EEG
EEG
electroencephalography
EEG qualities
high temporal resolution, low spatial resolution (because it is affected by the layers of the head)
How do EEGs work? in short
EEG and averaged ERPs measure electrical potentials generated in the extracellular fluid as ions flow across the cell
membranes and neurons talk to one another via neurotransmitters.
In other words, it measures postsynaptic potentials
The forward and inverse problem
Forward: computing the EEG signals on the scalp given a known source configuration; predicting the scalp potentials (EEG data) from known brain activity.
easy
Inverse: estimating the brain sources (location and strength) from the measured EEG signals on the scalp; inferring where in the brain the EEG signals originate.
hard
What is the contingent negative variation?
The period of brain activity related to the preparation for performing an action.
important ERP components
- polarity
- timing
- scalp distribution
- sensitivity to task manipulation
- nomenclatures (pre, peri, post)
C1
p/n depending on where in the visual field the stimulus is
50-70
sensory processing
P1
90-100
sensory/ perceptual preprocessing
N1
170-200
called N170 for faces
perceptual processing, expert recognition, visual discrimination
N2
225-250
object recognition and categorization
N2PC
deployment of covert attention
P3
300
stimulus evaluation, encoding and maintainance
which two components evaluate performance
ERN; event-related negativity (negative)
Pe; error positivity (positive, but after negative response)
LRP
preperation of motor response
Improving methodological rigor in EEG
- Standardize electrode placement, especially reference placement
- Make sure that research has not been conducted before
- multiple comparison correction
- Make sure there is no baseline bias + baseline correction
- sufficient number of trials (or participants)
- remove artifacts (eye movement, muscle movement, noise)
- Keep in mind that different stimuli inherently create different ERPs
- Look at the entire ERP, not just the peak
what is the sampling rate?
The number of times per second the analog signal is converted to digital data. A higher sampling rate captures more data points. (normaly 512 Herz)
what are the three types of electrodes?
- active
- reference
- ground
steps for preprocessing EEG signal
- importing
where raw digital data, which represents the voltage measured over time from each electrode relative to a reference, is loaded into the analysis software or environment to begin the preprocessing sequence
- filtering
Remove unwanted frequency bands.
High-pass filters remove low drifts (attributed to sweat or amplifier issues)
Low-pass filters remove high frequencies (often associated with muscle noise)
- segmentation (epoching)
Dividing the continuous EEG recording into segments time-locked to specific events (like stimulus onset or response)
- baseline removal
Subtracting the average voltage during a baseline from the entire waveform, to remove anticipation activity.
- artifact handling
done with ICA
Examples of things that can cause noise
- bad connection (electrode pop, skin impedance, sweat)
- external electrical noise
- amplifier saturation/ drift
- ocular artifacts
- muscle artifacts
three types of EEG analysis
- time-domain (ERP)
- time-frequency
- frequency
Three rarely discussed issues of ERP
- ERP refractoriness, the time it takes for the ERP to "reset"
- Offset transient, where sudden stimulus offsets elicit their own ERP components (long or short intervals)
- individual differences, e.g., due to unique patterns of brain folding
types of signals
- gamma
- beta
- alpha
- theta
- delta
Mismatch negativety
a negative ERP response when something unexpected occurs
Frequencies of bands
Gamma - 60-30
Beta - 12-30
Alpha - 8 -12
Theta - 8-4
Delta - 4-1.5
Most used amplifier
BioSemi
How does the midline run?
From inion to nasion