Method Flashcards
The development of human neuroimaging techniques
1930: Electroencephalography (EEG) was discovered
1970: First radiographic methods available
1980-1990: Positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) became available
1990-2000: U.S. president George H. W. Bush declared the “Decade of the Brain”
Ap recap
Post-synaptic potential is determined by integrating input of many synapses at the dendrites.
Action potential travels along the axon to all terminals.
-resting at -70mv, stimulus over threshold lead to depolarisation. follow by repolarisation also hyper polarisation
- We rarely get the chance to measure activity of neurons directly in humans
- We rely on techniques that measure activity “indirectly” and from ”the outside”
How can we measure neuron activity
- We (mostly) rely on methods that are non-invasive, meaning that we do not open the skull (or interfere with brain function)
- One class of neuroimaging methods available for human research detects frequencies in neural signals (i.e. the rate of change of the signal over time)
- 1 Hertz (Hz) = completing a full cycle (up and down) in one second
- Biological signals never contain just one frequency (as in artificial signals)
- Complex signals can be decomposed into frequency components, each has a particular frequency (e.g., 1 Hz, 2 Hz, 3 Hz, …)
- The amplitude describes how much it goes up and down
- The phase describes when it goes up and down
Methods that measure fast-changing electrical activity from outside the scalp
oMagnetoencephalography (MEG): Measures electrical activity through the magnetic fields produced by the electrical activity of neurons
oElectroencephalography (EEG): Measures small voltage fluctuations
picked up by sensitive scalp electrodes
hemodynamics
- The changes at the cell membrane leading up to the generation of action potentials and the neurotransmitter release at the synapse all require energy
- For this, oxygenated blood is transported to “active” brain regions, because oxygen is used there to produce the necessary energy
o functional Near-Infrared Spectroscopy (fNIRS): Optical imaging technique that uses light to study blood oxygenation through the skull
o Positron Emission Tomography (PET): Can measure the distribution of specific molecules in the blood (by radioactive labelling)
o functional Magnetic Resonance Imaging (fMRI): Measures local changes in blood oxygenation with high spatial resolution
How does neuron packing arragement affect signal mesurement
- Luckily for us, neurons with similar “interests” tend to “cluster” together
- This clustering happens at the level of small “columns” but also in larger “areas”, which serve (roughly) the same functions
The trade off
- We have to trade-off methods that have a high spatial resolution (e.g., fMRI) vs. methods that have a high temporal resolution (e.g., EEG)
- Finally, there are methods that are “causal”, because they rely on direct stimulation of the brain, meaning that we can study what effects this might have
- The most commonly used method is Transcranial Magnetic Stimulation (TMS)
Transcranial Magnetic Stimulation
transcranial Magnetic Stimulation is a “non-invasive” technique used to create “virtual cortical lesions”
- Studies on patients with real lesions have informed cognitive science for a long time as they allow studying what patients can’t do anymore
- E.g., Phineas Gage (1823-1860), an American railroad construction worker, who suffered a serious injury by an iron rod piercing his head and frontal cortex
- This led to severe changes in his personality
- Lesions can therefore tell us a lot about the functions of specific brain region
Why don’t we just rely on patients with natural lesions?
- Removing most parts of his hippocampus, parahippocampal gyrus and amygdala in famous patient H.M. led to severe anterograde amnesia
- In the same way, lesions in Broca and Wernicke areas have been linked to impairments of speech production and language comprehension, respectively
- However, there might not be enough patients with circumscribed lesions to study all cognitive functions
- Lesions in single, specialised areas are rare
- Recovery and brain plasticity might compensate for lesions over time è patients might become quite ‘special’ over time
How TMS work
TMS is a method to create a small, short-lived virtual lesion that is reversible
- TMS can be applied externally, using a coil placed on the scalp that produces a rapidly changing magnetic field to induce electrical currents in the brain
- These currents can depolarise neurons in a small, circumscribed area of cortex
- TMS-induced current causes neurons to fire randomly, acting as “neural noise”, thereby masking the neurons that are firing correctly
- In order to create the current pulse, which is required to generate the magnetic field, a capacitor is charged and then suddenly discharged
- In order to create a magnetic field strong enough for stimulation, very fast loading times (~100-200 μs) and short discharge durations (<1 ms) are required.
