Sign in to Peak Brain Path
Sign in to access your courses, books, and progress tracker. New here? Signing in creates your account automatically.
Want to explore courses first?
Browse courses and booksSign in to access your courses, books, and progress tracker. New here? Signing in creates your account automatically.
Want to explore courses first?
Browse courses and booksModule 12
Chapter 12 · 2 h · 8 quiz items · pass at 80%
This module applies IQCB Domain V (QEEG), 21% of the exam, to the act of reading a real report: maps, phenotypes, connectivity, asymmetry, and the state and stability effects that separate a trait finding from an artifact of the day. The quiz confirms the learner can interpret a QEEG report's findings and judge what a meaningful change looks like on serial assessment.
The pipeline produced metrics (Chapter 10). The database turned them into z-scores (Chapter 11). Now you read the result: a multi-page report of topographic maps and tables that has to become a clinical statement. This is where Domain V earns its 21 percent, because the exam tests not whether you can run an FFT but whether you can look at a brain map and say what it does and does not support. Interpretation is the deliverable. Everything before it was preparation.
The whole chapter rests on one principle carried forward from the database work: a flagged finding on a colorful map is a hypothesis, not a conclusion. The reds and blues communicate urgency that the underlying z-score may not justify. Your job is to read the map against that skepticism, to favor convergent patterns over isolated points, and to keep statistical deviation and clinical significance in separate boxes. The phenotype patterns below are condensed. The QEEG Field Guide phenotype atlas carries the full per-phenotype entries (detection rules, evidence grades, medication confounds, co-occurrence, stability), and this chapter forward-references it for depth rather than reproducing it.
A standard report is a stack of displays, each rendering a different metric across the scalp. Read them in a deliberate order rather than reacting to whichever blob is reddest.
Absolute power maps. One topographic head per band, showing how much activity (μV²) sits at each site. These answer "how much delta, theta, alpha, beta, and where." Read them first for the gross picture: is there a posterior alpha peak where it belongs, is slow activity anterior or diffuse, is the overall amplitude high or low. Remember that absolute power scales with skull thickness, impedance, and the low-voltage trait, so a globally low absolute map may be constitution rather than deficit.
Relative power maps. The same bands expressed as a fraction of total power at each site. These answer "what proportion of this brain's output is in each band." They cancel the global scaling that contaminates absolute power, so they are more stable for spectral shape, but they carry the denominator trap: a change in one band mechanically shifts all the others. Cross-check a relative finding against the absolute map before believing it. A finding that shows in both, same direction, is strong. One that shows only in relative power may be a denominator artifact.
Coherence maps. Displays of connectivity between electrode pairs, usually drawn as lines or a matrix, showing where coupling is excessive or deficient relative to norms. These answer "how are regions communicating." Read them after the power maps, because coherence is less reliable than power and an isolated coherence finding deserves more caution. Section 12.4 covers what over- and under-coherence mean.
z-Score maps. The interpretive layer. Each metric is shown as deviation from the age-matched norm, with color coding for direction and magnitude. These are where the clinical reading concentrates, and Section 12.2 covers how to read the color scale without being led by it.
The reading order, for the exam and the clinic: absolute power for the gross picture, relative power for spectral shape (cross-checked against absolute), coherence for connectivity, z-scores throughout for where the deviations sit. Then synthesize across displays rather than treating each as a separate finding.
The z-score map is the most-used and most-misread display in the report. Understanding its conventions prevents the errors the exam targets.
What red and blue mean. By near-universal convention, warm colors (red) mark deviation above the norm and cool colors (blue) mark deviation below, with saturation scaling to magnitude: deep red is a large positive z, deep blue a large negative z, and green or neutral marks the normal range. A deep red frontal blob means frontal power well above the age-matched mean; a deep blue posterior region means power well below it. The color is a direct rendering of the z-score and carries no information the z-score does not.
