QEEG-guided, z-score and LORETA neurofeedback

1. Introduction

Version followed: the International Society for Neurofeedback & Research (ISNR) position papers on QEEG standards (Hammond et al., 2004) and on standards of practice (Hammond et al., 2011) for the clinical frame; Thatcher, Lubar & Koberda (2019) for live z-score and LORETA z-score training as its developers describe it; Congedo, Joffe & Lubar (2004) for the first LORETA (tomographic) neurofeedback pilot; Coben, Hammond & Arns (2019) for evidence status. Where these differ, the difference is named.

This entry covers three overlapping ways of choosing and running EEG neurofeedback from a multi-channel brain map rather than from a fixed, symptom-based protocol:

• QEEG-guided neurofeedback: a quantitative EEG (QEEG) is recorded, compared with an age-matched normative database, and the deviations are used to individualise an otherwise conventional protocol (amplitude training of the kind in 045–049). ISNR describes guiding training as the main purpose of QEEG for neurotherapists, whereas neurology and psychiatry use QEEG mainly as a diagnostic aid for medical pathology [source: Hammond et al., 2004].
• Z-score neurofeedback (ZNFB): the live EEG is converted, moment by moment, into z-scores against a reference database (for power, asymmetry, coherence, phase and similar measures), and feedback rewards movement of the selected measures toward the database mean. Developed by Robert Thatcher's company Applied Neuroscience, Inc. and licensed to several device makers, as the developers state [source: Thatcher, Lubar & Koberda, 2019].
• LORETA neurofeedback (LNFB): low-resolution electromagnetic tomography, an inverse-solution method that estimates the cortical sources of scalp EEG, is used to feed back estimated activity (current density) in a chosen brain region or network, sometimes as a z-score. The first pilot trained the anterior cingulate (cognitive division) in three people [source: Congedo, Joffe & Lubar, 2004]. LORETA itself dates from 1994 and sLORETA from 2003, per the developers' account [source: Thatcher et al., 2019]; attribution of the original LORETA method to Roberto Pascual-Marqui is standard in the field [uncertain: original paper not verified].

Core idea of change (as developers state it). Feedback is operant conditioning of EEG activity. Thatcher and colleagues argue that rewarding moments when selected measures are close to the reference mean, and not rewarding unstable extreme moments, reinforces more stable network function ("homeostasis") through synaptic plasticity. They reject the description of the aim as making a brain "normal" and say the same logic serves peak performance [source: Thatcher et al., 2019]. This mechanism is the developers' hypothesis; an independent review found "virtually no evidence" that 19-channel ZNFB is effective or superior to standard protocols [source: Coben, Hammond & Arns, 2019] (see 2 and 4.17).

Who delivers it. Licensed health professionals with independent practice rights who have EEG/QEEG training, which ISNR states includes psychologists and is not limited to physicians; unlicensed technicians only under full-time on-site licensed supervision [source: Hammond et al., 2004; Hammond et al., 2011]. A psychotherapist can deliver it after neurofeedback and QEEG training (certification routes in 4.16); a psychotherapist without that training works alongside a provider (4.13–4.15).

Regulatory status (as sources state it). ISNR states that the US FDA classifies EEG biofeedback and monitoring equipment as Class II medical devices sold only to licensed providers [source: Hammond et al., 2011]. Coben et al. note that LORETA z-score neurofeedback is not FDA approved and that its registration appears to fall under the exempt biofeedback category [source: Coben et al., 2019]; the developers cite an FDA exemption for battery-powered amplifiers as evidence of low risk [source: Thatcher et al., 2019]. The QEEG has not been endorsed by the American Academy of Neurology (AAN) or the American Clinical Neurophysiology Society (ACNS) for psychiatric use [source: Wikipedia, citing Barrett, 2015; secondary]. UK, EU and other jurisdictions: [uncertain: not verified].

Evidence status in one line. Experimental: ZNFB and LNFB "cannot be recommended as a first line treatment option" [source: Coben et al., 2019]; the Dutch Psychiatric Association guideline advises against neurofeedback in general for children with ADHD [source: NVvP, 2020]; NICE NG87 does not list neurofeedback among its recommended non-drug options for ADHD per a secondary summary [source: Jones, 2019, National Elf Service; NICE text not verified, uncertain].

2. Framework

No single published manual sets out phases for QEEG-guided, z-score or LORETA neurofeedback. The sequence below combines the steps the developers describe for z-score work [source: Thatcher et al., 2019] with ISNR's practice and QEEG standards [source: Hammond et al., 2004; 2011]; the arrangement into phases and the movement criteria are [synthesis].

Format. Individual, in-office sessions with the practitioner present; ISNR treats leaving clients alone for much of a session, as in "treatment mill" settings, as below standard [source: Hammond et al., 2011]. Home or remote training only after in-office evaluation and sessions, under licensed supervision [source: Hammond et al., 2011].