History of TMS
- Fritsch & Hitzig (1870) were the first to electrically stimulate the cortex of animals
- D’Arsonval (1896) discovered that the magnetic stimulation of the visual cortex can elicit “phosphenes”
- Magnusson & Stevens (1911) developed the first “head coil” covering the entire head
Barker, Jalinous & Freestone (1985) developed the current TMS technique, which had the great advantage of not being painful (mostly…)
rTMS
fast sequence of pulses instead of a single pulse (called: “repetitive TMS”, or rTMS in short)
coils
Different coils have been used, but the most common one is the ‘figure-eight’ coil
- The figure-eight coil generates magnetic fields generating offset current loops that circulate in opposite directions, allowing for high precision in the stimulation
- A more focal area of the cortex is stimulated using the figure-eight coil compared to the round coil (usually 3-4 mm radius, but up to 1 mm possible)
- The advantage is that the researcher knows which part of the cortex was affected
Applications of TMS in Biological Psychology Research
1) The injection of “neural noise” approach using single-pulse TMS
2) The “virtual lesion” approach using repetitive TMS
3) The “probing excitability” approach using single-pulse TMS
4) The “probing information transfer” approach using paired-pulse TMS
5) Using paired-pulse TMS to test for the decay of activity
6) Clinical applications of TMS
injection of “neural noise” approach
Using single-pulse TMS to disrupt cognitive processing
- If a single TMS pulse to a specific region of the cortex disrupts a cognitive function, this is a powerful demonstration of its causal involvement in this process
- Testing for causality is impossible using most other neuroimaging techniques, which usually rely on correlations
- One way of doing this is to interfere with the process of interest at exactly the time window during which the regions is required; e.g., to delay movements or to disrupt visual processing
- Regions do not stop working completely, but “neural noise” interferes with normal functioning
classical study of injection of “neural noise” approach
the researchers used 3 alphabetical letters as stimuli presented under difficult viewing conditions using illuminated frames/background Amassian et al., 1989
- Magnetic stimulation (i.e. TMS) was applied ~ 2 cm above the inion over visual cortex
- Effects on letter perception were investigated when varying the interval between visual stimuli and time point of TMS stimulation
It was found that during a critical period (40 – 120 ms) stimulation affected
detection performance
When shifting the stimulation site from left to right, perception of letters in
the contra-lateral visual field was predominantly impaired
- When moving the TMS stimulation from top to bottom at midline, and letters were displayed vertically, stimulation above the reference line suppressed
letters at the bottom of the display
- Stimulating below the centre was not possible (the bone was in the way)
The “virtual lesion” approach
repetitive TMS (rTMS) is used to interrupt or enhance cognitive processing
- It is also possible to inhibit cognitive functions for a longer period of time by applying repetitive TMS (rTMS)
- It can then be measured whether (and for how long) a specific cognitive task is impaired (usually slowing instead of total loss of function)
- There are strict safety guidelines for rTMS (Wassermann, 1998)
“probing excitability” approach using
single-pulse TMS
(usually over the motor system)
- For the motor system in particular, one option is to test how responsive (or “excitable”) the motor cortex is during a cognitive task
- The idea is that if the motor cortex is required for a cognitive task, then it should already be activated when single-pulse TMS is delivered
- The measure of interest is how strongly the motor cortex “reacts” to the pulse itself (any “disruption” is ignored), i.e. how strong its output is after being stimulated
- The excitability of the primary motor cortex can be measured by recording “motor evoked potentials” (MEPs) using the electromyogram (EMG) – the electrical activity of muscles
- One can then measure MEPs for each stimulation and compare average
MEPs between experimental conditions
“probing excitability” approach amd M1
- Is the primary cortex (M1) involved in the mental rotation of objects?
