What's New in Neurofeedback

• Scientists who said ecstasy causes brain damage mixed up the labels
• "Brain training" link to hunger
• Epilepsy Trial Compares Brain Surgery to Drugs
• Babies" Brains Are Fragile Indeed
• Clues to ageing brain

  All links at: http://news.yahoo.com/fc?tmpl=fc&cid=34&in=science&cat=brain_research

I was recently involved in correspondence in which we debated what standards should be applied to QEEG analysis, particularly in its use in guiding neurotherapy practices. In 1994 Frank Duffy and his colleagues published a standards paper on QEEG but for medical use, and this paper loomed large in the minds of nearly everyone I was talking with. The use of the power transformation was also about to be mistakenly codified. The latter aspect has forced me to work up a scientific paper to show (again!, cf Sterman, Kaiser, et al, 1994) empirically and theoretically, why the power transform has no place in QEEG. I promise that this will be the last time I speak about power in this newsletter for some time. Next month I plan an article I call "Sudden Sleep" -- the role of sleep in prenatal development, and how I believe the various epilepsies are "sudden sleep" -- sudden slow-wave sleep, a slipping of the clutch during wakefulness, not a complete loss of arousal as with narcolepsy, but sleep in the waking brain just the same.

But that is next month. What follows is my edited letter on power and Duffy, which was to be published but no more (as it fulfilled its purpose and persuaded those involved to alter their paper accordingly).

I drop you in in the middle of the letter, three paragraphs in:

"When I commented how 10 years had elapsed since Duffy's 1994 publication, it was an indictment (a mild one) against the current authors, not Duffy. Have we no papers of our own to turn to from this century? Have psychologists all fallen silent on QEEG methodology for the past decade? Surely not. The aside was a criticism directed at the authors, to whom I was emailing.

Well, originally I was okay with including Duffy et al (1994) in the paper, but I've thought about it (been forced to due to this correspondence) and now I'm very much against its inclusion. Let me explain why: Incorporating Duffy's recommendations ... is poor politics. It is contrary to the goals of our field and the well-being of many of our practitioners. Duffy et al (1994) is a standards paper of QEEG for medical use, as the following quote makes explicit: "We believe to practice qEEG a physician (typically a neurologist or psychiatrist) must meet the following criteria..." (p. XVII). Do we want to couple our standards to medical practice so overtly? Shouldn't we do as much as possible to emphasize our field's psychological origins and leanings? Not only to fend off the inevitable hegemonic encroachment by medicine, but because it is central to the principles of this field. Neurotherapy is a form of learning....Neurotherapy teaches an individual, implicitly or explicitly, to (re)establish adaptive patterns of physiological function. Clients LEARN their way to better mental health. Learning and operant conditioning can never be wrested away from psychologists. (Or so I hope.) ...Learning is what differentiates psychological therapies (present) from medical ones (absent) -- in fact I believe it is the only thing that differentiates them.

POWER vs MAGNITUDE

Spectral magnitudes can be calculated as readily as power, and in multiple ways: by taking the square root of power, or by digital filters, or by wavelet analysis, or by other functions. Fourier analysis is but a single method of estimating spectral content and it plays no role in operant conditioning of EEG. (The time lag inherent in fourier analysis makes real-time feedback impossible.) Nowadays we rely on digital filters -- simple mathematical functions free of the complexities and requisites of fourier analysis - to provide us with estimates of spectral magnitude. We reserve FFT for evaluation, if even then.

But then why do so many psychologists still calculate spectral power in lieu of magnitude?

Historically, EEG has been analyzed as power spectra due to a conflation, an improper combination, of analysis and measurement. The utility and precision of analyzing EEG data with trigonometric series (DFT, FFT) overrode any reservations psychologists might have had with the power spectrum output (uV2, magnitude squared), notably the loss of the original measurement units (uV). Scientists confused calculation for measurement. The history of science is full of such prolonged misconceptions. The mathematical realm displaces the empirical world; math gets confused as science. The proper measurement of frequency amplitude is magnitude, not power. Given more space I could drive this point home, but let's just try to remain consistent in our own efforts as neurotherapists. Let's evaluate on what we train. Whatever we train is what we should measure beforehand. If we train magnitude, we should evaluate on the same dimension.

