Idealized waveform of auditory event‐related potential (ERP) recorded from the scalp to a brief stimulus such as a click. Upper tracing, amplified electroencephalogram (EEG). The ERP is not recognizable in the raw EEG and emerges gradually over many presentations of the auditory stimulus (S). Lower tracing, auditory ERP obtained by averaging many 1‐s epochs of EEG. Logarithmic time base allows visualization of early brain stem waves (I‐VI), midlatency components (N0, P0, Na, Pa, Nb), vertex potentials (P1, N1, P2), and task‐related endogenous components [Nd, N2, P300, and slow wave (SW)].

Exogenous and endogenous event‐related potentials (ERPs) recorded in 4 different experimental paradigms from a normal subject. In the first paradigm (top tracings), the subject's task was to listen to a sequence of tones and keep a running mental count of the number of improbable tones of higher pitch. The ERPs elicited by both probable and improbable tones contained N1 and P2 waves (open triangles above and below each tracing, respectively), whereas the ERP elicited by the improbable tone contained an additional P300 component (solid triangle). In the second paradigm, stimulus intensity was decreased to a sensation level (SL) of 40 dB; note the lack of change in endogenous P300 component. The third paradigm required the subject to count the number of times a tone was omitted from a repetitive train of stimuli. A P300 wave occurred after this omission, although there were no earlier N1‐P2 components. In the fourth paradigm, the subject ignored auditory stimuli and read a book; there was no late positive wave associated with ignored, improbable stimuli. For each condition, two replicated ERPs recorded from a central scalp electrode are superimposed.

Process of event‐related potential (ERP) averaging. First column, 16 single‐trial ERPs to 70‐dB sensation‐level tone. Second column, groups of 4 responses averaged together. In upper tracings of each pair, responses were added together, and in lower tracings they were alternatively added and subtracted so as to cancel the ERP and reveal the level of residual background noise—the socalled ± reference. Third column, average of all 16 responses, with the ± reference in lower tracing. Bottom of fourth column, the average of 64 responses to the tone. In upper right corner are superimposed two averages of 32 responses each, which illustrates replicability of the response. Amplitude calibrations (bottom) are adjusted according to square‐root scaling so that amplitude of background noise (as estimated by the ± reference) remains constant through the averaging process.

Electric fields produced by neurons. Left column, different anatomical configurations of neuronal aggregations. Right column, fields evoked during depolarization of cell bodies of these neurons. Top row, closed field. Current flows from peripheral, dendritic part of the nucleus to the depolarized cell bodies. Current stays within region of nucleus, and no field potentials are generated outside of its limits. Middle row, open‐field potentials generated in a cortical region. When cell bodies in deeper levels of the cortex are depolarized, a positive potential occurs at the surface and a negative potential occurs in the depth of the cortex. Bottom row, neuronal assembly with both open‐ and closed‐field properties.

Scalp distribution of auditory event‐related potentials (ERPs) in response to tones that provide feedback about performance on a perceptual task. The ERPs were recorded from 15 scalp locations using noncephalic reference (R). Negativity at scalp electrodes is shown by upward deflection. All ERP components were maximally recorded at vertex on this normal subject. The P3 wave was distinctly more widespread and more posterior in its scalp distribution than were the N1 or P2 components.

Distribution of pattern‐reversal visual evoked potentials over occipital scalp. A: averaged evoked potentials of 50 healthy subjects recorded from a chain of electrodes going from left to right around the posterior scalp. Potentials were recorded against a midfrontal reference, and negativity at occiput is represented by upward deflection. Middle traces, averaged waveforms; upper and lower traces, ±1 SD. Responses are shown to full‐field (left column), left half‐field (center column), and right half‐field (right column) stimulation. The major peak in the response is a positive wave occurring at 100‐ms latency. This positive wave is recorded maximally over left hemisphere for left half‐field stimulation and over right hemisphere for right half‐field stimulation. B: diagrammatic explanation for this paradoxical scalp distribution of the event‐related potential. Stimulation of left half field evokes a positive wave from the medial aspects of right occipital cortex. Its maximum amplitude appears over the left posterior scalp.