The symmetric-display assumption. The color scale is symmetric around zero: the same intensity of red and blue corresponds to the same absolute z-magnitude in opposite directions. This is a display choice, and it has a perceptual consequence the exam may probe. The eye reads a saturated red as "alarming" regardless of whether z = +2.0 or z = +4.0 if both saturate the scale, so a map looks dramatic when the underlying deviation is only moderate. Always check the color-scale legend for what saturation corresponds to: a map scaled to ±3 reads differently from the same data scaled to ±10. The display is calibrated by whoever set the scale, and a striking map at a compressed scale may be unremarkable at a standard one.
Reading the map against the multiple-comparison reality. A z-score map shows hundreds of metrics at once, and at a p < 0.05 threshold some will be colored by chance (Chapter 11). The map will show scattered flecks of color in a perfectly healthy brain. The skill is distinguishing a coherent region of consistent deviation (multiple adjacent sites, same band, same direction) from isolated colored specks that are statistical noise. A single bright electrode surrounded by neutral neighbors is far more likely a false positive or artifact than a real focal finding. The map rewards pattern-reading and punishes point-reading.
A phenotype is a QEEG pattern that recurs reliably across individuals, associates with functional or clinical characteristics, and can be detected by specified rules. Phenotypes are the bridge from "this map shows elevated frontal theta" to "this is a pattern that commonly co-occurs with attention difficulty." They are associations, not diagnoses, and the reverse-inference caution (Section 12.9) governs every one of them. What follows condenses the patterns the IQCB blueprint names. The Field Guide phenotype atlas carries the complete catalog of 31 phenotypes with full entries, and you should treat the atlas as the depth reference behind these summaries.
ADHD-associated patterns. The most-studied territory. The classic finding is elevated frontal theta (relative theta excess at F3, F4, Fz), accompanying an elevated theta/beta ratio, mapping to an underarousal-and-engagement-failure picture. A distinct subset shows frontal alpha excess (frontal alpha intrusion rather than theta), which is a different mechanism and a poor match for theta-down protocols. A further subset shows excess high beta (a fast-dominant variant, sometimes with anxiety or perfectionism). Low sensorimotor rhythm (deficient SMR at the central strip, C3/C4) co-occurs with the theta pattern and maps to motor restlessness and sleep-onset difficulty. The critical exam point: elevated frontal theta is a real group-level finding in ADHD but is sensitive and not specific (it appears in a meaningful fraction of typical children and is heavily confounded by drowsiness), and the theta/beta ratio has been shown to lack individual-level diagnostic validity (see Section 12.9). These patterns inform, they do not diagnose.
Anxiety and OCD-associated patterns. The headline is excess high beta (20 to 30 Hz), frontally predominant, read as cognitive hyperarousal: worry, rumination, mental racing, difficulty downshifting at rest. OCD-adjacent presentations show frontal high beta with the ruminative-checking quality. A second feature is failure of alpha to recover, deficient posterior alpha or impaired alpha reactivity (the alpha that should appear with eyes closed is reduced or fails to block normally), consistent with a brain that cannot drop into a resting state. The frontal high-beta finding carries a heavy artifact caveat: 20 to 30 Hz power at frontal and frontotemporal sites is frequently muscle (EMG) from jaw, scalp, or forehead tension, so rule out EMG before calling cortical high beta. This is one of the most common false-positive traps on the exam and in the clinic.
Depression-associated patterns. The canonical finding is frontal alpha asymmetry: more alpha on the left frontal site (F3) than the right (F4), implying reduced left-frontal activation. In the Davidson approach-withdrawal model, reduced left-frontal activation associates with withdrawal motivation, anhedonia, and depressive presentations. Posterior alpha excess appears in psychomotor-retarded presentations. The asymmetry finding carries a specific reliability caution that the exam may test: frontal alpha asymmetry has poor single-session reliability (intraclass correlation roughly 0.3 to 0.6), so a single recording is insufficient to characterize trait-level asymmetry, and multiple recordings are needed. The effect is also small-to-moderate and heterogeneous: a substantial fraction of people with depression show no asymmetry or the reverse, so this is a correlate of withdrawal motivation more than a marker of major depression as a category.