As for violating the communicate laws of Algebra (CLA) by using magnitude, nonsense. Ironically I claim power violates CLA. Which is right? This may be John Doe's reasoning against magnitude, how it violates CLA (it doesn't): the FFT involves calculation of real and imaginary components which are squared and summed together to get power. At this point of the calculation, Johnny calls time-out. He's done calculating. Anything more we add to our output that is non-linear in nature will violate CLA, so he says. Let's ignore the fact that at this point in the calculation we can no longer decompose the result into the original real and imaginary values because they were squared and mixed together. This squaring, however, was completed before the whistle was blown -- before Johnny decided the calculation was complete -- so even though it violates CLA in some vein that doesn't count. What counts is the CLA violations after this. Johnny decides power has empirical meaning, so he gets a fresh start to ignore all previous violations of CLA and starts the clock now.

So here I come and perform a square root on the power value to get an estimate of spectral magnitude, thinking that I'm just completing the calculation -- stopping at power was premature in my opinion.

"Foul!" Johnny claims foul, that I am violating CLA -- performing a non-linear operation with the intention of averaging them later.

But look at this: I don't have to follow Johnny's approach to get magnitude. I can also get to this estimate by using a digital filter, which can be as simple as moving averages or moving differences. These don't violate CLA. When I get to a magnitude value -- HOWEVER I GOT THERE -- I now say that the clock is running, all future CLA violations count. Why do I feel justified in doing this? Because the calculation were performed in the context of the physical world - the physical world tells me when to stop. The physical world and psychophysiology. Spectral magnitude is a measurement, not a calculation. in EEG, power is a calculation, not a measurement. So how did I avoid violating CLA by coming at magnitudes "from below" (moving differences, say) but violated CLA by coming "from above" (FFT). Where was the sleight-of-hand. No sleight. It's because MATHEMATICS DOESN'T KNOW WHAT IT IS DOING. Mathematics doesn't have independent empirical reality. Mathematics exists apart of the physical world, probably in a higher dimension than our poor bodies. Mathematics is a tool which which to model reality, but it is not reality. It is contentless, we are content-full.

Just look at the units. uV for magnitude, uV2 for power. We are recording microvolts, uV. Nevertheless, the power spectrum calculation is very useful, and reasonably accurate with stationary signals, but these calculations throw us into a higher dimension, a voltage by voltage field perhaps. (I'm being a bit fictitious.) So we can take this useful, precise calculation technique and rectify its output so that it reflects the measurement we want, so that it reflects the real world, brain or cognitive function.

The square root used to calculate magnitude rectifies the necessary squarings inherent in power spectrum calculation. By taking this square root, we transform a somewhat questionable index of brain function (EEG power) to a meaningful measurement, magnitude, the spectrum of electrical potentials coming off the head. This is just a continuation of what started this process. At the outset of analysis we transformed voltages to spectral voltages because we believed the raw record didn't reflect psychological function very well.

The magnitude spectrum is not the only domain where square-root rectification is commonly performed, in which a contrived parameter is converted into a meaningful one. In stats, we square differences from the mean in order to ignore the sign when summing them, much as we square the real and imaginary components of FFT in order to sum them. We take the square root of this averaged sum to calculate standard deviation, a measure which better reflects the real world than variance. So we do the same for power in EEG analysis. As the god of psych statistical textbooks William Hays, explains: "The square root of the variance for a distribution is called the standard deviation, and is an index of variability in the ORIGINAL MEASUREMENT UNITS" (Ahh, measurement, the world of science -- so close yet so far from the world of mathematics, of calculation)

In QEEG we usually take voltages recorded from the scalp and identify their spectral components. Why? As Barry Sterman is apt to say, "Always ask yourself, what am I trying to do here?" The answer is, because we believe that mental or brain function is better reflected in the periodicities we observe than the raw values we record. We want an index of brain activity that correlates strongly with human thought, and vice versa. So which index is better, power or magnitude?

The unfortunate aspect of this magnitude/power distinction is that we are arguing about it all wrong. We shouldn't be bickering over mathematics calculations, but psychophysiological relevance of each index.

One hundred and fifty years ago, Fechner brilliantly realized that psychological sensation corresponded to the log of a signal's physical magnitude, e.g., loudness increased by the log increase of acoustical energies. It's one of the few principles of psychology to earn the title of law, Fechner's law (1860) or the Weber-Fechner law. Steven's law (1957) followed, but what we want in our final index of brain function is not fully addressed by either. All of these early psychophysicists compared the physical energies with psychological intensities; but we are comparing neural energies with the psychological.