Scalp distribution of normal auditory event‐related potentials (ERPs) to binaural clicks, recorded using noncephalic reference. Four different scalp‐distribution maps are shown, each based on voltages recorded at a particular latency. Contours of each scalp distribution are plotted with broken lines representing negative potentials and continuous lines representing positive potentials. □, ERP recorded from an electrode over temporal scalp, to left of each map; Δ, ERP from a midfrontal electrode, to right of each map; * on waveform, the latency at which each scalp‐distribution map was calculated. Waveforms are plotted with scalp positivity upward. A: 2 scalp‐distribution maps for early part of the vertex N1 component, at latencies of 88 ms (top) and 78 ms (bottom); both suggest a dipole with negativity recorded in frontal regions and positivity recorded in temporal areas below the sylvian fissure. B: top, scalp distribution at the latency of the P2 component (170 ms), showing widespread positivity over frontal and central regions; bottom, scalp distribution at 115 ms, showing a negative potential with maximal amplitude over lateral scalp.

Event‐related potentials (ERPs) elicited during a visual target‐detection task, recorded from contacts on right midtemporal (RMT) and posterior‐temporal (RPT) depth electrode probes as well as from cranial pins inserted into the outer table of the skull at approximate locations C4 and P4. Recordings are from an epileptic patient being evaluated for possible neurosurgical intervention. The RMT electrode enters the brain anterior to motor cortex and terminates in anterior temporal lobe; RPT electrode enters the brain at the parieto‐occipital border and also terminates in anterior temporal lobe. Solid lines, ERPs elicited by counted, rare (P = 0.2) male names; dashed lines, ERPs elicited by frequent female names. Note late deflection in RPT probe at ∼600 ms that is positive in surface recordings and reverses polarity between depth probes 10 and 11 (posterior aspect of hippocampus) and again between probes 14 and 15 (anterior to hippocampus). Scalp recordings (top tracings) show long‐latency P300 waves. All ERPs were recorded simultaneously against a balanced sternovertebral reference. Calibration, 20 μV for depth recordings and 10 μV for cranial pin recordings, positivity upward.

Auditory evoked magnetic fields recorded from scalp probe placed at different locations over perisylvian cortex. Graphs, isofield contours for magnetic fields evoked by auditory tones with frequencies of 200, 600, 2,000, and 5,000 Hz. Horizontal axis, distance along a line connecting the ear canal and corner of the eye; vertical axis, distance along meridian connecting the ear canal and vertex. Arrows, estimated positions of equivalent‐current dipole sources for fields evoked by different tones. Calculated current sources are located in primary auditory cortex on the superior aspect of the temporal lobe. Distances between poles of the magnetic fields provide an estimate of the depth of current sources. Current sources responding to tones of low frequency are closer to the surface than those responding to high‐frequency tones. These current sources generate a dipole potential field recorded with different polarities above and below the sylvian fissure (cf. potential fields in Fig. 7A).

Evaluation of hearing impairment in 3‐mo‐old infant with brain stem auditory evoked potentials. Evoked potentials were recorded in response to monaural clicks presented at a rate of 61/8 at various intensities above normal adult threshold (nHL). Each tracing represents the average of 4,000 individual responses recorded between vertex and mastoid, with negativity at the vertex upward. Triangles, wave V in the brain stem responses. Responses to left ear stimulation are recognizable to 20‐dB nHL, whereas responses to right ear stimulation can be recognized only to 50‐dB nHL. These results indicate mild to moderate hearing impairment in the right ear.