- Some neuroimaging studies found activation of M1 during mental rotation – which is odd as nothing is ‘really’ rotated!
- Maybe M1 involvement simply reflects in inner speech?
- Or do we use our brain in the same way for rotating real and imagined objects?
- Eisenegger and colleagues found that stimulation of M1 during mental rotation elicited stronger MEPs(Motor evoked potentials) as compared to baseline, reading aloud and reading silently
- Evidence that M1 is more excitable during mental rotation èmight be already activated, and hence, “involved” in the cognitive process
- Maybe the involvement of primary motor cortex (M1) can be explained by people imagining using their own hands for mental rotation? Kosslyn, Thompson, Wraga & Alpert, 2001; Kosslyn, Ganis, Thompson, 2001; Bode et al., 2007
- Then, M1 involvement might depend on how mental rotation is instructed?
- M1 could potentially depend on whether the stimuli automatically trigger the use of a particular strategy?
The “probing information transfer” approach
It is also possible to utilise two pulses – a technique used in the “probing information transfer” approach using paired-pulse TMS
- Uses two pulses, delivered in brief succession – the first pulse is usually sub-threshold while the second one is supra-threshold
- The question is how strongly the first pulse influences the effect of the second
Other paired-pulse TMS applications
Paired-pulse TMS is not exclusively used to test for transfer between two regions. It can also be used to test for the decay of induced activity within the same brain region.
- In Schizophrenia, abnormalities in inhibition in the motor cortex have been suggested. There is evidence that the cortical silence period (CSP) – a period of suppression of tonic motor activity that follows descending excitatory activity – is reduced
- The researchers produced the excitatory activity by a first (sub-threshold) TMS pulse to the left motor cortex, and measured the excitability by assessing the effect of a second (supra-threshold) pulse (via MEPs)
- The results show that, compared to controls, patients with and without medication showed stronger responses to the second pulse
- This result points to general deficits in motor inhibition, i.e. the induced “artificial” activity does not decay as quickly as it should
Measuring electrical activity via eeg: the basic
- Electrical currents flow from high to low voltage. We are interested in the current flow from the scalp to the ground, and the EEG measures these voltage signals
- EEG recorded at the scalp is non-invasive – however, it is also possible to record intracranial EEG by measuring activity directly at the exposed cortex
- Normal scalp EEG is cheap and (relatively) easy to conduct
The discovery of eeg
- Hans Berger (1873-1941) detected the first EEG signal in 1924 with electrodes attached to the scalp of a human (his wife) and reported the results in 1929
- Berger initially studied medicine because he was convinced that there is “psychic energy”, which might allow for telepathy
- He wanted to discover the objective activity in the brain and “psychic phenomena”, but he did not realise the basis and potential of his discovery at the time
- Initially, he used two electrodes – silver wires placed under the scalp – one attached to the front of the head and one to the rear, and recorded the potential (i.e. voltage) difference between them
- Later, he used sliver foil placed on the scalp
- Berger also first described the alpha rhythm – when people closed their eyes, the electrical signal varied with a characteristic frequency of 8-13 Hz
Measuring electrical activity via egg: nodes
- Electrode gel/paste is applied to the gap between scalp and electrode to decrease the impedance (i.e. get a better signal)
- Typical systems have 32, 64, 128, or even 256 head channels
- Some universal location of nodes: International 10-10, International 10-20, modified 10-10 for infant
- In addition, we need a ground and a reference channel, and we use other electrodes to measure eye movement and blinks
- The reference should be a neutral point (e.g., tip of the nose, mastoids), but some researchers reference to the average of all scalp electrodes
The eeg signal itself
- EEG signals are very small and have a typical amplitude of 10 μV to 100 μV
- These signals need to be amplified, typically by a factor of 1,000 to 100,000
- The typical sample frequency is between 256-1024 Hz (1000 Hz = 1 data point per millisecond)
- The signal is band-pass filtered to remove the low (<0.5-1 Hz) and high frequencies (typically >35-70 Hz) because they cannot reflect brain activity.
