Request PDF on ResearchGate | On Jan 1, , G. A. Stefanatos and others published Event related potentials: A methods handbook. Event-related potentials: A methods handbook Request Full-text Paper PDF In contrast, event-related potentials (ERPs) provide a real-time measure of the. The first comprehensive handbook to detail ERP methodology, covering experimental design, data analysis, and special applications. The study of event- related.
|Language:||English, Spanish, Portuguese|
|Genre:||Business & Career|
|Distribution:||Free* [*Registration needed]|
1 Interpreting Event-Related Brain Potentials. 3. Leun J. Otten and Michael D. Rugg. 2 Ten Simple Rules for Designing ERP Experiments Steven J. Luck. Event‐Related Potentials: A Methods Handbook. Edited by Todd C Handy. A Bradford Book. Cambridge (Massachusetts): MIT Press. $ xi + p + 9 pl; ill. Event-Related Potentials: A Methods Handbook. edited by Todd C. Handy, pp., Bradford Book, , $ Over the past ten to fifteen.
In addition, many disorders are characterized by a change in the timing of one or more neural processes, and this can be measured much more readily with ERPs than with fMRI or PET.
Recent developments in equipment have also made it easier to record EEG in well-controlled environments outside the laboratory, such as clinics, schools, and hospitals. Moreover, although there are differences in waveshape, size, and timing of ERPs between individuals, ERPs tend to be highly stable within an individual.
Indeed, high internal consistency and high test-retest reliability of ERPs have been demonstrated in typical research participants and individuals with psychiatric disorders 22 — This high reliability, coupled with the fact that ERPs can be recorded many times from the same individual, means that ERPs can be used to examine changes in brain activity resulting from treatment intervention or disease progression.
Furthermore, animal models exist for some ERP components, which can be particularly useful in the early stages of drug development 26 ; Collectively, these features make ERPs promising candidates for biomarkers of psychiatric disorders 24 ; Designing an ERP Experiment Although the temporal resolution of the ERP technique makes it possible to see the many processes that occur between a stimulus and a response, many processes operate simultaneously, and the voltages from these processes are summed together in the ERP waveform.
Thus, one major challenge in conducting ERP research is to isolate a single operation from the many other operations the brain is performing at the same time. Isolating a component from the ERP waveform is necessary to make conclusions about the presence, size, or timing of a specific mental operation as opposed to conclusions about brain activity, more generally.
Given that individuals with clinical disorders often exhibit deficits in more than one operation, isolating a single ERP component can be especially important for drawing clear conclusions from ERPs in clinical research. Importantly, the conclusions that can be drawn from an ERP study also depend on how well the ERP component has been linked to a specific mental operation in previous research, which may or may not be well determined discussed further below; see also 11; 29; One factor that plays a significant role in how well an ERP component can be isolated is the design of the experiment.
Although it is certainly possible to take any experiment, put electrodes on participants, record EEG, and extract ERPs, this approach is very likely to yield ERP waveforms that collapse multiple operations, making it difficult or impossible to tell which operation or operations varied among conditions or groups. One effective design strategy is to focus the experiment on a single ERP component, holding all factors unrelated to that component consistent across the experiment.
For example, we were interested in whether people with schizophrenia exhibit delays in stimulus evaluation time For a more detailed discussion of the factors that affect the polarity of an ERP, see Luck , chap. In other words, the voltages measured on the scalp at a particular time reflect synaptic activity at that particular instant, with no measurable delay.
Thus, ERPs provide a direct and instantaneous millisecond-resolution measure of activity related to neurotransmission. Summation of Components in the Observed ERP Waveform It is important to note that although the ERP waveform at a particular instant reflects synaptic activity at that moment, it does not reflect only the neural activity that began at that particular instant. In other words, multiple groups of neurons are active simultaneously in different regions in the brain.
If we think of this neural activity in terms of dipoles, this means that multiple equivalent current dipoles are active simultaneously. In fact, source localization studies have shown that as many as 10 separate equivalent current dipoles may be active at a given time Di Russo et al. If we return to our conception of ERP components, in which we define an ERP component as a signature of an individual neural process, each equivalent current dipole is essentially a separate ERP component.
In other words, when we say that multiple equivalent current dipoles are active simultaneously, this really means that multiple ERP components are generated simultaneously.
In some cases, neurons engaged in one mental process may be distributed in different areas of the brain, such as the simultaneous processing of a single auditory signal in both the left and right temporal lobes. This would essentially lead to two equivalent current dipoles. They are typically treated as parts of a single component under the assumption that both hemispheres are engaging in essentially the same mental process.
