EEG BIOFEEDBACK APPLICATIONS FOR THE MANAGEMENT OF ATTENTION DEFICIT-HYPERACTIVITY DISORDER
- Joel F. Lubar, Ph.D.
- University of Tennessee
- Southeastern Biofeedback and Neurobehavioral Institute
- Knoxville, TN. 37919
- Phone: 865-584-8857
- FAX 865-584-8721
- Joel F. Lubar, Ph.D.
- University of Tennessee
- Southeastern Biofeedback and Neurobehavioral Institute
- Knoxville, TN. 37919
- Phone: 865-584-8857
- FAX 865-584-8721
ADHD is a pervasive lifelong disorder of great significance. It affects perhaps as high as 10 percent of the population depending on which review article one reads and how it's classified and how it's characterized (Whalen & Henker, 1991). ADD/ADHD overlaps with other disorders. Attention Deficit Disorder (ADD) comes in three forms, inattentive,hyperactive and combined. Unfortunately it's been classified as Attention Deficit/Hyperactivity Disorder (ADHD) but in the new 1994 DSM-IV ADD and ADHD are classified more separately and there are even different subtypes. The term hyperkinesis, an older term which is still used represents Attention Deficit with hyperactivity. The overlap of both ADD & ADHD is as high as 70% with specific learning disabilities (LD) (Hynd, Marshall & Gonzales, 1991). There are over 250 different kinds of LD (Hooper & Willis, 1989). Also in many individuals ADD/ADHD and LD overlap with conduct problems (Barkley, 1990). The Attention Deficit Disorder without hyperactivity is sometimes referred to as ADD-, to use Barkley's (1990) classification, and it tends to overlap more with anxiety disorders as well as learning disabilities. The Attention Deficit Disorder with hyperactivity, ADD+, also overlaps with learning disabilities but it is also more concurrent with oppositional defiant disorder (ODD).
ODD is diagnosed behaviorally when the individual, often a child, purposely does not follow rules, does not see the significance of them and therefore doesn't believe they are important. If this is not corrected or treated, it can evolve into a conduct disorder. Conduct disorder is more severe and more premeditative and even has more purposeful destructive behavior than the ODD. If the conduct disorder is not brought under control, especially in late adolescence and early adulthood, it can evolve into a much more serious disorder known as antisocial personality disorder. Many individuals who have antisocial personality disorder spend part of their time in prison and some with conduct disorder do also, but when there is only oppositional defiant disorder, they usually have not reached that point in the legal system. There is a great deal of overlap of this whole constellation with some other disorders such as depression and anxiety Some ADD/ADHD children have movement or tic disorders ranging from benign tic disorders to the more severe Tourette's Syndrome. Occasionally they may have seizure disorders and there is a higher incidence of substance abuse in the ADD and ADHD population than in the non-ADD/HD population.
There is evidence, from Kenneth Blum and colleagues (Noble, Blum, et al., 1991) that there is a strong genetic component in this whole constellation of problems located on chromosome 11. They've identified the series of alleles that are associated with and ADHD that lie close to another set that are associated with alcoholism, depression, and Tourette's Syndrome. Blum has designated a constellation of these disorders as the "reward deficiency syndrome ". This may be one of the reasons why these disorders sometimes occur together. He also postulated that many of the addictive behaviors including gambling, and risk taking are symptoms of reward seeking. In ADD/HD the reward valence of stimuli habituates rapidly. We find in our evaluation of children with ADD/HD that it's very common to be able to follow it back one to three generations in the family. In doing an intake evaluation, one should ask to what extent the disorder is seen in other family members particularly in males since it is between 3 and 5 times as common in males as in females (Lubar, 1990).
There are some considerations about criteria for ADD/HD listed in DSM IV (1994) that require elaboration. The main deficit as far as behavior is concerned is difficulty in task completion, particularly if the task is perceived by the child, adolescent or even adult as irrelevant for them. So, for example, if you tell the child, I want you to go home and do the 50 math problems on page 100, the child will probably not get the work done. A common excuse is "there is no homework tonight" or "I lost the assignment on the way home", or if the assignment is brought home, there'll be a pitched battle between the child and parents over each math problem. The child will plead "I want to go out and play", "I want to watch TV." The parent may say, "If you do the next problem, I'll give you a nickel, if you don't do the next one, I'll send you to your room." This kind of behavior goes on and on and then the assignment may get done, very sloppily and then the next day, it doesn't get turned in. The child either doesn't return it, destroys it, or loses it.
Consider this scenario: The mother is called into the principal's office because the child is not getting the assignments done and the principal says "your child has attention deficit disorder" and the parent says "that's not possible. My child will come home and sit in front of a TV set for 4 hours, play Nintendo or play in the arcade with his friends for hours without stopping. He also likes to build models all the time, beautiful models, and does it very meticulously, how can you say my child has an attention deficit disorder?" The answer is that attention deficit disorder is extremely selective, it is not global. ADD children are not necessarily distractible, they are only distractible for something that they don't see the point in doing. This is only one aspect of ADD/HD behavior, the other is that in regard to the child's life space, the child is living now, everything is right this minute. If you say to the child, "now look, you've got to get this homework done, you have an exam this week, if you don't get these problems done, you're going to have difficulty getting through the exam." The typical response is "oh that exam isn't till Friday, why do I have to do it today, Monday? I'll do it Thursday night." The child does not process the consequences of behavior. These are some of the aspects of ADD/HD that are not really well pointed out in the current DSM-IV.