Session counts: reported figures, not prescriptions. ISNR asks that session estimates be realistic and decided jointly; it notes ADHD research commonly cites 40–50 sessions and discourages promising lasting results after few sessions [source: Hammond et al., 2011]. The z-score developers report score reductions in "10 sessions or less" in a clinical-practice figure [source: Thatcher et al., 2019]; Coben et al. state that claims of fewer sessions rest on uncontrolled data and that the only LNFB randomised trial in ADHD used a session count similar to traditional studies [source: Coben et al., 2019]. The Dutch guideline reports a typical neurofeedback course of 25–50 one-hour sessions [source: NVvP, 2020].

3. Guiding principles

• The QEEG supplements clinical judgment and never alone makes a diagnosis. Practical meaning: say to the client and in reports that a map does not diagnose ADHD, depression or brain injury [source: Hammond et al., 2004; Thatcher et al., 2019: "Normative databases themselves do not diagnose"].
• A QEEG is not a clinical EEG. Practical meaning: it does not screen for tumours, epilepsy or dementia; ISNR's sample consent wording says so, and suspicious raw EEG goes to a neurologist or electroencephalographer [source: Hammond et al., 2004].
• Inspect the raw EEG first. Practical meaning: no map is read until the raw trace has been visually reviewed and artifact-edited, without over-editing away usable data [source: Hammond et al., 2004].
• Expect false positives. Practical meaning: the number of measures times the significance level estimates how many "abnormal" scores chance alone will produce; require consistency across repeated samples before acting [source: Hammond et al., 2004].
• Know the database you compare against. Practical meaning: check its sample size, age range, inclusion and exclusion criteria, equipment matching and statistical corrections, and compare only with an age-congruent sample [source: Hammond et al., 2004].
• Deviation is not automatically pathology. Practical meaning: abnormal findings may not match the client's symptoms or be clinically meaningful; diagnostic categories are heterogeneous [source: Hammond et al., 2004]. The developers' own rule is to choose targets that are both deviant and linked to the complaint [source: Thatcher et al., 2019].
• Stay inside your licence and competence. Practical meaning: collecting QEEG needs documented training; interpreting it needs a higher standard (QEEG diplomate, EEG and Clinical Neuroscience Society certification, a neurology board for physicians, or supervised analysis) [source: Hammond et al., 2011].
• Consent is an ongoing educational process. Practical meaning: separate written consent for QEEG and for training, naming experimental status, alternatives and side effects; revisit it when the plan changes [source: Hammond et al., 2004; 2011].
• Be honest about status. Practical meaning: ISNR calls it unethical to tell clients side effects never occur, or to claim cure or remission from neurofeedback [source: Hammond et al., 2011]; Coben et al. say "better and faster" marketing for ZNFB/LNFB is not justified [source: Coben et al., 2019].
• Standard protocols first. Practical meaning: Coben et al. suggest ZNFB/LNFB be considered only off-label after standard neurofeedback protocols and other established treatments have failed, with informed consent to experimental status [source: Coben et al., 2019]. The developers disagree and present z-scores as reducing guesswork [source: Thatcher et al., 2019]; this is a contested point.
• Coach, do not just connect. Practical meaning: be with the client throughout the session [source: Hammond et al., 2011].
• Neurofeedback is one part of a multimodal plan. Practical meaning: weigh psychotherapy, medication, peripheral biofeedback and other options with the client [source: Hammond et al., 2011].
• Stop what does not help. Practical meaning: do not encourage continued training when there is no benefit; at least 20% of patients may not improve significantly, by ISNR's estimate [source: Hammond et al., 2011].
• Declare commercial ties. Practical meaning: several key sources are written by people with company or method interests (the z-score method is the developers' licensed product; review authors disclose company ties); tell clients about your own [source: Thatcher et al., 2019; Coben et al., 2019; Hammond et al., 2011 on conflicts of interest].

4. Intervention inventory

Items 4.1–4.12 describe delivery as ISNR and the developers publish it, at a descriptive level: no equipment settings, reward thresholds, electrode site lists or session prescriptions are given, and these belong to accredited training and supervision. Items 4.13–4.15 are the psychotherapist's own work whether or not they deliver the training; 4.16–4.17 cover certification and evidence. For the conventional amplitude protocols that a QEEG may be used to select or individualise, see 045 Alpha training (Kamiya), 046 Alpha-theta training and the Peniston protocol, 047 Sensorimotor-rhythm training (Sterman), 048 Theta-beta training (Lubar) and 049 Slow cortical potential training (Birbaumer).

