Week 2 - EEG Flashcards
1
Q
Define spatial resolution
A
- Spatial resolution refers to the accuracy with which one can measure where an event is occurring. How precisely can the origin of a neural signal be located?
- Can spatially separate sources be differentiated?
2
Q
Define temporal resolution
A
- Temporal resolution refers to the accuracy with which one can measure when an event if occurring. How precisely can changes in the signal be tracked?
- High sample rate, no lag: good temporal resolution
- fMRI has a lag of a second
3
Q
What is the temporal and spatial resolution for EEG and ERPs?
A
EEG, MEG and ERP have good temporal resolution but bad spatial resolution
4
Q
What is EEG?
A
- Electric potentials in the brain being recorded through electrodes placed on different points on the scalp.
- Raw EEG signal:
- Voltage = difference in electric potential between two sites.
5
Q
What is the origin of the EEG signal and explain what happens during this
A
- Neurons communicate with each other through quick pulses of electrical current called “action potentials”. They bring about the release of neurotransmitters, which are absorbed by adjacent neurons. Action potentials occur at a rate of over 200Hz and are highly localised, they don’t create a diploe which makes them impossible to pick up by electrodes placed on the scalp.
- The type of electrical acitivity, that makes up the EEG signal, is the post-synaptic potential. Once a neuron receives an action potential from a neighbouring neuron and chemical transmitters have been released, this generates a current of ions in the cell. The ion current then causes a build up in electrical potential - this is the postsynaptic potential. If the postsynaptic potential reaches a certain level, it will cause an action potential to be released from the neuron. Postsynaptic potentials create a tiny dipole due to change in ion distribution allowing an EEG signal to be measured.
6
Q
What are the requirements for a measurable EEG signal?
A
- A large number of simultaneously active neurons are needed to generate a measurable EEG signal
- Neurons must be highly synchronous – if they fire with a delay, there is not enough charge to affect the electricity on the scalp side.
- Currents must have same direction (mostly inhibitory or mostly excitatory neurons) – otherwise they might cancel each other out.
- Neurons must have same orientation (of the cell itself):
- Subcortical areas (e.g. basal ganglia) - neurons are not aligned
- Cortical layers – neurons are aligned
7
Q
What is the inverse problem?
A
- Inverse problem – we can’t trace where the original EEG signal comes from.
- EEG systems typically have 32-256 electrodes:
- Each electrode = one observation
- Each dipole = one variable to solve
- Increasing the number of electrodes does not solve the inverse problem because we would still have millions of dipoles.
- Because of the inverse problem, the spatial resolution of EEG is poor
8
Q
What is the history of the EEG signal?
A
- 1875 - Richard Caton used galvanometer to observe electrical impulses from surfaces of living rabbit and monkey brains
- 1890 - Adolf Beck found that frequency of electrical activity of dog and rabbit brains is modulated by light intensity (input changes frequency of electrical activity)
- 1924 - Hans Berger measures first human EEG (from scalp sites), finds different frequencies for open and closed eyes
9
Q
How do you measure an EEG signal?
A
- Participant with EEG cap – with single active electrodes:
- Active electrodes show impedance (resistance) via LEDs
- Green < 5k
- Yellow <20
- Red > 20
- Active electrodes improves the signal-to-noise ratio by amplifying the signal at the scalp.
- Shielded cabin – to avoid other signals interfering:
- Reduction of electrical noise is crucial
- Faraday Cage – positive and negative ions. Negative ions move to one side which causes a second magnetic field to build up in the opposite direction of the field on the outside. Electrical charges in cage’s material redistributed → cancel out the magnetic field’s effects in the cage’s interior.
- Eye and muscle movements and sweat need to be reduced (drifting electrodes)
- Standardisation of scalp locations
- High viscosity contact gel (necessary for low impedances) – air and hair bad for electrical conductors
10
Q
How do you decompose a raw EEG signal?
A
- With a Fourier Transform, the raw EEG signal can be decomposed into frequencies. Similarly, chords can be decomposed into notes.
11
Q
What are the frequency bands in the EEG signal?
A
- Neurons tend to fire in temporal synchrony with each other; but at different frequency rates
- Different frequencies are related to different states in the brain
12
Q
What are the two ways to analyse EEG data?
A
- Spontaneous EEG:
- Measurement of voltage differences throughout longer periods
- Useful for:
- Activiational state (e.g., sleep research)
- Clinical research (e.g., epilepsy)
- Event-related potential (ERP):
- Voltage difference in time window relative to specific ”event“
- Measure for cognitive, motoric or sensoric process
13
Q
How is the spontaenous EEG signal used for the sleep-wake cycle?
A
- Different oscillation frequencies characterise the different phases of sleep-wake cycle
- Awake - beta and alpha bands.