Common tools used to determine diagnoses and comorbidities include extensive interviewing and family history, medical history, and, in some cases, detailed medical workups involving EEG, neurological scanning (MRI or PET) techniques, bloodwork, and sometimes, genetic studies. Psychodiagnostic tools such as the MMPI and metrics designed specifically for children can be useful as part of the assessment of possible thought and affective disorders.
Neuropsychological measures are particularly useful for assessing organic brain dysfunction and Specific Learning Disabilities. These techniques include but are not limited to the Halstead-Reitan battery, Luria Neuropsychological battery, Wechsler Intelligence Scale for Children (WISC-R) or WISC-III) or the Wechsler Adult Intelligence Scale-R (WAIS-R), Woodcock Johnson Psychoeducational Evaluation-Revised (WJ-R), and other Neuropsychological and achievement measures.
For the behavioral-observational assessment of ADD and ADHD, there are excellent rating scales available. These include Barkley (1987) Conners' Parent and Teacher Questionnaire (Conners 1969), Hawthorne, Achenbach Child Behavior Checklist, the Child Activity/Attention Profile of Edelbrock, Home and School Situation Questionnaire, and others in development. Many of these are checklists filled out by parents or teachers and often include items based on the DSM categorization of ADHD, One should employ the same checklist before and after an intervention to determine the extent that an intervention has affected the behaviors described.
The selection of treatments that have been used as standard are the stimulant medications more for children who have Attention Deficit with hyperactivity. For example, the use of amphetamine sulfate (Benzedrine) was first prescribed by Bradley in 1937. These are sometimes supplemented by tricyclic antidepressants, alpha blockers and in rare cases, anti-psychotic medications. There are no medications for children with only learning disabilities, such as dyslexia, dysgraphia, dyscalculia, receptive disorders or expressive disorders not in the constellation of ADD or ADHD.
Two main hypotheses have been developed to explain how stimulants may be useful to these children.
A) the low arousal hypothesis of Satterfield and Dawson (1971) and further elaborated by Satterfield, Lesser, Saul and Cantwell (1973).
B) the Noradrenergic Hypothesis developed over years and elaborated by Zametkin et al. (1990)
Basically, the two hypotheses propose that children with ADHD experience low arousal. This low arousal comes about as the result of decreased impact of sensory stimuli in all modalities, acting upon the central nervous system mechanisms for sensory integration and resulting in the child engaging in stimulus seeking behavior. Self-stimulation and object play are primary features of the hyperactive (hyperkinetic) syndrome (Lubar and Shouse, 1976; Lubar and Shouse, 1977 and Shouse and Lubar, 1978.) Children with hyperactivity often exhibit intense but brief interest in new stimuli., rubbing objects against their bodies, smelling them, tasting them if they can. They often engage in other stimulus-seeking behaviors, including excessive movement, picking up one object after another and then examining it, spinning around, running from place to place within a room. If in a room with minimal stimulation, they will often fall asleep after a short flurry of activity.
This stimulation-seeking behavior may be based upon decreased noradrenergic activity, particularly in the brainstem reticular formation and possibly in the basal ganglia as well. Heilman et al., (1991) proposed that in many of these individuals, both adults and children with this disorder, a defect in response inhibition and particularly in the nigro-striatal frontal system exists, more on the right side of the brain than the left side. Motor restlessness may reflect not only frontal lobe dysfunction but also possible impairment of the dopaminergic system as well. However, Malone. et. al.(1994) proposed dual neurochemical mechanism which is somewhat different in that they view ADHD related to increased presynaptic release of norepinephrine particularly for the right hemisphere and decreased left hemispheric dopaminergic transmission. They further postulate that methylphenidate acts to increase the dopamine available and to inhibit the release of the excessive norepinephrine. The dual effect of the medication is to decrease the overstimulation due to the norepinephrine and to increase attention which is mediated by the dopamine system.
We also think of arousal 1) Mediated by connections from the brainstem reticular formation, which receives inputs from all the sensory modalities except olfaction 2) the transmission of this information to the diffuse thalamic (reticular) projection system, and 3) the basal forebrain. These input systems, in conjunction with the basal ganglia and the cerebellum, which is involved in programming the output of the motor cortex, are all affected by decreased noradrenergic activity. Norepinephrine is produced by a very important brainstem nucleus, the locus coerulus. This nucleus has extensive projections with these mesiolimbic-striatal and forebrain systems. The neuroanatomy of these systems is not only complex, but the resulting neurochemical dysfunctions and/or neuropathology associated with them is also quite complex. The latter include both movement and affective disorders. As will be discussed later, these neurochemical and neurophysiological abnormalities are reflected in EEG measures, in regional brain metabolism, and event-related potential measures for individuals with ADD/HD.
The other common treatments for ADD/HD are behavior therapy using complex schedules of punishments and rewards, (Barkley, 1990, ) cognitive behavior therapy, , individual psychotherapy and family systems therapy. Regardless of what treatment is employed, whether it's neurofeedback, behavior therapy, drug therapy, and all the possible combinations with and without family and psychotherapy, there is one sad reality, ADD and ADHD are not curable at this point in time. That doesn't mean however, that they can't be dealt with. Diabetes is not curable but one can become asymptomatic with proper management, live a normal long life and not suffer any significant consequences, if the diabetes is not too severe. There are many other disorders that can be treated to the level of becoming asymptomatic. The hope for ADD and ADHD is that one can experience long symptom-free periods and maybe permanently become relatively asymptomatic, except under extreme conditions of demand or stress. But for the present it appears that ADD/HD is a pervasive, lifelong, neurophysiologically based disorder.