4.1 Intake, history and medical clearance

• Purpose / when: Before any recording. Establishes appropriateness, risk level and the medical questions that are not the neurotherapist's to answer.
• Steps: (1) Take a history of medical and psychological problems, prior treatments and current medications [source: Hammond et al., 2011]. (2) Ask specifically about head or skull injury, seizures, and medical and psychiatric history, which ISNR also lists as background for the QEEG [source: Hammond et al., 2004]. (3) For any medical condition, confirm that an appropriate medical evaluation has taken place, and work with the treating professionals [source: Hammond et al., 2011]. (4) Screen ISNR's listed risk factors (complexity, severity, chronicity, suicide history, personality disorder, emotional lability, dissociation, abuse history, sensitivity to past treatments, litigiousness, comorbidity); where present, plan more thorough assessment, stronger consent, careful documentation and consultation [source: Hammond et al., 2011]. (5) Ask for referral to a physician first where there are seizures, an unexplained neurological change, a recent head injury with new symptoms, or a medical cause not yet excluded [synthesis]. (6) Standard suicide and self-harm risk assessment and safety planning as in any psychotherapy intake [synthesis].
• Example language (illustrative): "Before we look at your brain activity, I need to know what doctors have already checked. A brain map here is not a medical test, so if something medical needs ruling out, that comes first."
• Watch for: Clients who hope the map will replace a medical or diagnostic work-up; untreated seizures; recent head injury; medication started or changed recently (affects the EEG and confounds results) [synthesis, informed by Hammond et al., 2004 on recording recent medications].

4.2 Informed consent for the QEEG and, separately, for training

• Purpose / when: Written consent before the QEEG and again before neurofeedback [source: Hammond et al., 2004].
• Steps: Cover: assessment procedures; that the QEEG is not a clinical EEG and does not detect tumours, epilepsy or dementia [source: Hammond et al., 2004]; fees; confidentiality limits; recommended treatment; practitioner qualifications; expected frequency and duration; the right to stop at any time; the supervisor's name if a technician or student is involved; alternatives (for example medication or CBT) with their benefits and drawbacks; experimental status where the condition lacks adequate peer-reviewed support; side effects (ISNR lists fatigue, headache, anxiety and sleep difficulty) and the instruction to report them promptly; that neurofeedback is not always effective; written permission for any anonymised research or teaching use [source: Hammond et al., 2011]. For ZNFB/LNFB, state that these approaches are considered experimental and are not first-line [source: Coben et al., 2019]. Adapt for language, hearing or cognitive limits [source: Hammond et al., 2011].
• Example language (illustrative): "This kind of training is still considered experimental. Some people feel tired, get headaches or sleep worse for a while; tell me straight away if that happens. The standard alternatives are…"
• Watch for: Consent given under parental or partner pressure; a client who has heard marketing claims of cure; children's assent [synthesis].

4.3 Recording the QEEG

• Purpose / when: Phase B; obtaining data clean enough to compare with norms.
• Steps (descriptive): Record eyes-open and eyes-closed conditions, ideally also a task such as reading or arithmetic; note the background variables (birth date, handedness, time of day, recent medication, food, caffeine, nicotine); monitor eye movement and pulse artifact; pause eyes-closed recording regularly to keep the client alert, and recognise drowsiness patterns; record more rather than less so that enough usable data remains after artifact removal [source: Hammond et al., 2004]. ISNR notes a full 19-channel recording is not required for effective neurofeedback, but many clinicians use one to guide training; the z-score developers start from a 19-channel normative comparison [source: Hammond et al., 2004; Thatcher et al., 2019]. Electrode preparation, montage and technical recording standards are taught in accredited QEEG training and are not reproduced here.
• Example language (illustrative): "Now just sit with your eyes closed, staying awake; I'll check in every so often. Try not to clench your jaw or move your eyes much."
• Watch for: Drowsiness, muscle tension, eye movement; anxiety about the cap; sensory-sensitive clients and children who need a slower set-up [synthesis].

4.4 Raw EEG review and artifact editing

• Purpose / when: Before any quantitative analysis.
• Steps: Visually inspect the raw EEG, which ISNR says QEEG must not replace; edit artifacts carefully; check for over-editing that leaves too little data [source: Hammond et al., 2004]. Identify findings outside your scope and refer them for medical evaluation [source: BioSource summary of Collura et al., 2025; secondary].
• Watch for: Possible epileptiform or other unusual activity: stop and refer to a neurologist or electroencephalographer rather than training through it [synthesis, consistent with Hammond et al., 2004 on physician consultation].
• Example language (illustrative): "Before anything is calculated, I look through the raw recording by eye and take out stretches spoiled by blinks or movement." [synthesis]