Use of evoked potentials in localization of lesions of sensory pathways. Auditory evoked potentials were recorded from a patient presenting with right‐sided sensorineural hearing loss. Brain stem auditory evoked potentials were recorded between vertex (Cz) and either left (M1) or right (M2) mastoid in response to clicks at 80 dB above normal adult threshold presented at a rate of 11/s. Triangles, three components (waves I, III, and V) in vertex‐to‐ipsilateral mastoid recordings. Response to right ear stimulation shows delay between wave I and wave III that exceeds normal limit of 2.6 ms. This indicates some dysfunction between cochlea and pons. On further examination, patient was found to have right‐sided acoustic neuroma.

Use of event‐related potentials (ERPs) in the demonstration of subclinical disorders. Visual ERPs were recorded in response to pattern reversal of a checkerboard stimulus. Evoked potentials were recorded between the occiput and a midfrontal reference, with negativity at occiput represented by upward deflection. A prominent component in the ERP is a positive wave with a normal latency of ∼100 ms (triangles). In this patient, its latency is normal for left eye stimulation but is severely delayed for right eye stimulation. Patient presented with weakness of left leg and no history or signs of visual impairment. The ERPs were clinically helpful in arriving at a final diagnosis of multiple sclerosis.

Use of event‐related potentials (ERPs) in prognosis. These somatosensory evoked potentials were recorded from a patient who was still comatose 1 wk after severe closed head injury. Responses were evoked by electrical stimulation of left and right median nerves. Waveforms were recorded after a delay of 4 ms to eliminate electrical‐stimulus artifact. Top tracings: from Erb's point (EP), normal peripheral nerve response at 10 ms latency. Middle tracings: from the neck over vertebra prominens (CV7) against a frontal (Fz) scalp reference. A clear negative component is recorded with peak latency of 14 ms. Bottom tracings: from electrodes located over primary somatosensory areas (C3′/C4′); no definite cortical responses were recorded. Normally there is a prominent negative wave at 20 ms. Absence of any cortical response was considered a poor prognostic sign. Patient continued in chronic vegetative state 1 yr after accident.

Schematic illustration of movement‐related potentials accompanying voluntary hand movement. [Components are labeled according to terminology of Vaughan and colleagues 508 and Kornhuber and Deecke 234.] Also shown is integrated muscle activity [electromyogram (EMG)] of the responding arm. The event‐related potentials are recorded from the scalp overlying sensorimotor cortex and are averaged with respect to onset of EMG activity at time zero. RP, readiness potential; PMP, premotion positivity; MP, motor potential; RAF, reafference potential.

A: contingent negative variation (CNV) (shaded area) arising in interval between warning flash (S1) and subsequent tone (S2) to which the subject made a prompt motor response. B: CNV developing in interval between warning flash (S1) and near‐threshold tone (S2), which was presented on a random 50% of trials. The CNVs were averaged separately on trials in which the subject correctly detected the tone (hits) and on trials in which tone absence was correctly reported (C.R.). The CNVs were averaged over 15 and 30 trials in A and B, respectively. Vertex recording from normal adult subject.

Paradigm for demonstrating early event‐related potential (ERP) changes with channel‐selective attention. Randomized sequences of tones are delivered to left (800 Hz) and right (1,500 Hz) ears at interstimulus intervals shown on upper axis. Asterisks, slightly deviant target tones that subject attempts to detect in one ear at a time. Typical averaged ERPs to tones in each ear are shown as a function of attend‐left and attend‐right conditions. Shaded area, difference waveform between ERPs to attended and unattended tones, is called the negative‐difference (Nd) component.

Averaged event‐related potentials (ERPs) associated with selective attention to 1 of 2 channels of tones distinguished by frequency cues alone (300 Hz vs. 700 Hz). High‐ and low‐frequency tone pips were presented in random order at a rapid rate (∼3/s). Subjects attended to one channel at a time and attempted to detect targets of slightly longer duration. Attended‐channel tones elicited a broad negative ERP, seen most clearly in the attended‐minus‐unattended difference wave (right). The ERPs recorded from frontal (Fz), central (Cz), and parietal (Pz) scalp areas were averaged over several hundred stimulus presentations using a computer routine that extracts overlapping time epochs for stimuli presented at short intervals.