PTSD-associated patterns. Patterns reported in trauma populations include hyperarousal features (elevated high beta, deficient posterior alpha), coherence deviations (hypercoherence in some networks), and findings described as right temporal theta or right-lateralized abnormality. The evidence is less consistent and less diagnostically specific than the ADHD or depression literatures. PTSD QEEG findings are heterogeneous, and arousal-state fluctuation is prominent (a recording during versus between symptomatic periods can differ markedly). Treat PTSD-associated patterns as low-to-moderate confidence and heavily state-dependent.
TBI-associated patterns. Traumatic brain injury produces some of the more reliable QEEG findings: focal slowing (increased delta/theta over the injury region), reduced or disrupted coherence (the injury degrades connectivity, particularly across the lesioned tissue), and a slowing pattern whose persistence tracks with severity and outcome. Acute injury shows diffuse and focal slowing that partly normalizes over months; chronic injury shows persistent slowing and coherence disruption that can become constitutionally embedded. The Thatcher TBI discriminant work is the anchor literature. The critical caution, developed in Section 12.7 and in the Field Guide database chapter: a TBI discriminant was trained on TBI versus healthy, not TBI versus other causes of similar slowing, so untreated sleep apnea, post-viral cognitive complaints, polypharmacy, and metabolic states all produce a "TBI-positive" pattern in someone who never sustained a head injury. The pattern is real. The etiologic inference is the trap.
A closing note on all five: each phenotype is a probabilistic association documented at varying evidence grades (A through D in the atlas), and most clients show two to four phenotypes at once, not one in isolation. Single-phenotype presentations are the minority. Multi-phenotype presentations are the norm, and they require reading the whole map rather than fixing on the dominant finding. The atlas composite signatures catalog the common combinations.
Coherence findings split into two clinical directions, and the exam expects you to read both. The metric mechanics (magnitude-squared versus lagged coherence, the volume-conduction problem, the zero-lag handling) are Chapter 10. Here the question is what a coherence deviation means.
Over-coherence (hypercoherence). Excessive coupling between sites: the regions are too tightly locked, rising and falling together more consistently than the norm. The clinical reading is rigidity, an inability to reconfigure, networks that cannot flexibly decouple to meet changing demands. Over-coherence appears in perseverative and obsessive presentations and in some trauma patterns. The mental model is a system stuck in a fixed coupling pattern when it should be able to engage and release connections as tasks change.
Under-coherence (hypocoherence). Deficient coupling: the regions are too weakly or inconsistently connected, communication fragmented. The clinical reading is disconnection, failure of regions to coordinate. Under-coherence appears in some ADHD presentations and in disconnection following injury. TBI is the clearest case: the injury physically degrades the white-matter pathways, and coherence across the affected tissue drops.
The interpretive cautions specific to coherence. Three, all testable. First, coherence is less reliable than power (test-retest roughly 0.5 to 0.7), and long-range coherence is less reliable than short-range, so isolated coherence findings on a single recording warrant caution, especially as the sole basis for a connectivity-focused interpretation. Second, the volume-conduction problem means magnitude-squared coherence inflates apparent coupling between nearby electrodes that are merely seeing a shared source. Favor lagged coherence or phase lag index findings, which are built to suppress that artifact, when the question is genuine inter-regional coupling. Third, scalp coherence is an imperfect proxy for actual cortical connectivity: it describes the relationship between two scalp signals, not a verified anatomical or functional connection between two brain regions. Report coherence findings as relationships between sites, with the proxy caveat, not as proven connectivity.
Asymmetry findings compare homologous sites across hemispheres and quantify lateralization (Chapter 10 covers the amplitude-asymmetry metric). Clinically, the dominant asymmetry is frontal alpha asymmetry, covered under depression above. The general reading principles extend to other asymmetries.
A genuine asymmetry finding requires the same convergence discipline as any other: the lateralization should be consistent across conditions, survive artifact review (frontal asymmetries are biased by eye-movement spread from Fp1/Fp2 into F3/F4, and linked-ears reference imbalance can manufacture a false asymmetry, so average reference or current source density is preferred), and ideally reproduce across recordings. A unilateral finding at a single site in a single condition is weak. Asymmetry can also reflect a benign heritable trait: some families carry hemispheric asymmetries with no functional consequence, so a familial history of the same pattern in high-functioning relatives is relevant context before calling an asymmetry pathological. The reading principle: asymmetry describes lateralization, and lateralization is meaningful only when it is reliable, artifact-cleared, and corroborated by clinical context.