4.5 Normative-database comparison

• Purpose / when: Expressing the client's measures as deviations from an age-matched reference group.
• Steps: Before relying on a database, know its number of subjects, age range, inclusion and exclusion criteria, whether recordings used the same equipment and how differences were handled, and what statistical corrections were applied to make data normally distributed [source: Hammond et al., 2004]. Compare only with an age-congruent sample. Use databases whose measures match the training you plan; reliability (test-retest or split-half) is ideal, especially in medical-legal work [source: Hammond et al., 2004]. Apply discriminant functions only when the client's presentation matches that function's criteria [source: Hammond et al., 2004]; classification methods should not be a primary diagnostic screen because of high false-positive rates [source: BioSource summary of Collura et al., 2025; secondary]. Replicate significant findings across samples [source: Hammond et al., 2004].
• Contested: The validity of normative databases for psychiatric use is disputed. The AAN/ACNS assessment (Nuwer, 1997) placed QEEG for psychiatric disorders and learning disabilities in its rejected, still-experimental category and gave a negative recommendation for head injury, as a critique describes it; that critique (drafted with AAPB and SSNR support) argued the report ignored work on normative databases, cross-validation and reliability [source: JNP page, "Limitations of the AAN and ACNS Paper on QEEG"; the 1997 report itself not verified, uncertain]. Coben et al. describe but do not analyse the database and "deviation equals pathology" assumptions [source: Coben et al., 2019].
• Watch for: Many "significant" scores in a large array (expected by chance); medication effects; age mismatch [source: Hammond et al., 2004].
• Example language (illustrative): "The comparison group is people your age recorded in a similar way. With this many measures, some scores will look unusual just by chance." [synthesis]

4.6 Interpretation and feedback of the map to the client

• Purpose / when: After analysis; a separate feedback session.
• Steps: (1) Present findings as patterns relative to a reference group, not as a diagnosis [source: Hammond et al., 2004]. (2) Separate findings that fit the complaint from those that do not; ISNR notes abnormal findings may not match symptoms [source: Hammond et al., 2004]. (3) Say what is uncertain and what will be referred. (4) Link to the proposed training plan only through findings you and the client agree are relevant [synthesis]. (5) Diagnoses stay within licence and training; non-physician neurotherapists generally do not make medical diagnoses from QEEG [source: Hammond et al., 2004].
• Example language (illustrative): "These colours show where your recording differs from people your age in the database. Differences are not damage, and some will have nothing to do with why you came. The ones that might fit your sleep problem are these two."
• Watch for: Clients who hear "abnormal brain"; catastrophising; relief that becomes over-attribution of all difficulties to the map [synthesis].

4.7 Protocol selection from the QEEG

• Purpose / when: Phase C.
• Steps (as developers describe, z-score route): Link the client's symptoms to brain regions and networks using clinical knowledge and the functional-imaging and lesion literature (the developers suggest a literature search by symptom); choose locations and measures by symptom relevance and size of deviation; for source-localised training, choose Brodmann-area hubs and connections; review instantaneous z-scores before training [source: Thatcher et al., 2019]. The developers say the location of a deviation matters more than its direction [source: Thatcher et al., 2019]. ISNR gives no rules for mapping findings to protocols [source: Hammond et al., 2004].
• Steps (QEEG-guided conventional route): Use the map to choose and individualise one of the conventional protocols described in 045–049 [source: Hammond et al., 2004 on QEEG guiding training; detail of the matching is not in the sources cited, uncertain].
• Watch for: Choosing targets for deviation alone with no link to the complaint; Coben et al.'s recommendation that standard protocols be tried first [source: Coben et al., 2019].
• Example language (illustrative): "I'd suggest we train only the patterns that fit what you came for, and keep the others in mind rather than chase them." [synthesis]

4.8 Surface z-score training sessions

• Purpose / when: Phase D, z-score route.
• Steps (descriptive): The live EEG is converted into z-scores (power, asymmetry, coherence, phase and similar) by joint time-frequency analysis; feedback (visual or auditory) is given when the selected measures move toward the reference mean, so there is no separate inhibit/enhance band logic; strongly deviant artifacts are said to be excluded automatically [source: Thatcher et al., 2019]. The developers note live z-scores run smaller than the offline (FFT) z-scores used for assessment, which is why assessment comes first [source: Thatcher et al., 2019]. Progress is tracked by session-by-session charts and topographic or LORETA maps [source: Thatcher et al., 2019]. The practitioner coaches throughout [source: Hammond et al., 2011].
• Example language (illustrative): "You don't need to try hard. Let the sound keep playing; notice what you are doing inside when it plays more."
• Watch for: Falling z-scores in session charts are a change in the trained measure, not proof of symptom change; keep symptom ratings separate [synthesis, consistent with Coben et al., 2019 on needing clinical endpoints].