Event‐related potentials recorded from central scalp area in 3 subjects (KM, SH, and KK) in response to standard stimuli presented in 2 patterns under 3 attention conditions. The LHL pattern occurs when the tone changes in pitch from low to high and back again to low as it moves across auditory space from one ear to the other. The HLH pattern changes in frequency and direction in the opposite manner. Note increased broad negativity elicited by attended tones.

Grand‐average event‐related potentials from 12 normal subjects in a two‐channel selective‐listening task in which high‐frequency tones were delivered from a speaker on the right and low‐frequency tones from a speaker on the left. Interstimulus intervals were 250–550 ms. Subjects listened selectively for occasional shorter‐duration targets in one channel at a time. Note that target tones in attended channel elicit a large P300 wave parietally (Pz) and a smaller slow wave (S.W.) frontally (Fz), whereas targets in unattended channel do not. Basic negative‐difference (Nd) attention effect (greater negative amplitude for attended‐channel tones) is seen for both standard and target tones.

Poststimulus time (PST) histograms of auditory cortex single unit for a sequence of nonattend and attend conditions. Monkeys were trained to press a lever in response to sounds in one ear and ignore sounds in opposite ear. Bar graphs represent the mean evoked activity in each PST histogram over a period of 20–200 ms after stimulus onset and spontaneous activity for the 100‐ms period preceding stimulus onset. Behavioral conditions are arranged in chronological sequence beginning at top for histograms and at left for bar graphs. Stimulus: 8.0 kHz, 65 dB sound pressure level. Stimulus duration indicated by bar below time axis.

Grand‐average event‐related potentials (ERPs) from 12 normal subjects in response to standard (nontarget) flashes located at the left position. Solid tracings, ERPs during attend‐left condition; dashed tracings, ERPs during attend‐right conditions. ERPs are shown for midline frontal (Fz), central (Cz), and occipital (Oz) scalp sites and for lateral parietal sites on left (LP) and right (RP). Mirror‐image attention effects were elicited by right‐field stimuli (not shown).

Comparison of effects of visual‐spatial attention upon unit activity in posterior parietal cortex of a rhesus monkey and upon event‐related potentials (ERPs) recorded from parietal scalp contralateral to visual field of presentation of a human subject. In the monkey experiment 538, the animal was trained to fixate a central spot while a peripheral flash elicited unit discharge in contralateral parietal lobe. Two poststimulus time histograms (PSTH) compare conditions when evoking peripheral flash was not attended vs. when it was attended. Human ERPs were recorded from left parietal scalp electrode in response to right‐field flashes under conditions in which those flashes were attended and when attention was directed to opposite field. Note that modulation of evoked activity occurs with similar time course and magnitude in the 2 species.

Event‐related potentials (ERPs) during somatosensory selective attention to random sequences of shock stimuli applied to 4 fingers. Subjects counted target stimuli to the third finger of either left hand (thicker traces) or right hand (thinner traces). The attended finger is represented in black; small arrow points to stimulated finger that evokes the ERP in question. Tracings: A, D, vertical eye movement controls; B, E, somatosensory ERPs recorded from contralateral parietal scalp (Sc) and elicited by stimuli applied to the third (B) or second (E) fingers of left hand; C, F, corresponding ERPs recorded simultaneously from ipsilateral (Si) parietal electrode. The N140 wave is larger contralaterally for both target and nontarget stimuli to the left hand; the P300 (P400) wave is symmetrical and occurs only after target stimuli. Early P45 component appears only contralaterally in B and E and is not affected by task.

Visual event‐related potentials (ERPs) to checkerboards with different check sizes, one of which was attended. Difference waves were formed by subtracting ERPs to each stimulus under different conditions of attention. Left, ERPs to each of 8 stimuli when 12′ check was attended minus ERPs to same stimuli when diffuse light was attended. Difference potential to diffuse light is inverted because the ERP to the attended stimulus was subtracted from the ERP to the nonattended stimulus. Right, same as left except that 35′ checks were attended. Note enhanced N150–350 to attended check sizes.