The disorder of ADD/HD manifests itself in the form of poor educational achievements, inappropriate emotional behavior, and all of the problems seen in terms of compliance, academics, etc., but the problem is not a primary emotional disorder, an educational disorder, nor a behavioral disorder; it is neurophysiological. This means that ADD/HD is reflected in the way in which the child, adolescent or adult is processing information. The way in which his or her brain is processing input is affecting the way the individual perceives the world and therefore responds to it. This disorder also involves a decrement in motivation or persistence of effort to be able to complete tasks which are perceived as boring, irrelevant or to difficult. Our point of view in working with this disorder is that if we can change the underlying neurophysiology, perhaps we can effect a more long term change in this disorder. This is the reason why behavior therapy has not been particularly successful. It works very well while it's being done. You can set up complex schedules of time outs, and reinforcements for behaviors and you say to the child, "Okay, you have 25 long division problems, if you get them done, we're going to go out to your favorite restaurant for dinner." Half an hour later, they're done perfectly. The next night, "I'm not going to do the homework." What is the parent going to do now? There is very little carryover of the seemingly potent reward. The hope is that if you continue to do the behavior therapy, from age 6 to somewhere around age 16, finally the message will get through and the child will develop insight as an adolescent and will be able to self-integrate the task-appropriate behaviors without continuous reinforcement or punishment. But very often, that's not the case.
Stimulant Medications & Guidelines
The stimulant medications are powerful for many children with hyperactivity, and 60 to 70% of children on medication show a very marked improvement of the hyperkinetic component of the syndrome and may perform better educationally. Currently, three primary stimulants are used in attempting to restore neurochimical balance. Dextroamphetamine is commonly used in younger children. Methylphenidate (Ritalin) and pemoline (Cylert) are commonly used with children from six on into adolescence and even adulthood. Adults and children who have extreme difficulty with organizational and planning skills or impulsive behavior may respond positively to small amounts of tricyclic antidepressants such as imipramine or norpramine. Imipramine has more side effects than norpramine. For some children, when impulsive behavior also becomes very aggressive, clonidine, an alpha adrenergic blocker, is sometimes helpful. Between 60 and eighty percent of children with ADHD show varying degrees of positive response to these medications when they are administered properly. In the ideal response cases, the child no longer exhibits all components of hyperactivity nor any great attention problems. Patients who have this ideal response can stay on their medication regimens for extended lengths of time without any significant adverse side effects. Other patients respond partially to the medications and may exhibit a variety of side effects, including anorexia, mood changes and varying degrees of sleep disturbance. Ritalin sometimes produces an almost totally flattened affective response in some children, and they are described as "zombie-like" by parents or others who know the child. This kind of response would warrant a change in medication or a non-medication alternative. .
The twenty-five percent of children who do not respond to medications are usually not good candidates for neurofeedback, though there are exceptions, particularly children who had allergic, toxic or medical reactions to the medications. Of the twenty to twenty-five percent of children who do not respond to medications well, there are a number of possible reasons:
1) serious side effects, such as gastrointestinal disturbances, 2) urinary problems , 3) seizure activity, 4) headaches, 5) increased tic activity ( particularly a possible problem with Ritalin), or , 6) unacceptable changes in affect and emotional behavior .
One of the main problems with stimulant medications is that as soon as the medication is out of the system, the ADD/HD behaviors return. If one is using a very short term medication such as methylphenidate or Ritalin, the poor behaviors will reappear within 3 to 6 hours after the last dose is given. Then you have to give the drug in the morning, again at noon, and maybe another small dose in the afternoon just to get through the homework. Use of a time-release form is one solution to this, but this is not always as effective as separate doses. If you don't give the medication during weekends, or on vacation, ADD/HD behaviors return. Parents often despair in taking their hyperactive children on vacation because they say they are unmanageable. Some of the experts in the field such as Wender (1985) believe that it doesn't matter if they're 8 or 80 years old, you keep them on medication all the time every single day, even during vacation periods. Since the disorder is not curable, the idea is to alter the distribution of neurotransmitters in the brain and maintain the appropriate behaviors pharmacologically. Unfortunately, pharmacologic treatments do not seem to have a long term carryover. If you've had a child on Ritalin for 10 years, and you try to phase the child off you may find that the behaviors will continue to be maintained for approximately 3 months and then the child starts to go downhill and then maybe you have to put the patient back on medication (Whalen & Henker, 1991).
The other fallacy that was in the older literature is if the child has hyperkinesis or is hyperactive, if you don't do anything, it will go away (Swanson & Kinsbourne, 1979). What will happen in many cases is that the hyperactivity does decrease, but the attention deficit becomes worse and so then you have an adolescent or a young adult who has ADD+ characteristics. They start failing their courses in high school, or they get thrown out of school. They can't go to college or during college, they can't get through and have to leave. If they get a job, they very often get fired because they don't meet the employment demands, or they get in power struggles with peers or superiors. I've had ADD adults who come in and talk about their marital situation saying, "I've been married 4 times," and then I ask, "Do you have an idea of what some of the problems may have been that caused this to happen? They'll say "Yes, I got bored with the relationship, it wasn't interesting anymore. I found somebody else." Their ability to stay in any particular setting is very difficult. They will often go from job to job, from location to location, sometimes from marriage to marriage, exemplifying the aspects of this pervasive disorder. Sometimes children with ADD\HD will perform well up to about grade 3 or 4, if they're very bright, and then they will become worse and worse until eventually they just can't make it as the demands of school become greater.