4.9 LORETA and LORETA z-score training sessions

• Purpose / when: Phase D, source-localised route.
• Steps (descriptive): Feedback is driven by estimated current density, or its z-score, in a region of interest or network. In the first pilot, three people trained to change low-alpha and beta activity estimated in the anterior cingulate, with baseline recordings before training and a final test in which they tried to earn many rewards in some trials and few in others [source: Congedo, Joffe & Lubar, 2004]. Studies since have trained the anterior cingulate (ADHD), precuneus and posterior cingulate (tinnitus distress) [source: Coben et al., 2019]. The developers describe swLORETA, an MRI-voxel and boundary-element variant, and propose deep targets (cerebellum, thalamus, subthalamus) for balance and parkinsonism as future work [source: Thatcher et al., 2019].
• Limits named by sources: LORETA and sLORETA have localisation limits from simplified head models [source: Thatcher et al., 2019]; in the ADHD trial, control over the targeted region was limited and benefits may reflect artifact reduction; in the precuneus study, clinical participants did not show the expected region changes [source: Coben et al., 2019].
• Watch for: Adverse effects from some LNFB protocols have been reported (Cannon, 2012, cited by Coben et al., 2019; original not verified, uncertain): ask about side effects every session.
• Example language (illustrative): "This version estimates activity in one deeper area. It is newer and less tested, so I'll ask about side effects at every session." [synthesis]

4.10 In-session coaching and monitoring

• Purpose / when: Every training session.
• Steps: Stay with the client, coach, and prevent drifting, loss of focus and drowsiness [source: Hammond et al., 2011]. Ask frequently about side effects; if they occur, document, discuss, and modify the protocol, change session frequency, add adjunctive treatment or consult [source: Hammond et al., 2011]. Obtain symptom ratings at clinically appropriate intervals; pre/post EEG and outcome measures where feasible [source: Hammond et al., 2011]. Keep records of consent, emergency procedures, procedures used, side effects and progress [source: Hammond et al., 2011]. Bill biofeedback services under biofeedback codes, not psychotherapy codes (US) [source: Hammond et al., 2011].
• Example language (illustrative): "Since last time, any headaches, tiredness, changes in sleep or mood? On the 0–10 scale we agreed, where is the main problem this week?"
• Watch for: Side effects, symptom worsening, low engagement [source: Hammond et al., 2011].

4.11 Generalisation and transfer

• Purpose / when: From mid-course onward, so that change is used in daily life rather than only at the screen.
• Status: Not addressed in the ISNR standards or the z-score/LORETA sources cited [source: Hammond et al., 2011 (not addressed); Thatcher et al., 2019 (not addressed); uncertain]. Transfer trials without feedback are part of slow cortical potential training as its developers describe it: see 049. Everything below is [synthesis].
• Steps [synthesis]: (1) In each session, ask the client what state or strategy went with reward and give it a name. (2) Agree one daily situation (homework, a meeting, bedtime) in which to recall that state. (3) Review use of it at the start of the next session. (4) Track the complaint with the client's own measures, not session z-scores. (5) Where psychotherapy runs alongside, link the named state to existing skills (for example relaxation, attention or emotion-regulation work).
• Example language (illustrative): "When the feedback was playing most, what were you doing in your head? Let's call that 'steady'. Where this week could you try being 'steady' for a minute?"
• Watch for: Improvement only in session data; clients attributing change only to the machine [synthesis].

4.12 Home and remote training (if offered)

• Purpose / when: Only after in-office evaluation and adequate in-office sessions [source: Hammond et al., 2011].
• Steps: Train the on-site helper (for example a parent) in equipment, sensor placement and coaching; obtain weekly progress data, ideally with live internet supervision and periodic face-to-face visits; use equipment that prevents the home user changing protocols; put in writing the data frequency, office visits, client-only use, fees, damage, return of equipment and a ban on offering services to others; tell clients that research on remote outcomes is very limited; side effects reported immediately; check licensing rules for clients in other states or countries [source: Hammond et al., 2011]. ISNR: home training "should never be done" without a licensed professional supervising it [source: Hammond et al., 2011].
• Watch for: Unsupervised use, sharing equipment, protocol changes at home [source: Hammond et al., 2011].

4.13 Psychoeducation about brain maps (for any psychotherapist)

• Purpose / when: When a client arrives with, or is offered, a QEEG "brain map" from any provider.
• Steps: (1) Ask what they were told the map shows. (2) Explain, in plain terms, that a QEEG compares their recording with a reference group; that it is not a clinical EEG; that it does not diagnose ADHD, depression or brain injury on its own; and that some differences are expected by chance [source: Hammond et al., 2004]. (3) Explain that the neurology bodies have not endorsed QEEG for psychiatric diagnosis and that the method is contested [source: JNP critique page; Wikipedia citing Barrett, 2015; secondary]. (4) Return to the client's own goals and the full range of options [synthesis].
• Example language (illustrative): "A brain map can be one piece of information, but it isn't a diagnosis. What were you hoping it would explain?"
• Watch for: Shame or fatalism ("my brain is broken"); a map used to override a clinical diagnosis or prescriber's advice [synthesis].