- The signal is also notch-filtered (at 50 Hz or 60 Hz) to remove line noise, which is also not brain activity
artefacts
- Artefacts are signal from the muscle that is transmitted to the brain
- Different kinds of artefacts can contaminate the signal and need to be removed
- Some can be detected automatically, and some need to be identified manually
Eye movement in eeg
- Eye movements and eye blinks create very strong artefacts, much stronger than the brain signals we are interested in, because the eye is a strong dipole
- Since we record directly from electrodes next to and under the eye (to capture horizontal and vertical eye movements, respectively), we can identify them easily
- We can either exclude contaminated trials, or use mathematical algorithms, such as independent component analysis (ICA), to remove just the eye component
Neurophysiology of the EEG signal
- The EEG activity originates mostly from post-synaptic potentials – voltages that arise when neurotransmitters bind to receptors on the membrane of the post-
synaptic cell (and only to some degree from action potentials) - This causes ion channels to open or close, leading to graded changes in the potential across the membrane
- With these electrical changes, the neuron acts as a small “dipole” (with a positive and a negative pole)
- Signals from single neurons are not strong enough to be recorded outside of the head, but if many neurons spatially align, then their summed potentials add up and create the signals we can record
A recording unit of neuron in eeg
This pooled activity from groups of similarly oriented neurons mostly comes from large cortical pyramid cells
- The functional unit is >10,000 simultaneously activated neurons
Neuron orientation and eeg
- The orientation of the neurons determines the sign of the recorded potentials
- Some orientations result in signals which cannot be recorded
EEG biased
EEG is biased to signals generated in superficial layers of the cerebral cortex, i.e.
the gyri (ridges) directly bordering the skull
- Signals in the sulci are harder to detect and additionally masked by the signals from the gyri
- The meninges, cerebrospinal fluid (CSF) and skull “smear” the EEG signal
- We cannot (easily) locate the sources of the signal, because it is a mathematical inverse problem:If the sources are known, the resulting scalp configuration can be reconstructed
- The reverse is not true – one given scalp configuration of signals can have multiple dipole solutions (i.e. we do not know where the signal comes from)
Analysing the EEG signal
- When looking at frequency information (for example in sleep research), the raw signal can show systematic variations, i.e. a specific frequency is dominant
- The EEG signal we measure is a mixture of many frequencies (plus noise) originating from different locations in the brain
The signal can be looked at in the time domain (amplitude data) or in the frequency domain
Reading eeg
- We can ask which frequencies in the signal are dominant, and when they are dominant (plotted here as the spectrogram)
- We can also measure the amplitude of the signal at specific moments in time when people engage in cognitive tasks – this is the basis of the event-related potential (ERP) method which we will cover in the next lecture
Analysis of eeg event related potential overview
- The analysis of event-related potentials (ERPs) is a method that allows us to investigate fast neural processes related to specific events of interest
- Usually, we want to study what happens in the brain when participants engage in cognitive processes, such as perceiving, deciding, responding, etc.