However, this is a fine detail of the definition of an ERP component, with little practical significance for the use of ERP components. Furthermore, resolution of this issue would require a precise definition of what is meant by mental process in terms of the behavior of neurons, both individually and as a group.
That is, how do we determine whether the same mental process is occurring in two individual neurons, and on a larger scale, in groups of neurons?
This is a complex issue that remains to be resolved by future research. The combination of multiple ERP components on the scalp leads to the superposition problem, which is depicted in Figure 1. When multiple ERP components are simultaneously active, the recorded voltage at the scalp is based on the sum of the voltages from all the individual components. This is a simple additive process. That is, if you knew the true waveform for each individual component, you could add all the component waveforms together to get the ERP waveform at each electrode site scaling each component by a weighting factor that reflects the contribution of the component to the voltage measured at a specific electrode site.
Unfortunately, the true waveform for each component is not known in real recordings, and it is quite difficult to reduce the sum of the components in the observed data to the individual components. However, understanding with simulated data how the voltage recorded at a particular electrode site reflects the various internal generator sources can help us understand the properties and intricacies of the ERP signals.
The propagation of voltage from a single generator site to a particular electrode site depends on the position and orientation of the ERP generator source with respect to the electrode, along with the conductance of the brain, skull, and scalp. This can be quantified with a weighting factor: The contribution of a given generator to the voltage recorded from a given electrode site is simply the waveform at the generator multiplied by the weighting factor see Figure 1.
There will be a separate weighting factor specifying the relationship between each electrode site and each internal neural generator source. Together, the set of weighting values between each source and each electrode site provides a mixing matrix that defines how the different components mix together at each site. Some mathematical techniques for recovering the underlying components work by computing an unmixing matrix that reverses this process, passing the observed data through the unmixing matrix to compute the component waveforms see Chapter 3, this volume.
The contr but on of each component waveform to the observed waveform at a g ven e ectrode s te s determ ned by a we ght ng factor that ref ects the ocat on and or entat on of the generator re at ve to that e ectrode, a ong w th the conduct v ty of the t ssues that form the head.
The observed waveform at a g ven e ectrode s te shown at the bottom r ght s equa to the sum of each of the component waveforms, mu t p ed by the we ght ng factor between each component and that e ectrode s te. The we ghts are nd cated by the ws on the arrows between the component waveforms and the observed waveforms e.
When multiple ERP components are simultaneously generated in different brain areas, the voltages from these components sum together. The voltage recorded at each site will therefore be the sum of each of the internally generated ERP components, with each scaled by the weight between that electrode site and each of the generator locations.
The value at p. Consequently, the ERP waveform at each electrode site contains information about all of the neural generators in the brain, not just the generator sources located close to the electrode although nearby sources will usually have a greater weight.
The inability to relate the ERP waveform at a particular electrode site to the neural tissue directly below the electrode site is made even more severe by the properties of the head. Specifically, as electrical activity travels from the brain to the surface of the scalp, the activity must pass through layers of skull and scalp. Although these constituents of the head are sufficiently conductive to allow the electrical activity generated in the brain to appear on the surface of the head, they are not perfect conductors, and the high resistance of the skull relative to the low resistance of the underlying brain and overlying scalp causes the voltage to spread laterally as it travels.
The signals are therefore blurred together by the head, which further distorts the relationship between the voltage at a particular electrode site and the cortex directly under that site. Of course, anyone who has seen the ERP waveforms from multiple electrode sites knows that differences exist in the shape and size of the ERP waveform across electrode sites.
In other words, although the waveform at each electrode site reflects neural signals from all over the brain, the summated signals are not identical at each site.
It is tempting to use the scalp distribution information to estimate the location of the neural generator source by, for example, determining at which electrode site the signal is largest. However, the superposition of multiple components and the blurring of the voltages p.
In fact, an infinite number of internal generator configurations could produce any observed distribution of ERP activity over the scalp see Luck, , chap. Thus, there is no technique that can determine, with certainty, the locations of the sources and the waveform at each source without bringing in difficult-to-verify assumptions or other sources of evidence. To summarize, the ERP waveform reflects ongoing synaptic activity related to mental processing as it unfolds millisecond by millisecond.
However, because scalp-recorded signals require the simultaneous activation of large Page 5 of 30 ERP Components: The Ups and Downs of Brainwave Recordings groups of spatially similar oriented neurons, only a portion of the neural activity that occurs in response to a stimulus will be measurable from electrodes on the surface of the scalp.