- No-rapid-eye-movement (NREM) sleep:
- Stage 1: light sleep with low-amplitude waveforms. Theta bands.
- Stage 2: sleep spindles (the higher-frequency waves) and K complexes
- Stage 3 and 4: slow-wave sleep (SWS) with high amplitude waves and a deeper level of unconsciousness. Delta bands.
- Rapid eye movement (REM) sleep associated with activity and, in some individuals, with low-amplitude sawtooth waves and a higher level of consciousness. Beta bands.
- Finding out about sleep wake cycle can help with learning
14
Q
What are ERPs?
A
- Event-related potentials (ERPs) are very small voltages generated in the brain structures in response to specific events or stimuli
- The positive and negative peaks are labelled with “P” or “N” and their corresponding number. Thus, P1, P2, and P3 refer to the first, second, and third positive peaks, respectively. Alternatively, they can be labelled with “P” or “N” and the approximate timing of the peak. Thus, P300 and N400 refer to a positive peak at 300 ms and a negative peak at 400 ms ( not the 300th positive and 400th negative peak!). ERPs are measured by EEG signals.
15
Q
How do we get from spontaneous EEG activity to event related potentials?
A
- Steps involved in pre-processing:
- Segmentation (e.g. -200 – 500ms)
- Sort segments according to condition
- Baselining – necessary because of spontaneous voltage fluctuations in raw EEG:
- Principle - the average baseline activity (e.g. 200 ms epoch before stimulus onset) is subtracted from entire segment activity
- Result - the waveform is shifted towards zero - allows comparison between trials!
- Artefact rejection and correction – necessary because of noise:
- Artefact correction – correcting the signal based on what it would look like without environmental interference.
- Saccades, blinks, drifting voltage, high noise level all need to be removed.
- Averaging – for each condition and electrode separately
- Exporting and grand average:
- Export - calculate mean amplitude for all conditions and all electrodes in time window of interest –> run statistics
- Grand average - average of averages: across trials and subjects → Display purposes (this is what is typically shown in figures in EEG papers)
16
Q
Why do we need hundreds of trials?
A
- More trials allow us to identify the real evoked potentials more easily
- Random noise averages out to zero
17
Q
What are Topo maps?
A
- Amplitude is depicted in terms of colour
- Spatial representation (scalp distribution) of EEG signal – even though spatial resolution is not good for EEG, its important to look at topo maps to see what cognitive processes are affected.
- Typically for time windows, not single data points
- Allows comparison between conditions or time course
21
Q
Describe the P300 ERP and how is this related to early vs. late selective attention?
A
- The P300 (P3) wave is an event related potential (ERP) component elicited in the process of decision making
Early versus late selective attention?
- Early section theories (Broadbent, 1958) assume that after the sensory store, a selective filter picks out small information that goes onto higher level processing
- Late selection theories (Treisman, 1964) assume that a lot of information goes to higher level processing and then we filter out stuff.
- Early and late selection models were based on behavioural data but inconclusive.
Hillyard et al. 1973 – used EEG signals for early vs. late selective attention:
- Task - count number of single tones. Asked to attend left or right ear.
- Found the N1 component was always larger for the attended ear than the unattended ear
- Shows early on that we filter out right stimuli if we listen to the left ear or filter out left stimuli if we listen to the right ear.
- Used the P3 to see when there is a difference between the target and non-target
- P3 only observed for task relevant tones (high tones) (signals), not for standard tones (from 250 ms)
- Differences between attended and non-attended tones (regardless of target versus non-target) suggest filtering at ~80 ms post stimulus.
- Differences between target and non-target tones from 250 ms on. This requires higher level processing.
- Higher level processing happens AFTER filtering
- Evidence for early filter models
22
Q
Describe the N400 ERP and the study associated with it
A
- The N400 is part of the normal brain response to words and other meaningful (or potentially meaningful) stimuli, including visual and auditory words, sign language signs, pictures, faces, environmental sounds, and smells
- Detection of semantic incongruity/expectancy
Kutas & Hillyard, 1980:
- Participants saw sentences on the screen
- Divided into three categories:
- No incongruity – sentence is normal
- Semantic incongruity – last word doesn’t fit the sentence
- Physical deviation – last word is bigger
- EEG signal shows:
- Difference between semantic incongruity and no congruity
- Interpreted as a violation of semantic expectancy:
- Congruent to context (classical N400 effect)
- N400 is affected by:
- No relation (classical N400)
- Untypical member (bird – turkey)
- Typical member (bird – sparrow)
- Antonym (opposite)
- N400 tracks effort to retrieve stored conceptual knowledge
- Preceding words serve as cues and make retrieval easier
- Also, N400 independent of sensory modality
- Doesn’t have to be written, can be spoken or visual