Subtypes of Individuals with Attention Deficit/Hyperactivity Disorder
Children and adults with ADHD often experience comorbid difficulties including Oppositional Defiant Disorder, Conduct Disorder, Anxiety Disorder, substance abuse, depression, and in some cases, tic or movement disorder. To further complicate matters, between 50 and 70 percent of children with ADHD also experience learning disabilities. Barkley (1990) has already pointed out that children with ADHD may experience primary difficulties in terms of attention, impulse control or impulsiveness or hyperactivity. He has also pointed out that ADHD usually can be diagnosed before the age of 7 and is sometimes also associated with distractibility, poor planning, poor rule following behavior, and inability to engage in sustained activities for any significant period of time. Whereas some individuals are described as hyperactive, others are described as hypoactive or lethargic, easily bored, disinterested in most play activities and extremely disinterested in school-related activities. Children who have specific learning disabilities, may also be classified in terms of specific areas of disability such as subtypes of reading disability. Flynn and Deering (1989) and Flynn, Deering, Goldstein, and Rahbar (1992) have described two subtypes of dyslexics-- , diseidetic and disphonetic. The differentiation is based on a classification by Boder. The question arises whether the different subtypes of ADHD with or without learning disabilities, conduct disorder, oppositional defiant disorder, depression, etc., can also be characterized as somewhat neurologically distinct and if they are neurophysiologically distinct, it is important to tailor the Neurofeedback treatment for the neurophysiology that is presented by these different subtypes.
Recently Amen, Paldi, Thisted (1993) have been evaluating ADD using brain SPECT imaging (single proton emission cerebral tomography). SPECT scan is a technique similar to a PET (positron emission tomography) scan. SPECT scan measures cerebral blood flow and indirectly brain metabolism and has the advantage that it involves less radiation than PET scan, is less expensive, and allows for a longer data collection period while the tracer is being distributed through the nervous system. There appears to be a high correspondence between the findings reported by our group using quantitative EEG and topographic brain mapping and the SPECT scan findings. Both the quantitative EEG and the SPECT scan findings indicate that for the most part ADD/HD is a disorder which involves deactivation of the prefrontal lobes (reduced blood flow,) particularly during an intellectual or academic stress task. In other words, the more the child tries to concentrate, the poorer they perform on academic tasks and the more they show slowing or decreased cortical metabolism particularly in frontal and central areas. One of the advantages of SPECT and PET scan over EEG is that they allow us to examine what is happening in subcortical structures as well as changes in the cortical surface in terms of blood flow and cortical activity. The Amen et al. study of 54 children who met the DSM-III-R criteria for ADHD were compared with 18 controls. Their work then identified the following subtypes:
(1) frontal lobe deactivation without other findings: This is the more classic ADHD individual who responds best to stimulant medication if pharmacotherapy is the treatment chosen. These individuals show marked decreased frontal lobe metabolism, particularly during academic tasks as compared with a resting eyes open baseline. EEG studies show excessive slow activity .
(2) A second subtype is represented by temporal lobe dysfunction. These individuals will often have deep temporal lobe dysrhythmias or epileptiform activity, and still experience an Attention Deficit Disorder which seems to respond best to anticonvulsants sometimes combined with Ritalin if pharmacotherapy is chosen.
(3) A third subtype is ADHD with homogeneous cortical suppression. These individuals tend to respond to antidepressants in combination with Ritalin. The main difference between this subtype and the first subtype is that individuals with this type of disorder have more widespread cortical deactivation than the stimulant responders alone. Very often these individuals will respond well to tricyclics such as Desipramine or Norpramine.
(4) A fourth subtype are individuals who actually have increased activity in the anterior medial aspects of the frontal lobes. This refers to an anatomical region known as gyrus rectus or the region which lies in front of the septal region. These individuals are also hyperactive, distractible and restless. This pattern is also associated with Obsessive-Compulsive Disorder and overfocusing to the point of being unable to complete a variety of tasks hence experiencing a type of attention deficit in which there is an inability to shift attention and excessive overattending to often irrelevant details.
(5) There is a fifth subtype which is characterized as ADHD with hypofrontality at rest but normal frontal activity with the challenge of intellectual stress. These individuals are also distractible, impulsive and restless but there are also extremes of more oppositional-defiant behavior than some of the other subtypes. They also tend to respond well to methylphenidate (Ritalin).
Overall, the Amen et al. study found that in 87% of the cases, there is prefrontal lobe deactivation with intellectual stress (65%) or decreased activity in the prefrontal lobe at rest (22%). In our own work with quantitative EEG, we have some preliminary evidence that suggests that individuals that show more slowing in the right prefrontal lobes experience difficulties with impulse control and often act inappropriately in social situations but show good organizational skills. Children with left prefrontal lobe slowing tend to have poor organizational skills but are more appropriate in social settings and children with right posterior parietal slowing tend to be of the lethargic, hypoactive category and often complain that everything is "boring".
In a more recent study by Henriques & Davidson (1991), it has been shown that individuals experiencing reactive or monopolar depression have more left frontal alpha than right frontal alpha activity whereas nondepressed individuals tend to have more right frontal alpha activity than left frontal alpha activity. The overlap of hypofrontality in terms of theta activity and perhaps excessive left frontal alpha activity might represent a type of individual who experiences ADHD and depression as well. To make matters a little more complicated, it is well known from the research of Peniston and Kulkosky (1989) that many individuals experiencing alcohol addiction and certain other types of chemical dependency experience increased beta activity and decreased alpha and theta activity. This pattern was shown originally by the work of John et al. (1988).
