Event-Related Potentials-1

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Event-Related Potentials

NN Thigpen and A Keil, University of Florida, Gainesville, FL, United States


Ó 2017 Elsevier Inc. All rights reserved.

Event-Related Potentials: Introduction and Overview 1


Neurophysiological Underpinnings of Event-Related Potentials 2
Measuring Event-Related Potentials From Electroencephalogram Recordings 2
An Example Experiment 3
Plotting Event-Related Potentials 3
The Components of the Event-Related Potential: Repeatable Patterns in Averaged Brain Waves 4
P300 4
Brainstem Auditory Evoked Potentials 5
C1 5
P100 5
Readiness Potentials 5
Eye Fixation–Related Potential 6
Steady-State Evoked Potentials 6
Advanced Analysis of Event-Related Brain Electric Signals 6
Conclusions and Future Directions 6
Appendix A Supplementary data 6
References 6
Further Reading 7

Event-Related Potentials: Introduction and Overview

The function of the human brain relies on the interactions of billions of brain cells called neurons. Neurons communicate through
electrochemical processes, which change the electrical properties of the tissue in and around them. This communication is critical
for human experience and behavior. Across a wide range of scientific disciplines, researchers have therefore used recordings of neural
electrical activity from the human brain to examine cognitive, emotional, and motor processes characterizing human behavior.
Recordings of brain activity may also assist in clinical diagnosisdand even in the treatmentdof brain dysfunctions.
Recently developed neuroimaging procedures, based on brain metabolism and blood flow, provide finer-grained images of the
brain. These techniques (for example, functional magnetic resonance imaging, fMRI) highlight brain structures that differentially
respond to a given situation or vary in their reactivity in different groups of people such as healthy versus patient samples. However,
brain metabolism and blood flow are many times slower than the electrochemical interactions between neurons that characterize
the function of the living brain. Furthermore, the relations between primary neural activity and metabolism and blood flow are not
fully understood and are indirect and complex in nature. As a consequence, recordings of electrical brain activity using electroen-
cephalography (EEG) are widely used.
EEG recordings of electrical brain waves have many desirable properties: They possess high temporal resolution/precision, they are
directly sensitive to the electrical interaction between neurons, are more affordable than hemodynamic imaging techniques, are
widely available, and impose little if any discomfort to research participants. Thus, EEG-based methods have found wide
application in the cognitive and neural sciences (Handy, 2005). A weakness of EEG-derived measures is their poor spatial
specificity, however. With few exceptions, it is difficult to attribute a given EEG phenomenon to activity in a particular brain structure.
After German physician Hans Berger, first discovered in 1924 that brain electric activity could be measured through the scalp
using EEG electrodes (see, Berger, 1969; for an English translation), it took only a little more than a decade for other researchers
to examine the relationship between changes in the EEG readings and specific events. As early as the 1930s, Pauline and Hallowell
Davis began comparing individual EEG segments that were time locked to different events (Davis, 1939), an approach that even-
tually led to what is now known as the technique of event-related potentials (ERPs). ERPs are extracted from the ongoing EEG
activity by averaging EEG segments that are cut out of the continuously recorded EEG such that each segment shares the same
temporal relationship with an event of interest (Donchin et al., 1977). This process is explained in detail below. Thus, ERPs repre-
sent the prototypical neural activity following (or preceding) the onset of a specific eventda sensory stimulus (a word, sound,
touch), a motor behavior, or a cue prompting covert mental operations (eg, mental imagery, the anticipation of an event). Some-
times, ERPs evoked by sensory stimulation are referred to as evoked potentials, emphasizing their causal relation with a salient elic-
iting event.
There are few limits to the use of the ERP research in research and clinical practice. Not surprisingly, scientists have used ERPs to
study a wide spectrum of brain processes in healthy and patient samples, resulting in a steadily growing body of knowledge
regarding the timing and extent of brain electric changes in a variety of contexts. Here, we will consider the physiological basis,
applications, different variants, example studies, and future directions of the ERP technique.