Source-localized findings (LORETA, sLORETA, eLORETA) appear in reports that carry source analysis, estimating where in the brain volume a surface finding originates. The methods and their progression are Chapter 10. The interpretive question here is how to read a source finding without overreading it.
A source finding moves a surface observation toward a generator: "F3 shows excess theta" becomes "the theta generator estimates to anterior cingulate." This is useful for depth (distinguishing surface cortex from deep midline structures the scalp cannot resolve) and for locating an anterior-versus-posterior generator. But source localization is a model-based estimate built on the inverse problem's assumptions: a standardized head model that may not match the individual's anatomy, a smoothness or filter constraint, and total dependence on clean input (LORETA will localize artifact as confidently as brain activity). Its spatial resolution with 19 channels is one to several centimeters, so it estimates a neighborhood, not a point. And source findings carry the reliability cautions of the surface metrics they derive from, amplified, because the inverse solution adds model-dependent variance (LORETA-specific test-retest data are limited, and narrow-band localized estimates are less reliable than broadband ones).
The reporting rule the exam rewards and the forensic chapter demands: present source findings as model-based estimates with their caveats, never as measured locations. "Source analysis estimates the generator in the region of the anterior cingulate" is defensible. "The anterior cingulate shows elevated theta" overclaims what a 19-channel inverse solution can establish.
A QEEG captures the brain in a particular state, and the state is part of the finding, not a nuisance to ignore. The most testable state contrast is eyes-open versus eyes-closed.
The eyes-open versus eyes-closed difference. With eyes closed, posterior alpha rises (the occipital-parietal alpha rhythm dominates). With eyes open, that alpha is suppressed, a phenomenon called alpha blocking or alpha reactivity, driven by intact thalamocortical function engaging the visual system. The normal pattern is a clear drop in posterior alpha from eyes-closed to eyes-open. Both conditions are recorded precisely because the difference between them is informative.
What it means when they collapse. If posterior alpha fails to block (eyes-open alpha remains high, the eyes-closed-to-eyes-open difference is small or absent), that is a clinically meaningful finding: the alpha-generating thalamocortical system is not reacting normally to visual input. Failure of alpha reactivity appears in several presentations and is a feature worth flagging. Conversely, the two conditions providing the same picture (no state-dependent change anywhere) indicates the recording did not capture genuine state variation, which raises questions about whether the conditions were properly elicited (was the client actually resting with eyes closed, actually alert with eyes open). The exam may frame this as "what does it mean when eyes-open and eyes-closed QEEG do not differ"; the answer is that the loss of the normal state difference is itself a finding, either a real reactivity failure or a recording-condition problem to rule out.
State effects extend beyond eyes-open/eyes-closed. Drowsiness is the single most common state confound in clinical QEEG, inflating frontal theta and the theta/beta ratio so that a sleepy recording mimics an attention phenotype. Alertness must be verified before those findings are trusted. A finding present in one state but absent in another is weaker than one that persists across states, which is why convergence across conditions is part of the pattern-reading discipline.
A single recording captures a state. A phenotype worth treating is a trait. Distinguishing the two is the interpretive work that separates competent QEEG from pattern-matching, and it runs on two questions: is this finding constitutional or regulatory, and will it reproduce on repeat assessment.
Constitutional versus regulatory. Constitutional phenotypes reflect fixed brain architecture (genetic expression, white-matter organization), are stable on test-retest, are present from early development, and change slowly if at all. The low-voltage-fast trait (heritability roughly 80 to 90 percent), individual alpha peak frequency (heritability around 80 percent), and overall spectral shape sit at this end. Regulatory phenotypes reflect current state management (arousal, vigilance, affective balance), fluctuate with state, respond to intervention more readily, and reproduce less reliably across single sessions. Frontal theta has a mixed character (a constitutional slow component plus a state-modulated fast component); high beta is largely regulatory and state-dependent; frontal alpha asymmetry is largely regulatory with poor single-session reliability. The distinction sets interpretive expectations: a constitutional finding describes the client's stable wiring and reproduces; a regulatory finding describes the client's current state and may not.