One final point regarding the neurology of ADD\HD and associated disorders is that many of the executive functions including planning, judgment, appropriateness of behavior in social settings, is mediated not only by the prefrontal cortex but by the frontal pole and the area underneath the frontal pole known as the orbitalfrontal cortex. The orbitalfrontal cortex has extensive projections to the amygdaloid complex and other structures within the limbic system which have been known for nearly a century to strongly mediate emotion and motivational states. The orbitalfrontal cortex is one of the organizing cortical areas for the control of emotional behavior through the limbic system and ultimately through the output systems of the autonomic and skeletal motor pathways. It is most unfortunate that we cannot reach this area of the brain in terms of neurofeedback strategies. The orbitalfrontal cortex is for all practical purposes inaccessible via surface EEG; however, it is a region of the brain which is very sensitive to medication effects and perhaps individuals in which SPECT and PET scans show abnormalities in this region are really better candidates for pharmacotherapy than a neurofeedback approach. In contrast, those patients that show excessive EEG dysfunction in the superior frontal cortex or the midline central cortex are much better candidates for neurofeedback interventions which can train individuals to produce more optimal patterns of EEG activity.
The Neurophysiological Hypothesis underlying the Neurofeedback Model
If ADD/HD are associated with neurophysiological dysfunction, particularly at the cortical level and primarily involving prefrontal lobe function, and if the underlying neurophysiological deficit can be corrected, the child with ADD/HD may be able to develop strategies and insights (paradigms)
which non affected individuals already possess. Using these paradigms, the ability to organize, plan and understand the consequences of inappropriate behavior is facilitated. This results in a stronger carry over of the effectiveness of time-outs, rewards and other behavioral approaches. Medication needs may actually decrease if by changing cortical functioning we can also show a change in brainstem functioning, such as brainstem or cortical event related potentials or measures of sensory integration. The long term effects of Neurofeedback have already been documented (Tansey, 1990; Lubar, 1991,Lubar 1995).
In conclusion, this technique, which leads to normalization of behavior, can lead to normalization of neurophysiological dysfunctional behavior in the ADHD child, and can produce long term consequences
in social integration, academic achievement and over-all life adjustment.
A Rationale For Neurofeedback
A primary problem the ADD/HD individual experiences is difficulty completing long, repetitive tasks which are perceived as boring. This includes much of the activity which occurs in the school setting, including homework. Another characteristic is difficulty accepting that socially appropriate behavior is governed by rules and rule-following behavior are essential to progress through the developmental milestones needed to become a fully functional adult. Individuals who cannot follow these rules may ultimate develop Oppositional Defiant Disorders, Conduct Disorders , or in the worst cases, Antisocial Personality Disorders. Not all ADD/HD children are inevitably on this pathway as a consequence of their behaviors, but there is a higher incidence of these problems particularly in the ADHD population. Poor motivation and rapid habituation to the reward properties of stimuli are additional common characteristic of the ADD/HD population. For that reason. toys and play activities do not remain interesting for very long for these children. Thrill-seeking behavior is sometimes also seen in ADHD adolescents.
Stimulant medications such as Ritalin (methylphenidate) , Cylert (pemoline) or Dexedrine (amphetamine sulfate) are primarily used to increase arousal or the impact of stimulation. However, there is very little clear evidence that these medications, particularly Ritalin, change cortical function.(Swartwood,1994). Parents commonly complain that the child, even with the medication, does better in school, concentrates better, is less hyperactive, but still has trouble getting work done, still has difficulty following the rules and understanding why, when they act inappropriately, that the behaviors have to be corrected. These latter problems are caused because of altered cognitive functioning and to a great extent, problems with frontal lobe functioning. The frontal lobes, as discussed previously, are the executive portions of the brain and they have decreased metabolism in ADD adults and probably in ADD children, plus decreased fast EEG activity and excessive slow wave activity. With Neurofeedback we are attempting to change cortical functioning as well as arousal functions. We attempt to establish a cortical EEG signature template which responds similarly to that of non-ADD/HD individual in the situations and circumstances which cause problems for ADD/HD adults, adolescents and children. Many neurofeedback trainers, report that once neurofeedback is implemented with this population, when they misbehave and the behavior is pointed out to them, they can readily move to correct it and they don't repeat the inappropriate behavior as much once they understand why the behavior was not acceptable. Essentially, ADD/HD children trained with neurofeedback often experience long-term transfer of training to home and school settings because they begin to cortically function more like non-ADD individuals in that rule-following behavior increases, inappropriate social behaviors decrease, impulsiveness decreases, and motivation levels improve. Our own earliest studies in the 1970s
showed the carry over of effects into school settings (Lubar and Shouse, 1977).