Reference Module in Neuroscience and Biobehavioral Psychology http://dx.doi.org/10.1016/B978-0-12-809324-5.02456-1 1


2 Event-Related Potentials

Neurophysiological Underpinnings of Event-Related Potentials

ERPs reflect, mostly, activity from the cerebral cortex, the thin outer layer of the brain that is often called gray matter, containing
densely packed neurons including their cell bodies. By contrast, white matter (connections between brain areas) as well as deep
brain regions such as the thalamus and brainstem, does not strongly contribute to the EEG/ERP signal. Researchers now widely
agree that the majority of brain electric (EEG) signals measured at the scalp are due to synaptic activity at dendrites, the tree-like
structures at which neurons receive inhibitory or excitatory input. Thus, the EEG reflects communication between groups of neurons
rather than the “firing” of individual neurons as expressed in all-or-nothing action potentials (Nunez, 1989).
Further contributing to the robustness of the EEG signal, apical dendrites (emerging from the top, or apex, of pyramidal
neurons) tend to be aligned in parallel to each other and perpendicular to the cortical surface, which enables them to collectively
generate significant electrical fields. Pyramidal cells account for 80–90% of all cortical neurons, with some variance between brain
regions, thus representing a major portion of the neuronal population. Their number is a critical factor: Although electromagnetic
field changes of a single pyramidal neuron cannot be detected through the layers of cerebrospinal fluid, skin, and scalp that separate
the neural tissue from the electrodes, it is easily possible to measure the activity that is evoked by millions of favorably oriented
pyramidal neurons (Nunez and Srinivasan, 2006).
A minimal condition for reliable recordings to be visible outside the brain is the synchronized activity of several tens of
thousands of neurons oriented in the same direction. However, given the folded surface of the cortex, neurons can be aligned
in any direction relative to a given electrode. Voltage gradients generated in the brain are further distorted by their passing
through other brain tissue, meninges, skull, and scalp. Thus, by the time a cerebral electrical signal reaches the scalp, the loca-
tion of the voltage change may not reflect the location of the active cortex. EEG/ERP readings nevertheless represent a tempo-
rally precise representation of transient synaptic activity at the level of neural populations, often called “macroscopic” activity
(compared to microscopic processes occurring between or within individual neurons). To summarize, scalp-recorded ERPs
reflect neural mass activity in the near-surface gray matter, specifically the integrated synaptic afferent processes at dendritic
trees of pyramidal neurons.

Measuring Event-Related Potentials From Electroencephalogram Recordings

Modern EEG technology affords the recording of very small brain electric changes (at the order of microvolts, mV, ie, millionths of
a Volt) with sensor arrays often covering wide areas of the scalp, face, and neck using sensor counts between few (eg, three elec-
trodes) and hundreds (eg, 512) electrodes. Special sensors (electrodes) and highly sensitive amplifiers are required to measure
the small signals generated by the brain. Frequently used today, EEG setups with 64 sensors or more are referred to as dense-
array EEG systems. In these systems, sensors are typically arranged in structures that resemble caps or nets, which can be worn
and adjusted more easily and rapidly, compared to electrodes that are mounted individually. For example, the sensor array in
Fig. 1 contains 129 electrodes covering the head and neck regions.
Most EEG setups require participants to be in an electrically shielded, controlled environment and to avoid even small move-
ments of the body, head, and eyes. A recent trend in the field, however, has been to develop EEG for use in real-world settings.

Figure 1 Example electroencephalography array. Head models are shown that approximate the locations of 257 sensors. Video clip 1 provides the
approximate location of all sensors (shown in blue) on the head model.
Event-Related Potentials 3

These ambulatory EEG systems are increasingly affordable even for consumer-grade applications, and they allow participants to
record the EEG while in uncontrolled environments and when walking. Ambulatory EEG has been useful in studies ranging
from monitoring drowsiness in truck drivers during long distance commutes to monitoring daily activity in seizure patients.
Once the continuous EEG has been collected, the signal itself is rich in time information, and its digital sampling is limited only
by computer storage and other technical constraints, making it possible to obtain brain electric recordings at submillisecond (ie, one
thousandth of a second) accuracy. To extract ERPs from these recordings, EEG segments surrounding an event are extracted, and
these brief time segments are averaged together across like events. For example, two types of events (often called “trials”) may repre-
sent the instantiations of two conditions in a study comparing brain responses to two different visual patterns. The ERP technique is
best illustrated by an example experiment.