- ERPs can be obtained by time-locking the signal to the events we want to study, so we can analyse the signal amplitude at specific channels
The key assumptions for eeg analysis
- The event of interest is defined in time
- The event consistently evokes the signal
- The timing of the signal is consistent
- The signal and the noise are uncorrelated
- The noise is random with a mean of zero
Averaging amplitude data of eeg for many test
We want to know whether there is brain activity reliably related to the cognitive processes of interest
- However, usually the single-trial EEG trace is far too noisy to do that
- For example, in this experiment, researchers studied the EEG data observed for frequent events (“X”) and rare events (“O”)
- If our assumptions are met (e.g., the noise is uncorrelated to the signal and has a mean of zero), then we can align the trial segments from the event (e.g., when and X or an O was shown) and average over the respective trials
- All noise will average out, and we are left with a better ,estimate of the true neural response to the event of interest
Averaging amplitude data
- In order to obtain a useful estimate of the neural response to an event, many trials of the same kind must be averaged
- Even the averaged signals per session (i.e. “experimental block”) for the same participant still look (somewhat) different
- There is also a lot of variance between different participants who do the same experiment
- However, for most participants, we can see a stable difference between experimental conditions in the same time window
Classifying messeges from eeg
- ERPs are described by their polarity and their order (at least most of them)
- Specific ERP components are measured at specific (groups of) channels
- There are many well-studied ERP components, many of which do not have the typical labels (e.g., some are named by hypothesised “function”)
One major problem with ERPs is reverse inference:
Concluding what a component “reflects” in a specific experiment requires knowing what the component “usually” reflects, which again requires experiments… - However, we now have over 60 years of ERP research to draw on
options to derive a measure of the amplitude in eeg
oPeak amplitude (i.e. baseline-to-peak)
oPeak-to-peak
oArea under the curve
oLatency (i.e. the “onset” of the amplitude)
- Most studies (~70%) are interested in a baseline-to-peak measure; but then
researchers have to decide what the best way to estimate this is. One could extract:
oThe maximum peak (i.e. the most extreme point)
oThe mean amplitude (i.e. by defining a range around where the peak should be
and considering the mean)
Which option to use for eeg amplitude measurement
- There is no clear “rule” for which measure is the best, but one should be aware that results might be very different depending on which measure is used
- These measures might reveal different aspects of the cognitive process of interest
eeg latency
- The latency refers to the onset of the ERP component
- There are mathematical algorithms that help estimating the latencies
- It can be very difficult to see when exactly a component starts
- The tail end of the component is often neglected
Gehring and colleagues (1993) investigation on error
- A very good way to use ERPs for studying cognitive processes is to subtract the waves from one condition from the waves from a control condition
- Gehring and colleagues (1993) investigated whether there is a cognitive mechanism for the detection of and compensation for errors
- For this, they measured the error-related negativity ERN), a negative deflection of up to 10 μV in amplitude observed at central electrodes ~80-100 ms after an erroneous response
- The ERN comes so fast after we have committed an error (but can’t take it back anymore) that is has been termed the brain’s “oh shit” response
Studying Cognition using ERPs process example: Gehring and ERN methodologies
- Gehring and colleagues asked their participants to emphasise accuracy or speed in a simple Flanker-task in which participants had to respond to the central letter on the screen Gehring et al., 1993
- They reasoned that incongruent displays should lead to more errors, but error detection should only matter in the accuracy condition
Overall, they found a clear ERN on incorrect trial in comparison to correct trials
Result of Gehring
- The ERN was indeed strongest when people emphasised accuracy, and weakest for speed
- This confirmed their hypothesis that participants’ brains only really “cared” about error detection when this was also important
- However, this result does not show whether the ERN is also related to compensating for errors
- If the ERN was not only indicating error detection, but also compensation for making an error (i.e. avoiding the “oh shit” moment), one would expect that the ERN additionally reflected the attempt to break the error
- To investigate this question, Gehring and colleagues divided the ERNs from the entire experiment into quartiles from “small” to “extra large” (X large)
- They then investigated whether ERNs of different sizes (i.e. quartiles) were related to specific response parameters, which, in turn, might be related to correcting or avoiding errors
ERN relationship to other variables
The greater the ERN, the lower the response force èParticipants might be trying to correct for the error
The greater the ERN, the higher the probability to get it right on next trial èParticipants might be successful learning from errors
The greater the ERN, the slower the response on next trial èParticipants mightbe slowing down to avoid a subsequent error