Quantitative EEG, Brain Mapping and Electrode Placement
A question raised by the literature on QEEG is whether one needs an extensive neurological assessment before neurofeedback training or is brain mapping sufficient? In our work, we are basing our electrode placements on studies of blood flow and imaging. We have placed most of our electrodes bilaterally, halfway between CZ and FZ (FCZ), and halfway between CZ and PZ (CPZ). The reason for using these particular placements is that based on 19 channel topographic brain mapping, these locations represent the areas where the highest ratios of theta to beta activity are seen which we believe, at this time, to be the most relevant neurophysiological correlate especially for ADD. The success rate with this particular set of placements has been found to be very high by our group and many others who are using this protocol for training and because the signal has less artifact there. I wish to encourage both clinicians and researchers to consider other sites based on the individual neurophysiology that is presented and to do the training either with bipolar or referential electrode placements over those sites where EEG assessments indicate that abnormal activity is more focal and particularly if it is not along the midline. If the greatest amount of dysfunction is in the orbital frontal cortex, the best locations for recording this activity would be FP1 and FP2; however, eye roll and blink and frontal EMG artifacts make these sites virtually impossible to use. If one were treating depression, perhaps placing electrodes at either F3 and F4 or C3 and C4 or some intermediate point between F3 and C3, and F4 and C4 would be acceptable in terms of changing alpha ratios between left and right hemispheres. These same locations utilizing a combination of feedback to change alpha distribution and to decrease the amount of theta activity and increase the amount of beta activity might be ideal for working with ADD/HD individuals who also are depressed This protocol might be the analog for individuals who are placed on a combination of stimulants and tricyclic antidepressants. Individuals who show the slow lethargic pattern which is sometimes localized on the right posterior parietal region around P4 might be trained in that location with electrodes perhaps at C4, P4, the bipolar placement or to change the ratio of theta-beta activity in the right parietal region with respect to the left parietal region in order to equalize distribution of this activity between the two hemispheres. Individuals with significant hyperactivity would benefit considerably from training the sensorimotor rhythm (SMR) with bipolar placements at C1-C5, or referential placement at C3 with linked ear references. This should be done for 20-30 sessions before using the midline placements for theta-beta training to increase attention, focusing and executive functions. The combination of SMR and theta-beta training may take as long as 40-60 sessions but the long term results are well worth the effort.
One final point which is of relevance in regard to electrode placement is the question of generalization. We have seen in more than a dozen cases in which training was carried out along central cortex with electrode placements either at CZ, PZ or halfway between FZ, CZ (FCZ) and halfway between CZ, PZ (CPZ) pre-post training changes in the theta-beta ratios in all 19 standard electrode locations. Although the changes are greatest in the central cortex, they can even be seen in occipital and temporal regions where small improvements are noted. This indicates that even with a central location electrode placement, the effects of the neurofeedback are experienced in widespread cortical areas. The issue of whether there is generalization from locations off the midline to all other locations is unknown at the present time and represents a researchable question. At the present time, I do not recommend electrode placements that employ summation of activity from multiple sites. The algebraic sum of EEG activity from multiple sites bears no resemblance to the activity seen at each site individually and may result in
feedback which might be either irrelevant or perhaps even injurious.
Filter bandpass considerations
The question also arises as to which band passes are appropriate. We find that in working with children, typically we'll reinforce increasing beta activity between 16 and 20 Hz. and decreasing theta activity between 4 and 8 Hz. This has worked for the majority of patients. However, above the age of 14, we find that there are many individuals that show excessive alpha activity and lack of alpha blocking. With these individuals we often train to decrease activity between 6 and 10 Hz., high theta low alpha, and increase beta activity between 16 and 22 Hz., 16 and 24 Hz., or even 18 and 24 to 26 Hz. These decisions are based on spectral analyses to determine whether there are areas in which there is decreased beta activity or increased alpha and/or theta activity during task specific conditions. It is probably difficult to train activity above 28 Hz. with conventional feedback systems because of the overlap of the EMG spectrum and the possibility of training individuals to increase certain types of muscular activity in order to achieve reinforcement for these higher frequencies.
The essential requirements for neurofeedback instrumentation include very accurate signal processing with the ability to show changes of one tenth of a microvolt , for threshold-setting purposes. Software should always allow the clinician or researcher to observe the raw signal and how it is processed so as to produce reward or inhibition of the signal and to observe the relationship between changes in events in the raw EEG and the feedback. Excellent systems employ both analog and digital processing. Such systems have been described in detail previously (Lubar and Culver, 1978 and Lubar, 1989).The absolute requirement of seeing the raw signal and exactly how it is processed and what in the raw signal activates the reward and inhibits is paramount for fine tuning of the protocol in terms of the duration, amplitude, and frequency of the relevant signals. The displays should be so complete that one could use a signal generator with a 0.1uv sensitive voltage divider to evaluate the accuracy and response time of the system. Without this type of display there is no way other that blind faith to know if your feedback device is providing accurate analysis of the raw signal and appropriate feedback. Regulatory agencies and third party payers have the right and power to demand accountability in this regard.
Activities During Training Sessions
The neurofeedback trainer must simultaneously meet specific physiological criteria. These include: 1) increase either sensorimotor rhythm (SMR) between 12 and 15 Hz (especially if hyperactive) or 2) beta activity, often defined as 16-20 Hz, while at the same time, not producing theta, movement or EMG activity. The goal is to increase the SMR or Beta activity, duration , prevalence and, if possible, amplitude, while simultaneously decreasing the amplitude and percentage of theta and gross movement, as detected by EMG activity. The subject or patient has to be very alert, but also relaxed and quiet to do this.
Staying on task, often documented by poor performance on continuous performance tests such as the Gordon Diagnostic system or the TOVA (Test of Variables of Attention,) is one of the main problems of children and adults with ADD/HD. ADD/HD children are able to play Nintendo or computer type games for long periods of time because the games change the demands to new tasks every few seconds. But there is no significant transfer to homework or other situations which are long, continuous, repetitive and/or boring, such as school related tasks. Neurofeedback essentially involves engagement in a continuous performance type task under an altered EEG state which is like a non ADD person's EEG for significant periods. Training the children this way. to stay on task for extended time periods, in the "normalized EEG state seems to work, enabling transfer of the skill to situations away from the neurofeedback setting.
