrote a very clean program that would run the LORETA-Key modules after the user had input the necessary data details. This program we called the Workstation (Congedo, 2000). Upon the recommendation of our primary professor and supervisor Joel Lubar, PhD, we formed a company to distribute this program to the larger clinical field of neurofeedback and QEEG providers. The company was named Nova Tech EEG representing our mission and goal of providing “New Technology for EEG”. This very straightforward program, Workstation (released in 2001), was a hit because it allowed clinically oriented users the ability to perform LORETA analysis with a greater ease and utilize this incredible tool in performing and understanding the current source density localization in their subjects and clients, preventing possible errors operating the research-oriented LORETA-Key program. Later we created several other software tools and interestingly revolutionized the development of applications in the neurofeedback industry. Prior to the Workstation (Congedo, 2000) and later the EEG Editor (Congedo, 2001) and finally EureKa! (Congedo, 2005), many of the analysis software were still compiled in the DOS operating system despite many newer developer environments integrating Windows 32 gadgets being released. This made our applications more convenient, professional and provided higher visual acuity in displaying the EEG and the spectral output. It should be noted that since 2005, Nova Tech EEG has also released all of the in house software as freeware, stating unequivocally that our goal was to make inverse solution tools as accessible as possible in the neurofeedback community.

For some time these software along with the original LORETA-Key package were the only ones available for computing LORETA output “all-at-once”. Meanwhile, myself and Marco developed the first adult normative database for LORETA current density in the frequency domain using non-parametric statistics (Congedo and Lubar, 2003). Other third party commercial entities developed programs that would perform LORETA analysis seamlessly calling up the LORETA-Key modules, although there were considerable differences in the applications and the output options. It is worth noting that the directions for seamless integration of the LORETA-Key modules into third-party software have always been clearly explained in the help files of the LORETA-Key software. At some point through the development of third party software the KEY Institute and Pascual-Marqui felt it necessary to restrict the licensing of the software due to apparent misuse. In the June 2007 release of the LORETA-Key software there was a new screen following the previous End User License Agreement (EULA), screens which required the user to agree not to misuse the LORETA software. This screen reads:
This software computes LORETA from scalp electric potential differences (time domain EEG/ERP) or from EEG cross-spectra (frequency domain). One particular very incorrect usage is to cheat LORETA with the input.

Examples of misuse:
1. Inputting scalp electric potential spectral powers will not output LORETA (current density) spectral powers.
2. Inputting scalp electric potential square roots of spectral powers will not output LORETA (current density) square roots of spectral powers.
3. Inputting scalp z-transformed-maps will not output LORETA (current density) z-transformed-values. The three previous invalid inputs to LORETA violate the mathematics and the physics underlying all computations.

Furthermore, they violate any correct usage of statistical analysis. Some more technical details can be found in:
1. For time domain computations: Pascual-Marqui RD: Review of methods for solving the EEG inverse problem. International Journal of Bioelectromagnetism 1: 75-86, 1999.
2. For frequency domain computations: Frei E, Gamma A, Pascual-Marqui R, Lehmann D, Hell D, Vollenweider FX: Localization of MDMA-induced brain activity in healthy volunteers using low resolution brain electromagnetic tomography (LORETA). Human Brain Mapping 14: 152-165, 2001. See text and equations on pages 154-155 therein (Pascual Marqui, 2009) .

It wasn’t long after a wider spread distribution of the LORETA software and use in the community of QEEG and neurofeedback that the natural question arose of if the current source density of interior cortical areas could be operant conditioned in the same manner as the scalp neurofeedback was being conducted currently. This was actually our first concern as announced during a workshop at ISNR by Joel Lubar,  Marco Congedo, David Joffe and Leslie Sherlin (Lubar, Congedo, Joffe, & Sherlin, 2001). The workshop was the starting point for a 3 year project which would be Marco’s dissertation where he demonstrated and verified that in fact the deeper structures could be trained using “LORETA feedback” (Congedo, 2003). This work was published the following year in IEEE Trans. in Rehabilitation Engineering and Neuronal systems (Congedo, Lubar and Joffe, 2004) and is still today a pioneering study in multi-channel neurofeedback. Dr. Lubar’s lab continued to pursue these techniques with additional validation studies (Cannon et al, 2007, 2009). Currently the method is used in several other universities and is becoming available in several neurofeedback systems.

