When is Theta Alpha?
|
The world is becoming more customized every day, yet many of us continue to use off-the-shelf principles when we assess or train individuals. In his first report Hans Berger (1929) characterized the "waves of the first order" in human EEG -- which for the sake of brevity he designated alpha. The "alpha rhythm" had large sinusoidal waveforms at a rate around 10 cycles per second against a background of smaller waves, "waves of the second order" (i.e., beta). Alpha waves were pronounced in posterior regions during eyes closed resting states, and diminished markedly upon opening the eyes.
This is how the alpha rhythm was defined 70 years ago. Today we have a superior definition: Alpha activity occurs between 8 and 13 Hz, or is it between 8 and 12 Hz, or perhaps 7 and 13 Hz, or 7.81 and 14.06 Hz, or 8 and 15 Hz (Etevenon et al, 1990, Ray & Cole, 1985); you get the picture. What is more disturbing than the different intervals are their boundaries, which are artificial, a product of ease of communication and the limits of one's analytical technique. The alpha rhythm is defined as the dominant frequency rhythm in the resting state, the frequency band that dominates the spectral density distribution. At this scale the brain rarely uses integers. Perhaps we would do better to keep the names simple but not its designation. Klimesh (1999) developed a simple designation strategy; he identifies an individual alpha frequency (IAF) from each subject, then defines bands relative to this peak. Lower alpha is from 2.5 Hz below IAF up to IAF, and higher alpha runs from IAF to IAF plus 2.5 Hz. The theta band is also defined relative to IAF. Obviously the plus or minus 2.5 Hz is artificial and is one of those compromises plentiful to psychophysiology, based on empirical data and ease. Some subjects will have a narrow dominant frequency, others might hit the mark exactly. Perhaps a refinement of the formula is needed, a mixture of percent attenuation and topography. This might produce a truly customized dominant frequency bandwidth. From there we build towards the other bandwidths of interest. Eventually we may find out that restricting our analysis to such unique ranges can improve the reliability and validity of our conclusions.
Above is a statistical distribution of the dominant frequency for 124 subjects. The peak frequency in five posterior sites (P3, Pz, P4, O1, & O2) during eyes closed resting baselines were calculated. Each subject provided up to seven replications (though no more than four per day). As can be seen above, the dominant frequency spreads across the 7-15 Hz range, though a bin of 8-12 Hz captures 95% of the data. Still, important individual information is lost in this depiction. For instance, to what extent does the alpha rhythm varies across this sample in both peak and shape? A review of the subjects finds a factor of two in alpha width: some subjects exhibit 40% of all alpha magnitude in a single 1-Hz (peak) frequency band whereas others never exhibit more than 20% in 1 single Hz band inside the dominant frequency range (see below).
The large variance in peak and width begs the question: why do we use a large band to assess dominant frequency activity? Would it not be simple to calculate an IAF, even with a one-channel EEG system? The figures above are from a NORMAL ADULT population in the ALPHA RHYTHM. These three properties align to produce the most regular and consistent recording possible in human EEG. We are all aware of frontal slowing in ADHD children. Some argue convincingly that high theta activity in such a population is actually misnamed; it is merely an immature manifestation of the alpha rhythm (the child's dominant frequency). So 4-7 Hz may be theta for some and alpha for others.
Here are two subjects who were excluded from the 124 sample for obvious reasons. Both show a lower peak frequency. One is a 3 year old child and the other at 65 entering second childhood, at least electrophysiologically. And these outliers are likely normal in this band -- at least for their group. Obviously we should reevaluate how we define frequency bands for clinical and scientific investigations. Especially if we are evaluating and training clinical non-adult populations. And by non-adult, we may mean anyone under 30 and over 50, according to Neidermeyer's model (1993). Cerebral maturity, I guess, is a difficult state to achieve and to maintain (see figure).
Two months ago I was faced with a challenge. A 3 year old was about to start neurofeedback training. Given his young age and condition his EEG rhythms were immature, but the question was how immature: where was his SMR band located, for instance? The literature suggested a dominant frequency around 5 to 6 Hz, possibly higher in some toddlers. If the SMR was adjacent to the dominant frequency, as it is in normal adults, it would fall in line with some Russian research which indicated an SMR rhythm of 6fromof the child, I had little data to go on. One strategy often used in this field is to start training at the adult range (here, for SMR, 12-15 Hz) and adjust (lower) the band if the client's response is not what one expects. General knowledge in lieu of an assessment was better than nothing, so this strategy was used. Well, it was one of those experiences that can make or break a parent's confidence. The very first session changed everything about the child. Unfortunately it made everything significantly worse. He was trained at 12-15 Hz, site C4. Too high. After a year of holding his urine while he slept, that week he started urinating in his sleep, he became aggressive at his Special Ed school, so much so that they threatened to kick him out. The training somehow disinhibited him along both physiological (urination) and behavioral (impulse control & aggressive) domains. Needless to say, the parents of this child were not happy. Time to adjust the training band. And rapidly. But there was no way to gauge the results readily enough, given the child's problems. It was like trying to listen to a volume change of a jet engine during takeoff: was that 145 decibels or is he improving and now putting out only 143 dBs? Fortunately we were able to finally record a Q on the child. And as it happened he fell asleep in the chair -- and his EEG was soon populated with sleep spindles. Sleep spindles are produced by the same reduction of motoric input as SMR, by similar neuronal pathways, so right there, bingo! I had an easily quantified measure of the toddler's SMR rhythm. Here was the appropriate band to start training on.
Above is 5 seconds of raw EEG data during the initial stages of sleep.
As the spectral analysis shows, his SMR rhythm is 10-12 Hz. Perhaps a sleep recording for all clients undergoing SMR training might be helpful. It could even be achieved with a one-channel system. Place the client in a dark room, a comfy chair, and with an electrode at Cz or Fz. Wait for spindles and simply count the cycles. It's not even a Q in this EEG assessment. Statistical descriptions may be powerful and accurate tools, but rarely as powerful as individual data. - David Kaiser, Ph.D. |
An Odd Kind of Fame: Stories of Phineas Gage
Cognitive Neuroscience of Emotion
Acute Stress Disorder: A Handbook of Theory, Assessment, and Treatment
Learning Disabilities: Implications for Psychiatric Treatment
Pain: What Psychiatrists Need to Know
Improving Treatment Compliance: Counseling and Systems Strategies for Substance Abuse and Dual Disorders
Integrative Neuroscience: Bringing Together Biological, Psychological and Clinical Models of the Human Brain
Chronic Fatigue Syndrome, Fibromyalgia, and Other Invisible Illnesses