Recording EEG and Artifact Considerations
Recording EEG through the human scalp is not an easy task in one respect. Electrodes are placed on the scalp, and so they are recording activity from the skin, such as skin potential changes or electrodermal activity, then underneath the skin, there are several layers of muscle producing EMG. Next you have to record through bone, hence there is further attenuation of the EEG, especially the higher frequency components. As a result, what we are struggling with primarily are non-EEG artifacts. There is EMG artifact, which is part of the signal, starting from as low as 12 Hz. and ranging to greater than 500 Hz., although most of the EMG spectrum is between about 30 and 150 Hz. This is why a lot of EMG machines for both EMG recording and feedback will have a frequency range of 30 to 150 Hz. The other thing that we are concerned about is EKG, electrocardiogram, because this is a very powerful artifact that is often seen in the EEG. The EKG artifact is more prevalent as we get older because the EEG amplitude becomes smaller. EKG is occurring from the electrodes that are picking up activity from underlying pulsating blood vessels in the scalp. The reason EEG becomes smaller as we get older is because the skull becomes thicker as do the meninges and the connective tissues between the skull and the scalp, and the skin layers. Also, the amplitude of neurological activity decreases.
Trying to get a perfect multichannel recording of 19 or even 32 channels becomes very difficult in adults. One might think children are going to be very difficult to record from but I've obtained some of the most ideal recordings from children because their EEG signal is so large that these artifacts are less of a problem. If you're recording from frontal locations in addition to these others you also pick up ocular artifacts, blinks, and electroretinal activity. If you examine the EEG, you sometimes see these large excursions, that look like slow wave activity, however, it's often simply an eye roll. Eye movements and blinks are in the delta range from 0 to 4 Hz.; but they're seen in the anterior half of the scalp. If recording from any leads from the lateral surface, temporal, lateral frontal, parietal, posterior, specifically F3, T3, T5,P3,F4,T4,T6,P4, invariably we pick up muscle activity from the masseter muscles, temporalis and with every swallow. Every time that happens, even to a small degree, we get a burst of very fast artifact activity which appears in the EEG that can create problems in terms of signal processing, sometimes leading to feedback for non-EEG activity. If you record from the posterior half of the scalp, then we pick up EMG activity, from the occipitalis, trapezius, and supraspinal muscles.
The question is how do you get "clean" EEG from an individual? We spend a lot of time, especially if we're going to do quantitative topographic brain mapping, trying to clean up all these signal problems, with relaxation, or through positioning of the head. Sometimes you have to hold the head of the individual or use pillows to stabilize their head, so that you can minimize these artifacts. If you record from the central locations along the midline, such as FZ, CZ, or PZ, then you can get relatively pure EEG.
We have chosen a particular set of locations for training with our ADD and ADHD population which are far less prone to artifact. One location is halfway between CZ and FZ, and the other location is halfway between CZ and PZ. These two points represent the two locations on the head where there is less EEG to artifact than anywhere else.
Referential Versus Bipolar Recording For Training
Another point relative to feedback is an unresolved issue, concerning the difference between what is called referential or monopolar recording and bipolar. When you employ referential recording, what you're asking is what is the activity at a particular point on the head as compared with some point that is electrically neutral? The way we usually do this is to place a reference electrode on each ear lobe. Essentially you are examining the relationship between activity at CZ or some other point while making the assumption that the ears are neutral or act as a ground so any activity we see in that EEG recording, represents the activity at that point. This approach is good in terms of knowing what's happening at each point and differentiating the activity at each point from any other point. However, the energy is maximal right under the electrode and it decreases with the square root of the distance from the electrode so that the activity surrounding CZ, the central location falls rapidly as we move away from CZ. If we have a montage where we have one electrode with ear references that we use for feedback, you have to realize that what you're really going to be influencing is a limited area around the electrode and probably activity occurring more than a centimeter from that electrode will have relatively little effect. There is then the theoretical question of whether training at that one point can have very much effect on behavior, although it is being done and some clinicians are getting good results with that type of recording.
Since referential recording for feedback purposes is very prone to artifacts, another alternative is bipolar recording. Bipolar recording depends upon the algebraic subtraction of activity between two points. From an electrophysiological point of view or neurological point of view, we don't know exactly what that means, yet nevertheless training that type of activity seems to work very well for many different disorders. Bipolar recording has two advantages: one is that we are actually influencing activity over a larger area and that of course seems to be advantageous. Maybe it affects the spread of activity because if we're looking at activity of apical dendrites in layer 1 of the cortex, since that activity is linked to many cortical areas, it's going to have more of an influence on other regions of the brain.
Another advantage of bipolar recording is that anything which is occurring in both channels at exactly the same time the same amplitude and completely in phase is subtracted out. This is called common mode rejection. If you clench your jaw, there will be a burst of activity in temporal electrodes and if it occurs at the same time and in phase at both electrode locations, much of this EMG will be subtracted out. Certainly EKG is a problem when electrodes are placed over blood vessels in different scalp areas. EKG is usually rejected or decreased as common mode activity. Eye rolls and eye blinks will also be reduced. However, if one is interested in pure alpha conditioning and if you have your electrodes at 01 and 02, the alpha which is in phase and coherent between these locations would be factored out. So that's a consideration that might lead to the use of a monopolar referential montage in alpha signal detection and neurofeedback training.