An Example Experiment

To investigate neurophysiological differences when seeing two different visual patterns, a researcher might record EEG from a subject
while they viewed the two patterns shown at random with several seconds in between them. In our example, one pattern has dense
bars, the other, sparse bars (see Fig. 2). From the continuous data, the researcher then cuts out EEG segments 400 ms before and
800 ms after the onset of each dense and sparse pattern, producing a total of 40 segments. Averaging the 20 “dense” segments
produces one averaged ERP, representative of a prototypical “dense pattern” response, and the 20 “sparse” segments form a second
averaged ERP. As shown in Fig. 2, this results in two waveforms that are 1200 ms long, at every EEG sensor. Trial averaging is essen-
tial both because the ongoing EEG is noisy and because the amplitude of changes to specific events is on the scale of several micro-
volts, whereas the EEG involves voltage fluctuations that can be hundreds of microvolts in magnitude. After averaging, the ERP
represents the part of the signal in each trial that is time locked and phase locked to the event, which means that the peaks and
troughs of the waveform in each trial must occur at the same time, in the same direction (positive or negative). This procedure
results in an average waveform at each EEG sensor, representing the average patterns of brain activity surrounding (in our example:
following) the events.

Plotting Event-Related Potentials

ERPs can be plotted across the time domain and in the spatial domain across the scalp if sufficient numbers of sensors were used
during the recording. As shown in Fig. 1, ERPs at a selected EEG sensor are typically plotted across time, where time zero denotes
event onset, and the y-axis represents microvolt changes relative to zero (where zero represents the average voltage across a suitable

Figure 2 Typical averaged event-related potential (ERP) waveforms elicited by two similar visual stimuli (shown in the key) are plotted at two
sensors: Pz and Oz. The topographies (or spatial distribution of voltages) at the peak of each ERP component (here, the P1, N1, P2, N2, and P3) are
shown for the red condition. Video clip 2 provides a model of the topographical voltage distribution across time. These waveforms represent the
average voltage measured at two midline sensors, one over parietal cortex (Pz) and the other over occipital cortex (Oz). Note how the shape (and to
some degree the latency) of each waveform varies between these two scalp locations.
4 Event-Related Potentials

baseline segment). Further information often found on ERP figures includes labels that denote the location of the sensor on the
head. In the most widely used version (the so-called international 10–20 system, Jasper, 1958), they are a combination of letters
(indicating the scalp region where the electrode is located, eg, T for temporal, P for parietal, C for central, O for occipital, and F for
frontal) and numbers that indicate the hemisphere (even numbers are on the right) and distance from the top of the head (with
increasing numbers denoting greater distance).
To visualize the spatial distribution of voltage across a number of EEG sensors, the ERP voltage at a particular time point (or across
an average of several time points) can be mapped onto a head volume as shown in Fig. 2. A typical application for ERPs is the compar-
ison of two voltage time courses (ie, ERPs evoked by two types of events) as they unfold in time: In our example experiment described
earlier, researchers asked participants to press a key when they saw the dense pattern (the ERP waveform of which is shown in blue).
Comparing the red (ERP in response to the sparse pattern) and blue waveforms over time, researchers can identify time points of diver-
gence or similarity. To aid in the description of time segments and ERP peaks and troughs, labels are typically used to denote systematic
deviations from the baseline, which are referred to as ERP components. They will be discussed in the next paragraph.

The Components of the Event-Related Potential: Repeatable Patterns in Averaged Brain Waves

The ERP waveform typically consists of a sequence of positive and negative voltage deflections, which vary in their magnitude
(amplitude) and topographical distribution across the scalp at a given point in time (Kappenman and Luck, 2012). The shape
of the waveform and its scalp distribution depend on the nature of the event, the nature of the task, as well as being dependent
on the participants (eg, their age or their health status). The ERP waveform may be divided into temporal segments, called compo-
nents, based on the peaks and troughs apparent throughout the ERP waveform. Many components are defined only in the context of
a specific stimulus type or in the context of a specific experimental paradigm during which it is recorded. For example, in 1965,
Rainer Spehlmann first noticed that presenting a visual stimulus to subjects leads to their scalp voltage increasing at around
80–120 ms after stimulus onset, an ERP component now termed the P100.
The latency of ERP components has been used to categorize and understand the time dynamics of brain processes. The earliest
components visible in the waveform are often considered to reflect sensory processes because they are highly sensitive to brain activity
associated with the physical features of a component (eg, color, size, shape, location, orientation, etc.). Many different ERP compo-
nents have been described in the literature, known to exist at specific scalp locations and during specific time ranges, in the context
of well-defined experimental tasks (Luck, 2014). Although it is impossible to give a complete list of all ERP components, we provide
an overview of the most widely studied visual and auditory components, respectively, in Table 1 and Table 2.