Fortunately the relationship between neural and psychological processes was thoroughly investigated by Vernon Mountcastle, Jr, in the 1960s, and by Kenneth Johnson in more recent times, both of Johns Hopkins. According to their research, neural coding is typically found to be linearly related to psychological perception (cf. Johnson, Hsiao, & Blake, 1996, in a paper titled "Linearity as the basic law of psychophysics"). Mind and brain are linked one-to-one, neural coding to psychology by a linear-to-linear relationship. Therefore our transformations should maintain this mapping. Voltages measured from the scalp should be transformed to magnitude spectra, spectra being the dimension that bests reflects psychological functioning, attention, emotion, memory formation, etc. uV should map to uV, not to uV2. (Conversely, power measures are better when the processes under investigation relate to each other exponentially, such as in wing design and wind shear detection)

So although I'm partial to Fechner and his log:linear relationship between physics and psychology, and would like to implement this in EEG analysis -- to see whether log magnitudes reflect psychology better than magnitudes -- we may already have the best unit to work with, magnitude. Central neural coding to psychology may be log:linear or linear:linear (most likely the latter). Either way, an exponential-to-linear correspondence (that is, power transform) is going in the wrong direction. Log is compression, power is amplification, which is what I've said often in the past.

Fechner: Physical energies ---> Perceptual intensities
Mountcastle: Physical energies ---> Neural coding (e.g., EEG) ---> Perceptual intensities

We must resort to psychophysics to determine which index is best, which indices are even relevant. But psychophysics has yet to give us a definitive answer, or even a complete string of suspects. In fact I doubt no single index of brain activity can ever reflect the complexities and vagaries of human thought and its disorders in its entirety. Neither magnitude, nor power, not absolute nor relative. We should not require all evaluation techniques to include a specified index at this date, certainly not at our present level of understanding. For neurotherapists, this would be especially rash and unwise. Not only do we not know the best mind-brain index at this juncture, we don't know which is the most trainable, most conducive to change through learning. But I do believe that our index of assessment should match the index of evaluation, be it an amplitude measure (magnitude, power) or difference (asymmetry) or a between-site synchrony one (comodulation, coherence, phase) or even a temporal one (dimensional complexity) or wherever our research leads us to in our continual search for mind-brain unity. This is simple parsimony, a basic principle of science.

So the morale of the story is: We must always be careful not to conflate the contentless world of mathematics with the world of measurement, of meaning, the world of science.

Or as Einstein put it, "As far as the laws of mathematics refer to reality, they are not certain; and as far as they are certain, they do not refer to reality."

-DK

News & Reviews NEW BOOKS

Helping Children With Autism Learn: Guide to Treatment Approaches
by Bryna Siegel
Practical guide to treating the learning disabilities associated with this disorder. -www.amazon.com/exec/obidos/ASIN/0195138112/top100

Depressive Rumination : Nature, Theory and Treatment
by Costas Papageorgiou, Adrian Wells
The role of rumination (recyclic negative thinking) is evaluated in development, maintenance and relapse of recurrence of depression. -www.amazon.com/exec/obidos/ASIN/0471486922/top100

Psychiatric Neuroimaging Research: Contemporary Strategies
by Darin D. Dougherty, Scott L Rauch
Demonstrates how the variety of ways neuroimaging can be used for delineating pathophysiology as well as for diagnostic tests and predicting treatment response -www.amazon.com/exec/obidos/ASIN/0880488441/top100

Exposure Anxiety - The Invisible Cage: ...Self-Protection in Autism
by Donna Williams
A very helpful perspective on autism from an insider's point of view. Written like a manual for people dealing with this form of anxiety. -www.amazon.com/exec/obidos/ASIN/1843100517/top100

Fetal and Neonatal Brain Injury: Mechanisms, Mgmt and Risks of Practice
by DK Stevenson, WE Benitz, P Sunshine (Eds)
Survey of neonatal brain injury; from epidemiology and pathogenesis, to clinical manifestations and care, and long-term outcomes. -www.amazon.com/exec/obidos/ASIN/0521806917/top100

Guide to Neuroimaging in Psychiatry
by C Fu, T Russell, C Senior, Dl Weinberger, R Murray
Paperback guide to the increasingly important role neuroimaging plays in differential diagnoses, treatment, and prevention in psychiatry. -www.amazon.com/exec/obidos/ASIN/184184229X/top100