Newer methods were developed by Pascual-Maqui in 2002 and it was named standardized LORETA or sLORETA (Pascual Marqui, 2002). This new implementation had to its advantage the ability to localize test point sources with zero localization error in the absence of noise, which had not previously been accomplished. Since my goal here is not to distinguish the difference in the methods I will skip over these technical issues. It will suffice to say that despite the name, from a mathematical point of view sLORETA is very different from the old LORETA method, and much more accurate. The sLORETA-KEY software was released by Dr. Pascual-Marqui once again as free academic software, but now there were new licensing agreements. As compare to the previous package there was an additional clause that, “This free academic software package is intended for use in research… If you install and use this software, you have then accepted the “license agreement”, and from then on, by law, clinical use and commercial use are strictly forbidden (Pascual Marqui, n.d. b)” This means that any use of the sLORETA package as published and distributed by the KEY Institute cannot be utilized outside the scope of research only. The use of this package cannot be used for creating clinical reports. Not less important, the sLORETA-Key software modules computing sLORETA images cannot be called upon anymore by third party software. The days of utilizing third party software for automatic processing of LORETA data were over to protect the software against misuse and profiteering.

It wasn’t long after the original publications of Pascual Marqui (Pascual Marqui, 2002) on the new sLORETA method that others began to independently replicate the sLORETA work (Congedo, 2006; Congedo et al. 2006; Wagner et al, 2006). This, or any other, independent replication does not fall under the same licensing restrictions because the independently replicated sLORETA algorithms are generated independently, that is, they do not use the implementation in the sLORETA software package distributed as freeware from the KEY Institute. One implementation of this sLORETA transformation was in the EureKa! Software. It made use of the same head model of the old LORETA-Key software, making the sLORETA computation as straightforward as LORETA computations!  Due to the fact that the head model implemented in the new sLORETA-Key package is not open, the older LORETA-Key viewer had to be utilized rather than the newer sLORETA viewer but the computations using sLORETA in this way are completely legal and ethical. The end user must comply with the EULA of the EureKa! Software but this does not violate the KEY Institute’s EULA. This is only one example but I am aware of at least one other legal and ethical such use by a third party and that being the Mitsar Company of St. Petersburg, Russia. In fact the Mitsar Company has recently implemented their independent replication of the sLORETA algorithm in their neurofeedback software BrainTuner. In a recent study that is still under data examination, the clinical reports of using this modality with the sLORETA implementation was overwhelmingly positive (Ozier, Whelton, Mueller, Lampman, & Sherlin, unplublished dissertation in progress).

Recently, we have even successfully replicated and implemented independently the newer eLORETA method, which represents a further improvement over sLORETA in the EureKa! software.

So in summary, the end user must be aware and of course operate within the EULA of the software they are implementing whether for analysis or feedback. I might add that all of the implementations that I am aware of require that the end user understands and agrees that he/she is ultimately responsible for the use of the software; and that the distributor is not responsible and cannot be held responsible for any acts outside the intended use.

References

Cannon R., Congedo M., Lubar J.F., Hutchens T. (2009). Differentiating a network of
executive attention: LORETA neurofeedback in anterior cingulate and dorsolateral prefrontal cortices. International Journal of Neuroscience, 119, 404-441.

Cannon R., Lubar J.F., Congedo M., Thornton K., Towler K., Hutchens T. (2007), The
Effect of Neurofeedback Training in the Cognitive Division of the Anterior Cingulate Gyrus, International Journal of Neuroscience, 117(3), 337-57.

Congedo, M. (2000). Workstation (Version 1.0). Knoxville, TN: Nova Tech EEG, Inc.

Congedo, M. (2001). EEG Editor (Version 1.0). Knoxville, TN: Nova Tech EEG, Inc.

Congedo, M. (2003). Tomographic Neurofeedback; a new Technique for the
Self-Regulation of Brain Electrical Activity. University of Tennessee, Knoxville.