Furthermore, the ERP waveform at a given electrode site reflects the contribution of many simultaneously active ERP components that overlap in time, and it is difficult to mathematically unmix the observed waveforms and determine the original component waveforms. We will concentrate on the spatial variants of PCA and ICA, in which components are defined on the basis of scalp distribution information see Spencer et al.
In these three approaches, a component is defined solely by its scalp distribution, which is assumed to remain stable over the course of a single experimental session this is a reasonable assumption given that brain geometry is unlikely to undergo major changes within a few hours.
As mentioned in the previous section, these techniques provide an unmixing matrix that reflects the estimated scalp distributions of the individual components; the waveform for each component is computed by passing the observed waveforms through this matrix. That is, rather than passing the component waveforms through the weights shown in Figure 1.
Unfortunately, there is no unique solution to the problem of determining the underlying component waveforms from the observed scalp waveforms, and these three techniques use different assumptions to pick a single solution to this problem without any guarantee that the correct solution will be found.
In source localization techniques, a component is equivalent to a neural generator source.
These techniques use biophysical assumptions about the flow of current through the conductive tissues of the head to define the scalp distribution of each component and thereby compute a unique unmixing matrix. To obtain a unique solution, these techniques must also rely on additional assumptions, such as a specific number of discrete dipoles or maximal smoothness in the distribution of current flow over the cortical surface.
That is, these techniques find the set of single-component scalp distributions that can sum together to provide the best fit to the observed scalp distribution as it varies over time while also being consistent with a variety of assumptions for a review and critique, see Luck, , chap.
Thus, source localization techniques define a component as activity arising from a region of cortex, which is similar to our definition of an ERP component as reflecting a specific brain process on the assumption that most brain processes occur in discrete areas4.
However, our definition of the term ERP component goes further, because more than one brain process may occur in a given region of cortex. Moreover, source localization approaches differ considerably from the traditional approach to defining components in the procedures used to discover and define individual components.
Whereas source localization techniques use a variety of assumptions to select a set of scalp distributions that together provide a quantitative account of the data from a given experiment, traditional approaches to defining components are based on using experimental manipulations to test hypotheses about the link between a voltage deflection and an underlying neural or psychological process as discussed further in a later section.
Principal component analysis and ICA make no biophysical assumptions, but instead use the statistical properties of the data to derive the scalp p. That is, the observed scalp distribution changes from moment to moment and from condition to condition as the underlying components wax and wane, and the statistical relationships between the values observed at the different electrode sites are used to determine the scalp distributions of the individual components.
In PCA, for example, two electrode sites will tend to contribute strongly to the same component if they tend to covary in voltage. Principal component analysis is designed to find an unmixing matrix in which a small number of componentseach with its own scalp distributioncan sum together to explain most of the variations in the observed scalp distribution.
It reduces a large and complex set of observed scalp distributions for each time point, condition, etc. In contrast, ICA is designed to find an unmixing matrix that maximizes the independence of each Page 6 of 30 ERP Components: The Ups and Downs of Brainwave Recordings component so that every individual component represents the largest possible amount of information.
The scalp distributions of the components in ICA may be correlated with each other as would be expected for two independent but nearby neural sources , but the strength of activation of each component varies independently of the strength of the other components over time points and over conditions. Whereas PCA attempts to lump as much information as possible into a small number of components, ICA attempts to split apart the information into different components for a detailed comparison, see Chapter 3, this volume.
Because it is a lumping technique, spatial PCA by itself is unlikely to produce components that are related to individual neural and psychological processes. However, the essence of ICA corresponds well with a reasonable assumption about these processes. Specifically, for something to count as a unique process, it must be dissociable from other processes.
This is largely identical to saying that the process must sometimes vary independently of other processes, and this is exactly the type of independence that ICA uses to define components. Thus, although ICA uses a mathematical approach rather than a hypothesis-testing approach to derive the components, it shares much with the definition of the term ERP component that we have proposed in this chapter.
Music Neuroscience Philosophy Physical Sciences. Kappenman and Steven J. Luck Abstract Event-related potentials ERPs have been used for decades to study perception, cognition, emotion, neurological and psychiatric disorders, and lifespan development. Bibliographic Information Publisher: Dec ISBN: Sep DOI: Editors Emily S. Read More. Luck Beyond ERPs: Brunia, Geert J. Smulders and Jeff O. Gehring, Yanni Liu, Joseph M. Luck and Emily S.
Perez and Edward K. Swaab, Kerry Ledoux, C. Christine Camblin, and Megan A.