The concept of phase and coherence may be of considerable importance for what we are trying to accomplish. Phase is measured in degrees of lead or lag between the signals of the same frequency. For example if two 10 Hz sine waves rise and fall in each cycle at exactly the same time then they are in phase and the phase angle between them is zero degrees. If one wave reaches a maximum when the second wave reaches zero voltage then they are out of phase by 90 degrees. If one wave reaches its maximum positive voltage when the second wave reaches its minimum voltage then they are 180 degrees out of phase. Coherence is a measure of the degree to which the phase angle between two waves of the some frequency (not amplitude) remains constant over time, regardless of the phase angle between the two waves. Coherence is similar to a correlation coefficient and in this sense ranges from 0-1.0. If two waves of equal amplitude and frequency are recorded with a bipolar electrode montage and these waves are in phase then the resultant amplitude will be zero due to these being common mode signals. However if these two waves are 180 degrees out of phase then the resultant amplitude will be twice that of each wave measured separately. Since a considerable amount of neurofeedback concerns either increasing beta or decreasing theta activity in conjunction with ADD/HD or teaching individuals to increase alpha and theta activity for treatment of addictive behaviors (Peniston & Kulkosky, 1987), it is important to determine whether bipolar or referential recording and training is appropriate.
Figure 1 presents a scatter plot showing the relationship between the relationship of theta to beta activity in 32 cases for both
Insert Figure 1 About Here
referential and bipolar recordings. Each individual represented in the figure had 3 electrodes placed on their cranium at CZ and halfway between FZ and CZ and halfway between CZ and PZ. The bipolar and referential recording was done with the eyes open while the individual was fixating on a photograph placed 24 inches in front of them. The photograph was 2 X 4 inches in size and because of its size, produced minimal eye movement. Referential recording was done first in half the subjects and then the bipolar recording, each recording lasting for 90 seconds. The order was reversed in the other half of the subjects.
The figure clearly shows that there is a strong linear equivalent relationship between bipolar and referential theta-beta ratios. In other words, individuals tended to have approximately the same theta-beta ratios whether the recording was done using the bipolar montage or the referential montage. This has some very specific implications for neurofeedback training. First, the study indicates that the amount of common mode rejection in the bipolar recording is not of significance in that if there was considerable rejection of common mode EEG signals, then the ratios for bipolar recording should be very different from those for monopolar recording. The second implication is that the decision as to whether to use a referential montage at CZ or bipolar montage with electrodes placed halfway between FZ and CZ and halfway between CZ and PZ should be based on the behavior of the individual. If we are working with a child who is very hyperactive, then we would want to have common mode rejection of movement artifacts and EMG. In this case bipolar recording would be preferable. If the individual is primarily ADD without hyperactivity, then referential recording with an electrode placed at CZ could be used. However, linked ear references must be employed. It is interesting that in the many groups that are using the theta-beta paradigm for Attention Deficit Disorder, those employing monopolar and bipolar recording montages are obtaining approximately equal results. If using a bipolar montage, inter-electrode distance must be measured for every session and should remain plus or minus 1 mm., since signal magnitude depends on inter-electrode distance. Accurate electrode placement is essential if reliable conclusions about the learning taking place are to be drawn.
Figure 2 shows the alpha-beta ratio relationships for referential and bipolar
Insert Figure 2 About Here
recordings with the eyes open. Notice there is also a linear relationship but there is much more scatter. The reason for this is as follows: if for a particular individual with a bipolar montage, the two signals are in phase, there will be considerable common mode rejection. In that particular case, the referential ratio will be much higher than the bipolar ratio. If the individual's alpha activity is out of phase a bipolar recording will result in larger amplitudes of alpha activity and therefore the bipolar ratios will be higher than the referential ratio. There are individuals on the graph who represent both conditions. The decision then to use referential or bipolar recording in training a decrease in alpha-beta ratios should be based on measures obtained both ways to determine whether common mode rejection is of importance. And then, one should go with the greatest ratio for a decrease and the lowest ratio to train for an increase.
Insert Figure 3 About Here
Figure 3 shows a different representation of this type of data. In this figure the actual percent power of theta, alpha, and beta activity is represented for all subjects. It can be clearly seen that there is virtually no difference between the actual amount of theta and beta activity obtained with either montage. In the case of alpha, however, there is greater alpha power obtained in the bipolar montage than in the referential montage. This would tend to indicate that most alpha activity recorded along the midline between FZ and PZ is out of phase leading to a greater amplitude in the bipolar recording condition. Table 1 summarizes the relationships between a variety of bipolar and referential measurements in terms of their correlation coefficients. Even though for some measurements the correlation between referential and bipolar measurements may be as high as .9, it is still important to employ both types of measurements in determining for a specific individual whether training should be undertaken with referential or bipolar electrode placements. One final point and that is that the data presented here is only valid for central locations. New databases would have to be obtained with reasonable numbers of subjects using both types of recording for any other locations on the cranium. Therefore, decisions as to whether to record bipolar between locations 01 and 02 or referentially at OZ or bipolar for locations F3 and F4 versus referential recording at FZ or similarly recording between C3 and C4 as opposed to a referential recording at CZ is unknown at this time and will await development of databases similar to the one shown here for the central midline locations.
A third approach which hasn't been done systematically but it's theoretically possible is instead of subtracting the activity from two electrodes, adding the activity. So for example you might place an array of electrodes at F7, F8, C3, C4, and CZ, and train the summated activity from all of these at the same time. From a theoretical point of view, we do not know what is being recorded but the idea behind that kind of a montage is try to train large areas of the brain at one time. A few people such as Fehmi (1980) and Tansey (1991) use a 1 x 6 centimeter electrode soaked in saline. It is very difficult to assess what is being added, or subtracted with such an electrode. Therefore regardless of what type of recording you use, always observe the raw sign