Information delivery and the P300 component. Single or double clicks were presented at soft or loud intensity. In one condition the subjects judged whether the stimulus was soft or loud; in a second condition they judged whether it was single or double. Averaged event‐related potentials (ERPs) recorded between vertex and earlobe are shown for one subject. Late positive wave (P300) to first click was larger when the subject was judging soft or loud than when judging single or double. When subject was guessing single or double, a second positive wave occurred 300 ms after second click or at the point in time when second click would have occurred in the case of a single stimulus.

Stimulus probability and the P300 wave. Event‐related potentials (ERPs) were recorded from midparietal electrode during sequences of high‐ and low‐frequency tones. Probabilities of the 2 tones were varied across experimental conditions as indicated. Subject's task was either to count the number of high‐frequency tones in a session or to ignore the stimuli. Large positive wave with peak latency of ∼350 ms occurred in response to more improbable stimuli. The amplitude of this positive wave in response to counted tones was only slightly larger than that in response to uncounted tones at the corresponding level of probability.

Effects of stimulus sequence on P300 component. Two tones of different pitch were presented in random sequence with each tone having a probability of 0.5. The event‐related potentials (ERPs) to any particular stimulus (A) were averaged according to whether preceding stimuli were of the same frequency (A) or of the other frequency (B). From left to right: ERPs were averaged according to longer preceding sequences. The P300 wave was larger when the stimulus was preceded by tones of different frequency than when it was preceded by tones of the same frequency.

Event‐related potentials (ERPs) recorded from vertex of a normal subject during threshold‐level auditory signal‐detection task. ERPs were averaged separately according to signal‐presence or signal‐absence (noise) trials and according to the subjects' rated confidence 1,2,3,4,5,6,7,8 in their judgment. Ratings 1–4 signified decreasing confidence in signal presence; ratings 5–8 signified increasing confidence in signal absence. Lower tracings, ERPs averaged across all 8 ratings. Numbers beside each waveform indicate number of trials included in that averaged ERP. FA, false‐alarm responses; CR, correctly reported tone absence.

Event‐related potentials elicited by near‐threshold tones presented on a random basis at specific times after warning flash (W). On each trial there could be either 0 (top left tracing), 1 (second through fourth left tracings), 2 (top right tracings), or 3 (second through fourth right tracings) tones presented at times indicated by *. Subjects reported the number of tones perceived after a subsequent response cue (R) appeared. Arrows, Sa, Sb, and Sc, time points at which tones might occur. The Sb stimulus could occur 300, 600, or 900 ms after the Sa stimulus. Grand‐average vertex recordings from 5 subjects.

Effect of discrimination difficulty on the event‐related potential (ERP). Subjects listened to a tone sequence consisting of 80% standard tones and 4 different target tones, each having a higher pitch than the standard tone and a probability of 5%. Subjects made a reaction‐time (RT) response upon detecting each target. Latency of both P300 wave and RT increased with increasing difficulty of discriminating target. Amplitude of P300 wave was also reduced, probably because of decreased confidence of decisions. Fz, midline frontal scalp site; Cz, central scalp site; Pz, parietal site.

Single‐trial P300 latency vs. reaction time (RT) on the same trials for accurate and speed RT conditions. The task was to make a semantic judgment on each of a series of words, such as deciding if the word was a synonym of a key word, and to press a button when such targets were detected. Large crosses indicate trials on which errors were committed. Observed error rate was 3% for accurate RT condition and 9% for speed RT condition.

Developmental changes in human event‐related potential (ERP) elicited by novel stimuli. Recordings were obtained from subjects of different ages in response to occasional bizarre and unpredictable stimuli during performance of a simple visual‐discrimination task. Recordings are from midfrontal (Fz) and midparietal (Pz) electrodes, with positivity represented by upward deflection. The P300 (P3) wave is maximally recorded from the parietal electrode and decreases in latency with increasing age. Younger children show an enlarged negative‐positive complex (Nc‐Pc) recorded from the frontal region in response to novel stimuli.