4.14 Shared decision-making for clients considering a commercial QEEG/z-score/LORETA course

• Purpose / when: The client asks your view, or a family is about to commit to a costly course.
• Steps: Help the client ask the provider, drawing on ISNR's standards: Is the provider licensed for independent practice, BCIA-certified, and certified for QEEG interpretation? Who reviews the raw EEG? Which database is used and is it age-matched? Is the practitioner present and coaching in every session? What are the alternatives and the stated evidence for this condition? What side effects are possible? How will benefit be measured, and when will you stop if there is none? Are there claims of cure? [source: Hammond et al., 2004; 2011]. Share that independent reviewers class ZNFB/LNFB as experimental and not first line [source: Coben et al., 2019]. Never advise a change to medication; route any medication question to the prescriber [synthesis].
• Example language (illustrative): "It's your decision. Would it help to take these questions to them, and to agree with them what result after how long would tell you it is or isn't working?"
• Watch for: Pressure selling, packages sold up front, promises of fewer sessions or permanent cure (ISNR treats cure claims as unethical) [source: Hammond et al., 2011].

4.15 Collaboration and integration with psychotherapy

• Purpose / when: Throughout, whether you deliver the training or a colleague does.
• Steps: (1) Agree with the client who holds the overall plan; ISNR describes neurofeedback as one part of a multimodal plan alongside psychotherapy, medication and other options [source: Hammond et al., 2011]. (2) With consent, share symptom ratings and side effects between neurofeedback provider, therapist and prescriber [synthesis]. (3) Tell the neurofeedback provider about medication changes, which the QEEG record needs [source: Hammond et al., 2004, background variables] and which can confound training [synthesis]. (4) Continue psychotherapy goals in parallel; use sessions to process what the client notices in training [synthesis]. (5) Pass on warning signs (worsening mood, suicidal ideation, new neurological symptoms, sleep collapse) to the provider and medical team, and follow your usual crisis pathway [synthesis].
• Watch for: Split messages between providers; the client stopping effective treatment in favour of training [synthesis].
• Example language (illustrative): "With your permission, I'll share your weekly ratings with the neurofeedback provider and your prescriber so we're all working from the same picture." [synthesis]

4.16 Training and certification routes

• Neurofeedback (BCIA). ISNR recommends board certification from the Biofeedback Certification International Alliance (BCIA) [source: Hammond et al., 2011]. As summarised by a BCIA-accredited training provider: an entry-level licensed route (degree in a BCIA-approved health field plus a licence to treat diagnosed disorders), an entry-level performance route for unlicensed holders of approved degrees (no independent work with diagnosed disorders), and a technician route under a licensed BCIA-certified supervisor; requirements include a BCIA-accredited didactic programme, a BCIA-approved neuroanatomy, neurophysiology, behavioural-neuroscience or physiological-psychology course, mentoring (25 contact hours covering 10 personal self-regulation sessions, 100 client sessions and 10 case presentations), a proctored written exam, and recertification every four years [source: BioSource Software, BCIA neurofeedback certification page; secondary; uncertain: not verified on bcia.org].
• QEEG. ISNR's 2004 paper "strongly encouraged" QEEG certification and named the QEEG Certification Board (Technologist and Diplomate levels) and the EEG and Clinical Neuroscience Society (for MDs and PhDs), with BCIA and the Neurotherapy Certification Board for EEG biofeedback [source: Hammond et al., 2004]. ISNR's 2011 standards require a higher standard for interpreting than for collecting QEEG (QEEG diplomate, ECNS certification, neurology board, or supervised analysis) [source: Hammond et al., 2011]. The International QEEG Certification Board (IQCB) currently sets training, examination and practice guidelines for clinical QEEG and qualification to review raw EEG is shown by IQCB certification or equivalent [source: BioSource summary of Collura et al., 2025; secondary]. Whether IQCB is the successor of the QEEG Certification Board was not confirmed [uncertain].
• Training content ISNR lists for QEEG practitioners: functional neuroanatomy, EEG rhythms and age changes, instrumentation, pattern recognition, recording technique, montages, artifact recognition and its effect on QEEG, interpretation, statistics including false positives, and medical-legal issues; those recording their own EEGs should first study conventional EEG [source: Hammond et al., 2004].
• Licence. Deliver only within your licence; an introductory course does not qualify independent practice [source: Hammond et al., 2011].
• Purpose / when: when you plan to deliver QEEG-guided or z-score training yourself, or need to check a provider's credentials. [synthesis]
• Steps [synthesis]: check the provider's licence first, then BCIA neurofeedback certification, then QEEG-specific certification for anyone interpreting maps; verify each on the certifying body's own register.
• Example language (illustrative): "Could you tell me which licence you hold, and who is certified to read the QEEG?" [synthesis]
• Watch for: weekend courses presented as qualification for independent practice; certification offered in place of a licence. [synthesis]