Substance Abuse Treatment for Criminal Offenders: Evidence-Based Guide for Practitioners
by DW Springer, AMcNeece, EM Arnold
Overview of current research useful for practitioners offering treatment to adult or juvenile substance abuse offenders. -www.amazon.com/exec/obidos/ASIN/1557989907/top100

Assisting Survivors of Traumatic Brain Injury: Role of Speech-Language Pathologists
by Karen Hux, Victoria Sutton
For grad students and speech-language pathologists; overviews aspects of TBI including definitions, epidemiology, and mechanisms of injury relevant to speech. -www.amazon.com/exec/obidos/ASIN/0890798958/top100

Functional Neuroimaging and Neuropsychology: Convergence, Advances and New Directions
by David Silbersweig, Emily Stern
Role of neuroimaging in neuropsychology -www.amazon.com/exec/obidos/ASIN/9026518560/top100

Integrated Treatment for Dual Disorders: A Guide to Effective Practice
by Kim Mueser, Robert Drake, Douglas Noordsy, Lindy Fox
Treating mood disorders and other mental illness in the addicted client. -www.amazon.com/exec/obidos/ASIN/1572308508/top100

Quantitative EEG and the Frye and Daubert standards of admissibility. : Shows how Quantitative EEG meets all Daubert standards of scientific knowledge and how science and technical aspects of QEEG also match recent Supreme Court standards of "technical" and "other specialized" knowledge. www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=12784902

Magnetic Seizure Therapy in Major Depression : Magnetic seizure therapy, a novel means of performing convulsive therapy using rapidly alternating strong magnetic fields, may be just as effective as ECT and reduce side effects www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=12865903

Neuroimaging Communality between Schizophrenia and OCD : The caudate nucleus, orbitofrontal cortex, anterior cingulate gyrus and mediodorsal thalamic nucleus may be implicated in both schizophrenia and OCD, though the jury is still out on this. www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=12872207

Emotion regulation during separation procedure: addicted mothers : Emotional expression and regulation appears to be impaired by the stress and maternal disengagement common to cocaine-exposed children www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=12848434

Neuropharmacology of TBI-induced plasticity. : Neuroleptics and other central dopamine receptor antagonists, benzodiazepines and some anti-convulsants may be detrimental to brain injury recovery. www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=12850953

Elevated thalamic and prefrontal regional cerebral blood flow in OCD : OCD symptom severity correlated positively with inferior frontal lobe and right basal ganglia activity. Compulsive behavior was inversely correlated with right thalamus activity. www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=12850251

Quantitative EEG patterns following unilateral stroke: chronic stage. : Left and right hemisphere stroke patients could be differentiated in their EEG by comparing the changes from rest to challenge task conditions. www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=12856476

EEG deficits in chronic marijuana abusers during abstinence : During early abstinence, marijuana abusers show reduced theta and lower alpha rhythm activity during eyes closed compared to controls. These reductions persisted for the entire month of monitoring. www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=12853297

Methylphenidate enhances both intracortical inhibition and facilitation in healthy adults. : MPH enhances both intracortical inhibition and facilitation, which suggests it acts on the motor cortex using a neurotransmitter in addition to dopamine. www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=12734766

Brain sites of movement disorder: neurodevelopmental perturbations. : Neuroanatomy and neurochemistry of ADHD are reviewed with reference to metallic and organic environment and genes. The role of apoptosis and cellular plasticity are highlighted. www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=12832221

• Mental Health Net mentalhelp.net/mhn/yellowpages/db.cgi?signup_form=1
• At Health athealth.com/proform.html
• Neurofeedback Yellow Pages www.skiltopo.com/nfyp/ (mine)
• Find-a-Therapist www.findingstone.com/professionals/find-a-therapist.htm
• Drug Abuse USA www.drug-abuse.com/membership/nonmembers.html
• Psyfidential (psychotherapists only) www.clinicalregistry.com/app.html
• Therapist Finder www.therapistfinder.com/cgi-bin/tforder-therapist.cgi
• PsychWatch www.psychwatch.com
• Who's Who in Mental Health on the Web idealist.com/wwmhw/profiles/new.html
• Mental Health Network www.mentalhealth.net
• AAA Natl Dir'y of M.H. Professionals www.dirs.com/contact/list.htm