Effects of aging and dementia on latency of P300 wave. Top, P300 latencies of normal subjects. Regression lines show limits of normal range. Middle, P300 latencies of patients with dementia. Bottom, P300 latencies of nondemented patients.

Recordings of averaged readiness potentials preceding speech (K‐words) and oral‐facial motor activity (cough). Recording sites are vertex, left (LIF) and right (RIF) inferior frontal regions, left (LPC) and right (RPC) precentral regions, and cheek (glossokinetic). Note absence of left‐right asymmetry in negative readiness potential preceding speech and presence of artifact in glossokinetic channel.

Auditory event‐related potentials (ERPs) in response to spoken words in a passage of poetry (solid tracings) and to a sequence of tone bursts that were matched to the poetry in loudness and interstimulus intervals. Averaged ERPs were time locked to word or tone onsets. Subjects listened to poetry in order to answer context‐related questions afterward but listened passively (no task assignment) to tones. Note that ERPs to tones are approximately symmetrical in recordings from left and right hemispheres (over Wernicke's area and its right hemisphere homologue, respectively), whereas incremented negativity to attended speech sounds (shaded area) was larger over left hemisphere.

Grand‐average event‐related potentials (ERPs) from left and right temporoparietal scalp sites to each of the words in 7‐word sentences. Columns show ERPs of all right‐handed (R.H.) subjects (left) and of right‐handed subjects without (center) and with (right) left‐handed (L.H.) relatives in their immediate family. Next to each ERP pair is the P value for matched‐pair t test comparing left and right hemisphere amplitudes between 400 and 700 ms poststimulus.

Left: amplitude ratios for auditory probe evoked N1‐P2 peak/peak component comparing phonetic task (listen for target syllable in conversation), prosodic task (detect emotional intonations in speaker's voice), and control task (attend to probe click). Note relative probe ERP (event‐related potential) reduction over left hemisphere in phonetic task and reverse for prosodic task. Right: ERP waveforms from 2 subjects showing task‐related changes in probe evoked N1‐P2 over left (L.H.) and right (R.H.) hemispheres.

Event‐related potentials (ERPs) elicited by semantically and physically incongruous words. A: timing of word presentations for 3 sample sentences and typical ERP waveforms recorded over 7‐word sentence. In each comparison (B‐D), grand‐average ERPs for normal (solid line) and deviant (semantic, dashed line; physical, dotted line) seventh words are superimposed. The 300‐ms region used for quantitative analyses is indicated by shading. The means and SE of this difference area, formed by subtracting normal seventh‐word ERP from deviant seventh‐word ERP, are given in each bar graph. Fz, midline frontal scalp site; Cz, central scalp site; Pz, parietal site.

Grand‐average event‐related potentials (ERPs) from 17 subjects to semantically anomalous words at intermediate and terminal positions of sentences in text (dashed lines). Superimposed waveforms (solid lines) are ERPs elicited by semantically congruent words at corresponding positions. Content words that immediately preceded intermediate semantic anomalies were chosen as congruous words for these comparisons.

Grand‐average event‐related potentials (ERPs) to visual test words that either did (solid line) or did not (dashed line) fit with a semantic category that had been specified by a preceding phrase. Note enhanced negativity peaking at 400 ms (N400) to words that did not match the category. Cz, midline central scalp site.

Topographic mappings of scalp areas at which indicated event‐related potential (ERP) components were differentiated between groups of normal and dyslexic subjects. The VEP 282 refers to a mapping of the visual evoked potential (elicited by flashes) measured at a latency of 282 ms, and AEP 198 refers to a mapping of the auditory evoked potential (elicited by clicks) measured at 198 ms. The TTAEP is the auditory evoked potential to verbal stimuli measured at indicated latencies. Note that maximal between‐group differences are localized mainly over posterior scalp and over left hemisphere.