4.17 Evidence status (for consent and planning)

• Independent review: Of 15 studies examined, 6 met inclusion: 3 uncontrolled 19-channel ZNFB studies and 3 LNFB studies with control groups (one randomised ADHD trial; precuneus training; posterior cingulate training for tinnitus distress). Conclusion: ZNFB has "virtually no evidence" of efficacy or superiority; LNFB shows "some suggestion for efficacy" in ADHD and tinnitus distress, unreplicated; neither has shown superiority over standard protocols; both are experimental [source: Coben et al., 2019]. The review authors disclose ties to other neurofeedback methods and companies [source: Coben et al., 2019].
• Developers' position: reductions in z-scores in many studies and cases; no reported adverse reactions across 13 years, over 3,000 clinicians and over 60 studies, and "more published research is needed" [source: Thatcher et al., 2019]. Much of their cited support is case reports, conference material and newsletters, and the developers have a commercial interest [source: Thatcher et al., 2019].
• Guidelines: NVvP (Netherlands, 2020): neurofeedback is not recommended for children with ADHD; small, non-clinically-meaningful benefit, inconsistent results, possible conflicts of interest; QEEG-guided forms not discussed separately [source: NVvP, 2020]. NICE NG87 (2018): neurofeedback not among listed recommended non-drug options per a secondary summary [source: Jones, 2019; NICE text not verified, uncertain]. ISNR (2011) listed ADHD, uncontrolled epilepsy, insomnia and addictive disorders as having considerable evidence for neurofeedback in general, and many other conditions as lacking adequate research [source: Hammond et al., 2011]; this is a professional society's view and predates the reviews above.
• QEEG as diagnostic: contested; see 4.5.
• Purpose / when: at consent and at each review, so that the client's choice reflects the current evidence. [synthesis]
• Steps [synthesis]: summarise the independent review, the developers' position and the guideline position, naming who says what and their interests.
• Example language (illustrative): "Independent reviewers still call these methods experimental; the developers see it differently. Both views are worth hearing before you decide." [synthesis]
• Watch for: presenting developer case series as established evidence; omitting guideline advice against routine use. [synthesis]

4.18 Review, ending and follow-up

• Purpose / when: At planned review points and at the end of training.
• Steps: Compare symptom ratings and outcome measures with baseline; repeat EEG measures where feasible [source: Hammond et al., 2011]. Decide jointly on stopping, continuing, switching or referring, taking account of diagnosis, response, comorbidity and stressors, and explain the reasons; do not encourage continued treatment that shows no benefit [source: Hammond et al., 2011]. Detailed follow-up procedures are not specified by ISNR [source: Hammond et al., 2011]; arrange a follow-up review of symptoms and agree what would prompt a return or another treatment [synthesis]. Close with a written summary for the client and, with consent, for the referrer and therapist [synthesis].
• Example language (illustrative): "Your sleep ratings have improved and held for a month. Shall we stop here and check in again in a couple of months, or would you rather taper off?"
• Watch for: Dependency on sessions; ending forced by cost without a plan; no benefit but pressure to continue [synthesis; Hammond et al., 2011 on not continuing without benefit].