Situational findings. Some deviations are purely situational: drowsiness, acute anxiety during the recording, caffeine, sleep deprivation, a medication taken that morning. These are real in the recording but do not describe a stable feature of the brain. A situational finding misread as a trait launches treatment at a phantom. The guard is documentation (medications, sleep, caffeine, alertness, all captured before the recording) and the trait-versus-state reasoning that asks, for every finding, whether it reflects the client's wiring or the day's conditions.
One-time map versus serial assessment. A single QEEG is a snapshot. For findings of uncertain stability (regulatory phenotypes, anything recorded under possible state confounds, anything in a developmentally unstable adolescent), the appropriate response is repeat assessment under controlled conditions before committing to an interpretation, not a confident verdict from one recording. Constitutional-heavy presentations can be characterized from a single competent baseline. State-heavy presentations require multiple recordings. The exam expects you to match the number of recordings to the stability of the findings.
The single most common interpretive error in clinical QEEG underlies every phenotype in this chapter, so it gets its own section. The reverse-inference fallacy runs the evidence backward: observing a pattern and inferring the diagnosis, when the original evidence ran from diagnosis to pattern.
Valid forward inference says "people with ADHD show elevated frontal theta on average." Reverse inference says "elevated frontal theta means ADHD." The reversal is not logically warranted without knowing the base rate and specificity of that pattern across all conditions that could produce it, and for most QEEG patterns those are not favorable: elevated frontal theta appears in typical children, in drowsiness, in several conditions, and is sensitive but not specific. The error is not that the group difference is fake. It is that a group-level association does not license an individual-level diagnosis.
The theta/beta ratio is the worked example. The group finding (elevated theta in ADHD samples) is real. The field committed to diagnostic inference from it, and that commitment was the error. Recent reappraisals (Poil et al., 2024; the Strzelczyk et al., 2026 multiverse analysis across hundreds of analytic specifications) found that the theta/beta ratio has no diagnostic value for ADHD at the individual level, that previously reported group differences are highly contingent on analytic choices, and that some of the difference may reflect aperiodic spectral slope and individual alpha frequency rather than genuine oscillatory difference. The American Academy of Neurology advises clinicians not to rely on the theta/beta ratio to confirm or refute an ADHD diagnosis. The ratio retains value for treatment stratification (selecting a starting protocol from a baseline profile) but not as a diagnostic test. The Field Guide carries the full evidence trajectory.
The discriminant-function version of the same error: a TBI (or depression, or ADHD) discriminant trained on the condition versus healthy controls flags any brain that resembles the condition's training group, including brains with a different cause of the same pattern. A positive TBI discriminant means "this brain looks more like the TBI training group than like healthy controls," not "this person has a concussion." Before accepting any discriminant or phenotype as evidence of a condition, ask what else could produce this pattern in this person, and check medication, sleep, and medical history. The exam tests reverse inference directly; the clinic tests it daily.
How much a finding reproduces on repeat assessment determines how much weight it can bear. Test-retest reliability, expressed as the intraclass correlation coefficient, varies by metric, and the exam expects the rough ordering.
Absolute spectral power is the most reliable (intraclass correlation typically 0.7 to 0.9 for standard bands), higher at posterior sites and for broad bands, which is why it anchors phenotype detection. Individual alpha peak frequency is excellent (around 0.9), consistent with its high heritability, making it a reliable anchor. Relative power is good but lower than absolute (0.6 to 0.85), dragged down by denominator instability. The theta/beta ratio is moderate at best (0.5 to 0.75) and collapses when drowsiness is not screened. Coherence is moderate (0.5 to 0.7), with long-range less reliable than short-range. Frontal alpha asymmetry is poor on a single session (0.3 to 0.6) and needs multiple recordings. LORETA source estimates have limited reliability data and add model-dependent variance.