P300
Probably the most widely studied ERP component, the P300 (or P3), is evoked in the “oddball” paradigm, where a series of stan-
dard (eg, identical) stimuli are occasionally interrupted by a deviant stimulus (the “oddball” stimulus). The oddball produces
a larger positive-going amplitude around 300 ms after stimulus onset, compared to the more frequently presented stimuli
(Chapman and Bragdon, 1964) This response is thought to reflect the cognitive processing of rare or novel stimuli. This P300 effect
has been consistently observed regardless of whether the rare stimulus was visual, auditory, somatosensory, or even olfactory. The
P300 has also been thought to reflect working memory load and stimulus evaluation time. Many studies provide evidence that the
P300 component may be used as a biomarker for a number of health conditions (depression, anxiety, alcohol dependence, and
conduct disorder, among others).

Table 1 Visual event-related potentials

Component Typical peak latency


name (ms) Neurophysiological origin Sensitive to

C1 40–80 Response in the calcarine fissure (¼lower tier visual Physical stimulus features, some forms of spatial
cortex) selective attention
P1 80–120 Arrival of cortical activation in extrastriate cortex Spatial selective attention, some forms of feature-
based attention
N1 140–190 Spread across the visual system, reentry from Feature-based attention, match or mismatch with
frontal cortices expected semantic content, priming, face
processing
P2/N2 200–280 Engagement of wide cortical areas, occipitoparietal Higher-order attentive processes, semantic content,
focus of activity configuration, context
P3 300–500 Contributions from widely distributed cortical and Stimulus frequency, memory, attention capacity,
subcortical regions and many other higher-order task requirements
Late positivities >400 Contributions from widely distributed cortical and Motivation, emotion, attention, competition, many
subcortical regions other higher-order task requirements
Event-Related Potentials 5

Table 2 Auditory event-related potentials

Typical peak
Component name latency (ms) Neurophysiological origin Sensitive to

Brainstem auditory evoked potentials: 1–10 Responses in the deep nuclei in the brain Physical stimulus features, temporal
Labeled I, II, III, IV, V stem; require thousands of averages properties of the stimulus sequence
(stimulus repetitions)
Middle latency components: 10–40 Subcortical and primary auditory cortical Physical stimulus features, temporal
No, Po, Na, Pa, Nb, and neural activity; require several properties of the stimulus sequence
hundreds of averages
Pb or P50 50 First pronounced response of (primary) Attenuated for the second of two rapidly
auditory cortex sounds, spatial attention
N1 70–110 Engagement of wide cortical areas, Higher-order attentive processes,
occipitoparietal focus of activity semantic content, configuration,
context
P2 120–160 Contributions from widely distributed Stimulus frequency, memory, attention
cortical and subcortical regions capacity, and many other higher-order
task requirements
N2 Around 200 ms Contributions from widely distributed Motivation, emotion, attention,
cortical and subcortical regions competition, many other higher-order
task requirements
P3 >250 Contributions from widely distributed Motivation, emotion, attention,
cortical and subcortical regions competition, many other higher-order
task requirements

Brainstem Auditory Evoked Potentials


Brainstem auditory evoked potentials (BAEPs) are a sequence of very early ERP responses following an auditory stimulus, with the
first deflection occurring as early as 1–2 ms after the onset of an auditory stimulus. Given this early latency, the neural substrate
thought to underlie BEAPs is brainstem regions in the path to auditory cortex, such as the auditory nerve, the cochlear nucleus,
the superior olivary nucleus, and the inferior colliculus. Given the origin of these signals deep inside the brain, and the small volt-
ages this ERP typically shows, averaging thousands of trials is typically required to obtain clear BAEP waveforms. BAEPs have been
used to understand the earliest neural activity following a sound in typically functioning adults and also those with brainstem
damage. Brainstem damage in certain areas is associated with different waveform shapes in BAEPs.