5. Handling problems

• No response after a fair trial. ISNR estimates at least 20% of patients may not improve significantly; do not encourage continued treatment without benefit; explain and jointly decide to switch, refer or stop [source: Hammond et al., 2011]. Consider whether a standard protocol (045–049) or an established treatment has been tried first, as Coben et al. recommend [source: Coben et al., 2019]. Re-check the formulation: were targets chosen for deviation without link to the complaint? [synthesis].
• Side effects (fatigue, headache, anxiety, sleep difficulty). Ask every session; document and discuss; modify the protocol, change session frequency, add adjunctive treatment or consult [source: Hammond et al., 2011]. Adverse effects of some LNFB protocols have been reported (Cannon, 2012, cited by Coben et al., 2019) [uncertain: original not verified]. If symptoms are severe, new or neurological, pause training and refer for medical review [synthesis].
• Map findings that do not fit the client. Expected: abnormal findings may not match symptoms or be meaningful, and many scores reach significance by chance [source: Hammond et al., 2004]. Do not train them only because they are deviant; replicate before acting [source: Hammond et al., 2004; Thatcher et al., 2019 on linking to symptoms].
• Suspected medical finding on the raw EEG. Stop, do not interpret beyond your scope, refer to a neurologist or electroencephalographer, and explain to the client that the QEEG is not a clinical EEG [source: Hammond et al., 2004; BioSource summary of Collura et al., 2025].
• Client or family wants a QEEG diagnosis (for example of ADHD or brain injury). QEEG should never be the sole basis for diagnosis, and non-physician neurotherapists generally do not make medical diagnoses [source: Hammond et al., 2004]; refer for standard diagnostic assessment and explain why [synthesis]. In medical-legal cases (for example head injury), ISNR advises documented training and certification or neurologist consultation, and physician review of the raw EEG may be desirable [source: Hammond et al., 2004]; the diagnostic use in head injury was given a negative recommendation by the AAN/ACNS (as described by its critics) [source: JNP critique page; uncertain].
• Distress or fatalism after seeing the map. Normalise variation, re-state that the map is not a diagnosis, and work with the client's meaning in psychotherapy [synthesis; Hammond et al., 2004 on limits].
• Drowsiness, drifting, low engagement in session. Coach actively and remain present [source: Hammond et al., 2011]; for children, shorter tasks and frequent check-ins, and review whether parent coaching is properly supervised [synthesis; Hammond et al., 2011 on parent coaches].
• Non-adherence (missed sessions, home-training lapses). Explore ambivalence and practical barriers as in any therapy; re-check that the client still endorses the goal; in home training, review weekly data and the written agreement [synthesis; Hammond et al., 2011 on remote rules].
• Medication change mid-course. Record it (it is a QEEG background variable), tell the neurofeedback provider, and interpret changes cautiously; never advise starting, stopping or changing medication: route to the prescriber [source: Hammond et al., 2004 for recording medications; rest synthesis].
• High-risk clients (suicide history, dissociation, abuse history, emotional lability, personality disorder). ISNR calls for more thorough assessment, stronger consent, careful documentation and consultation [source: Hammond et al., 2011]. Keep a safety plan current, monitor mood and dissociation each session, and slow or pause training if states destabilise [synthesis].
• Rupture over cost or broken expectations. Name it, review what was promised against what sources support (ISNR rejects cure claims; Coben et al. reject "better and faster" marketing) [source: Hammond et al., 2011; Coben et al., 2019], and repair by renegotiating goals or ending with a referral [synthesis].
• Pressure from a commercial setting or "treatment mill" practice (clients left alone, technicians unsupervised, packages pre-sold). This falls short of ISNR standards; unlicensed technicians need full-time on-site licensed supervision and a written agreement; the licensed supervisor carries responsibility [source: Hammond et al., 2011]. If you learn of unauthorised home-equipment service provision, ISNR says it must be reported to licensing boards [source: Hammond et al., 2011].
• Conflicting expert claims. Tell clients plainly that developers and independent reviewers disagree, name who says what and their interests, and let the client decide with full information [synthesis; Thatcher et al., 2019; Coben et al., 2019].
• DIY or consumer neurofeedback devices. ISNR states home training should never be done without licensed supervision because of risks of harm [source: Hammond et al., 2011]; explore what the client is using and why, and discuss supervised alternatives [synthesis].

Sources

Online sources:

• Standards for the Use of Quantitative Electroencephalography (QEEG) in Neurofeedback: A Position Paper of the International Society for Neuronal Regulation (Hammond, Walker, Hoffman, Lubar, Trudeau, Gurnee & Horvat, 2004, Journal of Neurotherapy 8(1), 5–27) – abstract page and full text
• Standards of Practice for Neurofeedback and Neurotherapy: A Position Paper of the International Society for Neurofeedback & Research (Hammond, Bodenhamer-Davis, Gluck, Stokes, Harper, Trudeau, MacDonald, Lunt & Kirk, 2011, Journal of Neurotherapy 15(1), 54–64)
• Z-Score EEG Biofeedback: Past, Present, and Future (Thatcher, Lubar & Koberda, 2019, Biofeedback 47(4), 89–103)
• Electromagnetic Tomographic Neurofeedback: A Pilot Study (Congedo, Joffe & Lubar, 2004, Journal of Neurotherapy 8(1), 85–86)
• 19 Channel Z-Score and LORETA Neurofeedback: Does the Evidence Support the Hype? (Coben, Hammond & Arns, 2019, Applied Psychophysiology and Biofeedback 44, 1–8)
• Limitations of the American Academy of Neurology and American Clinical Neurophysiology Society Paper on QEEG (Journal of Neuropsychiatry and Clinical Neurosciences page)
• ADHD bij kinderen: Neurofeedback bij ADHD bij kinderen (Nederlandse Vereniging voor Psychiatrie, 2020)
• Alternatives to medication for children and young people with ADHD (Jones, 2019, National Elf Service)
• BCIA Neurofeedback Certification (BioSource Software)
• Dr John Davis explains the latest IQCB guidelines (BioSource Software blog, on Collura et al., 2025)
• Quantitative electroencephalography (Wikipedia)

Cited, not verified [uncertain]:

• Nuwer, M. (1997). Assessment of digital EEG, quantitative EEG and EEG brain mapping: report of the American Academy of Neurology and the American Clinical Neurophysiology Society. Neurology, 49, 277–292 (details as given in the JNP critique page).
• NICE (2018). Attention deficit hyperactivity disorder: diagnosis and management (NG87) (as cited by Jones, 2019).
• Collura et al. (2025), IQCB qEEG guidelines (as summarised by BioSource Software).
• Cannon (2012), on adverse effects of LNFB protocols (as cited by Coben et al., 2019).
• BCIA's own certification pages (bcia.org) [uncertain: not verified]; check current requirements there.