Congedo M., Lubar J.F. (2003), Parametric and Non-Parametric Normative Database
Comparisons in Electroencephalography: A Simulation Study on Accuracy, Journal of Neurotherapy, 7(3/4), 1-29.

Congedo, M. (2004). sLORETA zero-localization error as seen in a point spread functions: an animation Retrieved April 29, 2009, from http://www.lis.inpg.fr/pages_perso/congedo/sLORETA.htm

Congedo M., Lubar J.F., Joffe D. (2004), Low-Resolution Electromagnetic
Tomography neurofeedback, IEEE Trans. on Neuronal Systems & Rehabilitation Engineering, 12(4), 387-397.

Congedo, M. (2005). EureKa! (Version 3.0). Mesa, AZ: Nova Tech EEG, Inc.

Congedo M., Lotte F, Lécuyer A. (2006), Classification of Movement Intention by
Spatially Filtered Electromagnetic Inverse Solutions, Physics in Medicine and Biology, 51, 1971-1989.

Congedo M. (2006), Subspace Projection Filters for Real-Time Brain
Electromagnetic Imaging, IEEE Transactions on Biomedical Engineering, 53(8), 1624-34.

Congedo M., Joffe D. (2007), Multi-Channel Spatial Filters for Neurofeedback. In
“Neurofeedback: Dynamics and Clinical Applications “, (Ed) Evans J., Haworth Press, New York,

Lubar, J. F., Congedo, M., Joffe, D., & Sherlin, L. (2001). LORETA 3-D Neurofeedback, Normative Database and New Findings. Paper presented at the Society for Neuronal Regulation.

Ozier, D., Whelton, W., Mueller, H., Lampman, D., & Sherlin, L. (unpublished). Comparing the efficacy of thermal biofeedback and sLORETA neurotherapy as interventions for chronic pain., University of Alberta, Edmonton.

Pascual-Marqui RD, Michel CM, Lehmann D. (1994). Low resolution electromagnetic
tomography: a new method for localizing electrical activity in the brain. International Journal of Psychophysiology, 18:49-65.

Pascual-Marqui RD. (1999). Review of Methods for Solving the EEG Inverse
Problem. International Journal of Bioelectromagnetism, 1:75-86.

Pascual Marqui, R. D. (2002). Standardized low-resolution brain electromagnetic
tomography (sLORETA): technical details, Methods Find. Experimental Clinical Pharmacology, 24(D), 5-12.

Pascual-Marqui, R.D., Esslen, M., Kochi, k., and Lehmann,D. (2002b). Functional
imaging with low resolution brain electromagnetic tomography (LORETA): A review. Meth. Findings Exp. Clin. Pharmacol., vol. 24C, pp. 91–95.

Pascual-Marqui, R.D., Esslen, M., Kochi, k., and Lehmann,D. (2002c). Functional
imaging with low resolution brain electromagnetic tomography (LORETA): Review, new comparisons, and new validation. Jpn. J. Clin. Neurophysiol., vol. 30, pp. 81–94.

Pascual Marqui, R. D. (2009). LORETA: do not misuse Retrieved April 29, 2009, from http://www.uzh.ch/keyinst/NewLORETA/Misuse/Misuse.htm

Pascual Marqui, R. D. (n.d. a). Limitation of use: Retrieved April 29, 2009, from
http://www.uzh.ch/keyinst/NewLORETA/sLORETA/04Slor.html

Pascual Marqui, R. D. (n.d. b). LORETA: do not misuse Retrieved April 29, 2009, from
http://www.uzh.ch/keyinst/NewLORETA/Misuse/Misuse.htm

Sherlin, L. (2009). Diagnosing and Treating Brain Function through the use of Low
Resolution Electromagnetic Tomography (LORETA). In T. Budzynski, H. K. Budzynski, J. Evans & A. Abarbanel (Eds.), Introduction to Quantitative EEG and Neurofeedback, Advanced Theory and Applications (2 ed.): Elsevier.

Wagner, M., Fuchs, M., Kastner, J. (2004). Evaluation of sLORETA in the presence of noise and multiple sources,”. Brain Topogr., vol. 16, no. 4, pp. 277-280.