C1
The C1 is the earliest visual evoked component, peaking 65–90 ms following a nonfoveal stimulus over occipitoparietal areas of the
scalp (Jeffreys and Axford, 1972). The C1 is thought to be the result of activity in the primary visual cortex. The C1 reverses polarity
based on stimulus location, such that stimuli in the upper visual field produce a negative change in voltage, and vice versa, reflecting
the anatomy of primary visual cortex: Here, neurons sensitive to upper and lower visual field stimulation are located at opposite
sides along the upper and lower lips of the so-called calcarine fissure, a structure in the posterior part of the brain containing primary
visual neurons (Foxe and Simpson, 2002; Stolarova et al., 2006). The C1 component is unique in that it provides an index of the
initial afferent volley into the retinotopic visual cortex, allowing a method to noninvasively study a particular subgroup of human
neurons at exquisite temporal resolution.

P100
Like the C1, the P100 is considered an early visually evoked potential, often associated with sensory processing and thought to be
the result of activations in extended areas in the occipital (visual) cortex. It peaks between 80 and 120 ms poststimulus over the
lateral occipital scalp, contralateral to the visual field of the stimulus. Many studies have found a larger P100 amplitude when visual
stimuli are presented at attended locations, compared to unattended locations. Thus, the P100 has been used in a large body of
studies of visual spatial selective attention (Hillyard and Anllo-Vento, 1998).

Readiness Potentials
The readiness potential (also called the Bereitschaftspotential or the premotor potential) reflects activity prior to a motor response.
In figures showing readiness potentials, the ERP time zero does not indicate a stimulus onset but instead the initiation of a motor
movement. Two deflections typically seen prior to movement onset are termed the early (peaking around 1000 ms) and late
(peaking around 300 ms) readiness potentials. Proceeding unilateral hand and foot movements, these potentials often appear
6 Event-Related Potentials

on the contralateral side of the scalp, reflecting the contralateral organization of motor cortex. Ipsilaterally or bilaterally controlled
movements (ie, face and tongue movements) are not lateralized in the readiness potential.

Eye Fixation–Related Potential


The eye fixation–related potential (EFRP) is a special case of motor-related ERPs where instead of stimulus onset serving as the
temporal reference point for averaging (time zero), the reference point is subject driven. Specifically, an eye tracker detects when
a fixation is made, and a time window from the EEG recording is extracted before and after each eye fixation, followed by averaging.
EFRPs have been widely used to investigate early brain processes involved in reading but also have been used to examine brain
activity during scene processing, while playing video games or while using a brain–computer interface.

Steady-State Evoked Potentials


Another special case of the ERP is the steady-state evoked potential (SSEP). These waveforms can be elicited by rapidly modulating
the intensity (or another dimension) of a visual, auditory, or tactile stimulus many times a second. In response to the oscillatory
stimulation, the sensory cortices respond with an oscillatory waveform at the same frequency as the driving stimulus (eg, a square
flashing 20 times per second results in an SSEP waveform oscillating at 20 Hz over the occipital part of the brain, where visual cortex
resides). Many applications for SSEPs exist, including the possibility of using multiple concurrent stimulus presentations at different
frequencies. The brain responses to the different concurrent stimuli may then be extracted from the EEG signal at different frequen-
cies and thus enable researchers to study the processing of competing, simultaneous events.

Advanced Analysis of Event-Related Brain Electric Signals

Traditional ERP analysis as described earlier are useful for quantifying the temporal unfolding of electrocortical processes that are
time locked to known events. Methodological advances in signal processing over the past decades have resulted in additional
research tools for extracting information from the EEG signal. For instance, time–frequency representations of EEG signals (some-
times achieved through spectrograms or wavelet transforms) have allowed a description of temporal dynamics that overlap in the
time domain but are located in different frequency ranges of the EEG signal. They have enabled researchers to characterize the effects
of prestimulus and ongoing activity on evoked activity, which is difficult with ERP approaches alone. Importantly, they also allow
more advanced follow-up analysis using algorithms that are sensitive to the patterns of brain connectivity underlying the electro-
physiological data recorded.