The clinical translation: a finding's reliability sets how confidently you can act on a single recording. A high-reliability finding (absolute power abnormality, slowed alpha peak frequency) can be characterized from one baseline. A low-reliability finding (asymmetry, isolated coherence, a drowsiness-sensitive ratio) requires multiple recordings under controlled conditions before it can be treated as a stable trait. Building a treatment plan on a non-reproducing finding is clinical error, and the way to avoid it is to know which metrics reproduce and which do not.
A separate stability axis is developmental: pediatric and adolescent findings can be developmentally unstable (puberty shifts alpha and beta), so a young client's map should be treated as provisional and repeated after puberty before committing to an extended interpretation, and apparent session-to-session change at those ages reflects maturation as often as anything clinical.
Comparing a person to a normative database asks "how does this client differ from the population." Comparing a person to their own prior recording asks "how has this client changed," and for tracking change the second question is the more reliable one. Intra-individual comparison eliminates between-person variation entirely, detects change more sensitively than normative comparison, and works regardless of whether the client was within normal limits at baseline. It is the most reliable application of QEEG.
What serial QEEG is for. Tracking medication effect, neurofeedback progress, concussion recovery, and age-related change over time. A baseline recording and a follow-up under matched conditions reveal what moved, in that individual, independent of where they sit relative to norms. The classic example: baseline frontal theta at 18 μV (z = +2.2) drops to 12 μV (z = +0.8) after a course of training, a one-third reduction clearly detected even though the client remains slightly above the norm. The change is visible in the intra-individual comparison whether or not the endpoint reaches the normal range.
The requirement that makes it valid. Serial comparison is only meaningful if the recordings are comparable, which means matching everything across sessions: the same montage and reference, the same conditions (eyes open, eyes closed), the same electrode positions, and a similar state (medications, sleep, time of day, caffeine). A change in protocol between recordings manufactures apparent change that is nothing but the protocol difference. This is the most testable point: an uncontrolled serial comparison measures the difference between two procedures, not a change in the brain.
What counts as a meaningful change. Not every difference between two recordings is real. Some is the metric's own test-retest variability. A meaningful change exceeds what that metric fluctuates by on its own, so the reliability ordering from Section 12.10 sets the bar: a shift in absolute power or alpha peak frequency (high reliability) is more credibly real at a given magnitude than the same nominal shift in coherence or asymmetry (lower reliability), which could fall within normal test-retest noise. A meaningful change also tends to show the convergence that characterizes a real finding: consistent across adjacent sites, present in both conditions, in a direction that makes sense given the intervention, and ideally corroborated by symptom or performance change. A single metric moving in a single condition, within the range of its own variability, is not yet a meaningful change. The exam frames this as "what does a meaningful change look like"; the answer is a change that exceeds the metric's test-retest variability, shows convergence, and was measured under matched conditions.
Read the report in order (absolute, relative, coherence, z-scores), favor coherent regions over isolated specks, and check the color scale before trusting a dramatic map. Treat each phenotype as a probabilistic association at a known evidence grade, expect two to four phenotypes at once, and forward to the atlas for depth. Read connectivity as rigidity (over-coherence) or disconnection (under-coherence), with the volume-conduction and proxy caveats. Read asymmetry only when it is reliable and artifact-cleared. Read source findings as model-based estimates, never measured locations. Read state into every finding, especially the eyes-open/eyes-closed alpha difference and the drowsiness confound. Separate situational from trait, constitutional from regulatory, and match the number of recordings to the stability of the findings. Refuse the reverse inference at every phenotype. And when the question is change rather than deviation, use serial QEEG under matched conditions, where a meaningful change is one that beats the metric's own variability and converges across the map.
Above all, keep statistical deviation and clinical significance in separate boxes. The z-score map shows where the client differs from the norm. Whether that difference matters comes from symptoms, performance testing, and history, not from the map. The report is the beginning of the interpretation, and the interpretation is the beginning of the clinical question, which is the deliverable the IQCB Diplomate is certified to produce and the next chapter teaches you to write.