What's New in Neurofeedback

• $3 million grant to study non-drug treatments of ADHD, nada for NF -www.eurekalert.org/pub_releases/2001-10/lu-tai100101.php

• Ritalin May Cause Long-Lasting Changes In Brain-Cell Function -www.sciencedaily.com:80/releases/2001/11/011112073546.htm

• Brain Operates Differently In Deception And Honesty -www.sciencedaily.com:80/releases/2001/11/011112073302.htm

In the 1982 film Firefox, the Soviets develop a jet fighter controlled by direct neural links, allowing an individual to pilot his jet by thought, without using his arm or leg muscles. We were seriously working on one aspect of Firefox in the early 90s in Sterman's lab: having the computer monitor a pilot's EEG so it might recognize when he was too overloaded by a task or situation to respond appropriately. Unlike Firefox, refueling during a banked turn at Mach 3 using only your caudate nucleus required at least five more years of funding and the Cold War ended before we could give it a try. But realistic EEG-based human-computer interfaces do exist, as a recent review is Psychological Bulletin describes.

All human communication normally requires some degree of muscular control. However, more and more patients are surviving severe injuries of the brain and spinal cord or neurodegenerative diseases such as amyotrophic lateral sclerosis (ALS or Lou Gehrig's disease) and many of these patients experience locked-in syndrome: an active mind in a paralyzed body. In these patients, communication is extremely restricted or impossible and the inability to communicate often leads to low quality of life and rapid deterioration. Fortunately, the development of EEG-based communication devices can provide a muscle-independent communication channel. A recent review of brain-computer interfaces (BCI) based on EEG analysis describes other uses for operant conditioning of the EEG. Using operant conditioning, locked-in patients are taught to elicit specific EEG components - be they in spontaneous EEG, ERPs, or slow cortical potentials - with which they control a cursor on a personal computer. Like any signal system, using EEG signals for communication requires rapid and accurate generation of the signal. But both rapid are accuracy are relative terms for someone who has no other communicative avenues.

Brain-computer interface research has been underway for 20 years, using two generally approaches: those that require EEG self-regulation and those that do not - essentially a neurofeedback approach and a QEEG assessment approach. Identifying EEG correlates of specific tasks, the assessment approach, includes both temporal and spectral techniques. Some researchers look to evoked potentials, while Pfurtscheller and others are applying FFTs (or filters) to create primitive BCIs. Pfurtscheller examined ERDs during motor imagery and after 60 or so training sessions reports very high success. Keirn and Aunon examined cognitive tasks and were able to discriminate six responses by patients with high accuracy: (1) thinking about nothing, (2) solving a multiplication problem, (3) mentally rotating a complex figure, (4) mentally writing a letter to a freind, and (e) visualizaing numbers being written on a board sequentially.

In the neurofeedback approach, a patient is conditioned into producing specific neural events or rhythms as a signal. One advantage to this approach over the former is that the disease or injury which leaves the patient locked-in often has unpredictable consequences on the EEG, including the development of novel rhythms, that might interfere with detection algorithms (which are normally developed on normal intact individuals). One neurofeedback technique involved electrodes implanted directly into the hand area of the right motor cortex. The patient increases or decreases action potential firing rates to produce a binary signal, a yes or no. This technique yielded only 3 letters per minute, which is not too good but comparable to other BCI rates. In another (familiar) technique, patients increase or decrease mu-rhythm (8-12 Hz, centrally located, wicket morphology) amplitudes. Unsurprisingly, most were able to do so with great accuracy after modest training, though the spelling rates are again slow. Finally, other researchers examined self-regulation of slow cortical potentials, but these turned out to produce very variable (day to day) spelling rates.

I occasionally shout out, sometimes with little provocation, "where's my videophone?" My plaintive cry is a response to AT&T's 1964 exhibit at the World's Fair where they revealed a videophone prototype. For those of us not keeping up on current events, that was 37 years ago!! Surely 30-plus years of monopolistically-funded research should have produced a cheap and reliable videophone. So where is it? Why are they keeping it from us? The microchip underwent about a trillion-fold improvement over the same timeframe -- what's so difficult about sending pictures over phone lines. Jules Verne imagined it a thousand years ago (1863). With somewhat less vigor I might add loudly, "where's my Firefox?" Why hasn't Nintendo or Microsoft hacked through this problem and added a couple of electrodes to their Game-Boy X-box console. Twenty million adolescents are bored with using keyboards and mice to blow away friends in games like Quake and Doom; they want to use only their minds to "frag" cyberportrayals of their bestest buddies.

As it turns out, the greatest obstacles to Firefox are psychological more than psychophysiological. Humans habituate to the repeated stimulations needed in most techniques, thus altering the characteristics of the desired and undesired responses. Current formulations of BCI requires a patient to pay attention to the contents of his or her communication while simultaneously following the generative rules of the communication device. In other words, current BCIs are divided attention tasks to the extreme. In one technique a user has to perform complex visualizations in order to convey his verbal thoughts. Didn't the corpus callosum evolve to stop such cross-coding? I for one could never pat my head and rub my belly long enough to get through a simple conversation. Ultimate success depends on the secondary task (operating the BCI) fading into the background, to be second-nature or automatic, like most human signal systems.

One final obstacle to Firefox is the "instability of EEG frequency bands." Classifying EEG signals (in terms of their cognitive correspondences, for example) turns out to easy if you don't mind always being incorrect. EEG signals are apparent pesky phenomena (or epiphenomena). Like hyperactive chameleons, they change their basic composition rapidly and abruptly. One attempt to overcome this obstacle is the use of short trials, but still much of the complexity remains unexplained and uncorrelated.

The authors of the review conclude that, yes, more neuroscience research is required to better understand the behavior of the EEG, but the really tough issues appears to be something out of Psych 101: that is, the basics of learning. Until these are mastered, BCIs are doomed to failure. And in a similar vien, neurofeedback as a field might also gain by focusing more of its efforts on the process of learning. How we learn is still a mystery. We are closer to the starting gun than the finish line when it comes to understanding the processes of learning.

Related reading:

Kubler A, Kotchoubey B, Kaiser J, Wolpaw JR, Birbaumer N. (2001). Brain-computer communication: unlocking the locked in. Psychological Bulletin, 127, 358-75.
www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?form=6&db=m&uid=11393301

News & Reviews NEW BOOKS

Atlas of Epileptic Seizures and Syndromes
by Luders

Review of brain mechanisms involved in many of the various anxiety disorders. --www.amazon.com/exec/obidos/ASIN/0198522703/top100

Modernization of the Sacred Disease: The History of Epilepsy, 1865-1914
by Walter J. Friedlander

First recognized by the ancient Greeks, epilepsy has a long and distinguished history from St Paul to Dosteovsky. This volume describes its understanding prior to the WWI. --www.amazon.com/exec/obidos/ASIN/0313315892/top100

Affective Neuroscience: The Foundations of Human and Animal Emotions
by Jaak Panskepp

Summary of fundamental neural sources of emotions and a framework for studying relevant systems of the brain. --www.amazon.com/exec/obidos/ASIN/0195096738/top100

Alcoholism Sourcebook
by Karen Bellenir