Conclusions and Future Directions

Recent years have seen exciting developments in virtually every aspect of ERP research, ranging from innovation in data-recording
techniques to novel experimental paradigms and sophisticated analysis techniques. It is now possible to record from high-density
sensor arrays at hundreds of extracranial locations, providing impressive spatial sampling along with high temporal accuracy.
EEG technology is more affordable than hemodynamic imaging techniques, widely available, and relatively comfortable for partic-
ipants. The technical and conceptual advances have taken the field to a point where the neurophysiological processes underlying the
modulation of ERPs are increasingly understood. Thus, ERP-based research may move beyond a descriptive analysis of waveforms,
generating and testing hypotheses on the in-vivo, real-time, functioning of the human brain.

Appendix A Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/B978-0-12-809324-5.02456-1.

References

Berger, H., 1969. On the electroencephalogram of man. Electroencephalogr. Cin. Neurphysiol. (Suppl. 28), 37–73.
Chapman, R.M., Bragdon, H.R., 1964. Evoked responses to numerical and non-numerical visual stimuli while problem solving. Nature 203, 1155–1157.
Davis, P.A., 1939. Effects of acoustic stimuli on the waking human brain. J. Neurophysiol. 2 (6), 494–499.
Donchin, E., Callaway, E., Cooper, R., Desmedt, J.E., Goff, W.R., Hillyard, S.A., Sutton, S., 1977. Publications criteria for studies of evoked potentials (EP) in man. Report of
a committee. In: Attention, Voluntary Contraction and Event Related Cerebral Potentials. Progress in Clinical Neurophysiology, vol. 1.
Foxe, J.J., Simpson, G.V., 2002. Flow of activation from V1 to frontal cortex in humans. Exp. Brain Res. 142 (1), 139–150.
Handy, T.C., 2005. Event-Related Potentials: A Handbook. MIT Press, Cambridge, MA.
Hillyard, S.A., Anllo-Vento, L., 1998. Event-related brain potentials in the study of visual selective attention. Proc. Natl. Acad. Sci. U.S.A. 95 (3), 781–787.
Jasper, H.H., 1958. The ten twenty electrode system of the international federation. Electroencephalogr. Clin. Neurophysiol. 10, 371–375.
Event-Related Potentials 7

Jeffreys, D.A., Axford, J.G., 1972. Source locations of pattern-specific components of human visual evoked potentials. I. Component of striate cortical origin. Exp. Brain Res.
16 (1), 1–21.
Kappenman, E.S., Luck, S.J., 2012. ERP components: the ups and downs of brainwave recordings. In: Luck, S.J., Kappenman, E.S. (Eds.), Oxford Handbook of ERP Components.
Oxford University Press, New York.
Luck, S.J., 2014. An Introduction to the Event-Related Potential Technique, second ed. MIT Press, Cambridge, MA.
Nunez, P.L., 1989. Generation of human EEG by a combination of long and short range neocortical interactions. Brain Topogr. 1 (3), 199–215.
Nunez, P.L., Srinivasan, R., 2006. Electric Fields of the Brain: The Neurophysics of EEG. Oxford University Press, USA.
Spehlmann, R., 1965. The averaged electrical responses to diffuse and to patterned light in the human. Electroencephalogr. Clin. Neurophysiol. 19 (6), 560–569.
Stolarova, M., Keil, A., Moratti, S., 2006. Modulation of the C1 visual event-related component by conditioned stimuli: evidence for sensory plasticity in early affective perception.
Cereb. Cortex 16 (6), 876–887.

Further Reading

Keil, A., Debener, S., Gratton, G., Junghofer, M., Kappenman, E.S., Luck, S.J., Luu, P., Miller, G.A., Yee, C.M., 2014. Committee report: publication guidelines and recom-
mendations for studies using electroencephalography and magnetoencephalography. Psychophysiology 51 (1), 1–21. http://dx.doi.org/10.1111/psyp.12147.
Regan, D., 1968. Evoked potentials and sensation. Percept. Psychophys. 4 (6), 347–350.

You might also like