Bio psych Flashcards
What is Transcranial Magnetic Stimulation (TMS)
“Non-invasive” technique to create virtual cortical “lesions”
How is TMS applied?
externally - coil placed on scalp producing rapidly changing magnetic field inducing electrical currents in brain
What does TMS do?
○ Depolarises neurons in small circumscribed area of cortex
○ TMS-induced current causes neurons to fire randomly - acting as neural noise - masking neurons that are firing correctly
○ Creates so much noise that nothing gets through anymore
What is required for TMS
100-200mews and short discharge durations
Different approaches to TMS
- Injection of neural noise
- Virtual lesion approach
- Probing excitability approach
- Probing information transfer approach
What is the neural noise approach to TMS?
- single-pulse TMS to disrupt cognitive processing
- If a single TMS pulse to a specific region of the cortex disrupts cognitive function - powerful demonstration of its causal involvement in process
- to infer causality: interfere with process of interest at exactly time window during which region is required eg delay movements or disrupt visual processing
- Regions do not stop working
What is the alphabetical neural noise study?
- Researchers used 3 alphabetical letters as stimuli presented under difficult viewing conditions using illuminated frames/backgrounds
- Three letters next to each other
- Stimulated 2cm above inion over visual cortex
- Critical period - 40-120ms stimulation affected detection performance - how long it takes visual cortex to process information and when is it over
- When moving TMS stimulation from top to bottom at midline and letters were displayed verticallly, stimulation about reference line supressed letters at the bottom of the display
- Stimulating below centre was not possible (inion bone in the way)
What is the visual mask neural noise study?
- Investigated whether ‘visual mask’ can itself be masked using single-pulse stimulation - unmasking the stimulus
- Without TMS: 100ms between unmasked letters and masked letters - detection rate was 37%
- TMS following the mask detection rate increased to 90%
- Unmasking was found between 60 and 140 ms stimulation after the mask
What is the virtual lesion approach?
- using repetitive TMS to interrupt or enhance cognitive processing
- Possible to inhibit cognitive functions for a long period of time using rTMS
- Can then be measured whether (and for how long) a specific cognitive task is impaired (usually slowing function instead of total loss of function)
- Used so that you don’t miss the critical window which can happen with TMS
- Need to be safety restrictions (can stimulate epilepsy in epileptic people
What is the ‘probing excitability’ approach?
- single-pulse TMS
-For motor system - does not disrupt cognitive functions - makes neurons fire more strongly
- measured by recording motor evoked potentials (MEPs)
using electromyogram (EMG) (electrical activity of muscles) - Stimulating left side of motor cortex can make muscles twitch on right side
- cannot measure causality
What is the mental rotation study?
- ‘probing excitability’ of TMS approach
- found that there was a good chance you need primary motor cortex when mentally rotating
- does not depend on strategy
What is the ‘probing information transfer’ approach to TMS?
- uses paired-pulse TMS
- Using two pulses delivered in brief succession: one usually sub-threshold while the other is supra-threshold
- Give sub-threshold TMS pulse to one brain region, then give stronger one (supra-threshold) to another brain region you think would communicate with the other one
- Pulse should trigger communication with first region - should send signals which would be already met by excitable cortex
- If there is an effect (ie if it there is a higher response, then you know that there is communication)
- Tests how brain regions talk to each other
What is the schizophrenia ‘probing information transfer’ study?
- it is suggested that there are abnormalities in motor cortex inhibition in schizophrenics (Cortical silence period (CSP) (period of suppression of tonic motor activity follows descending excitatory activity) is reduced)
- Researchers produced excitatory activity by first TMS stimulus to left motor cortex and measured excitability by assessing effect of second pulse
Results showed with and without medication showed stronger response to second pulse - Schizophrenics have stronger excitability of motor cortex - takes them longer to get rid of extra activity
What are the clinical applications of TMS?
- can be used for depression
What are the experimental t-test designs?
- one sample design
- between groups/independent measures design
- within-groups/repeated measures design
What is a one sample experimental design?
○ One group with values coming from different people -
compared to a single values
○ Advantages:
- Can be used to compare group data to
known values
○ Disadvantages
- May not always know population values
- May want to compare two groups, or
investigate change of behaviour over time
What is an independent measures experimental design?
○ Two groups and values come from different people
○ Results of the two groups are compared to each other
○ Advantages:
- Independent measurements
- Don’t have to worry about learning effects due to repeated exposure
○ Disadvantages
- People in the different groups might vary in
various ways - need large sample sizes to
average out these effects/need to
counterbalance all factors that we know
might have an influence on the results
- Cannot study behaviour over time
What is a repeated measures experimental design?
○ Single group provides data for both conditions at different time points
○ Advantages:
- Don’t have to think about differences
across groups
- Can study changes in behaviour over time
- Can usually test less people
○ Disadvantages
- Measurements are not independent - need
to calculate variance (t-test) differently
- People know the treatment after first
condition - can’t be naïve in second round
- Need to counterbalance conditions to
avoid unwanted order effects
What are the t-test assumptions?
- Observations must be independent (people must not influence other people’s values; no systematic biases when assigning people to groups)
- Populations from which samples are drawn must be normal
- If comparing two populations, sample must have equal variances
What is an EEG?
• Electroencphalography - method of directing neural activity by placing electrodes on scalp
• Electrodes pick up small fluctuations of electrical signals originating from activity of (mostly cortical) neurons
○ Reason we get this is because neurons communicate using electrical pulses when they generate actual potentials
○ Essentially listening to brain while it is thinking
• While raw signal recorded are noisy - systematically related to cognitive processes
• Can use these signals to learn something about cognition when people perform tasks
• Non-invasive
• Also possible to record intra-cranial EEG by measuring activity directly at exposed cortex - would only do for people who have skull already open and exposed like surgery patients
• Cheap and relatively easy to conduct
• Put a sensor on back of ear because there is a big bone there so you won’t receive neural signals, therefore can use it as a baseline to compare other results to
What is an advantage of EEG?
Temporal resolution is great
What is a disadvantage of EEG?
Spatial resolution is not so good
What is used to record an EEG?
- Electrode cap
- Amplifier
- Experimental stimulation
- EEG recording
What is the neurophysiology of an EEG?
• EEG activity does not reflect action potentials but originates mostly from post-synaptic potentials - voltages that arise when neurotransmitters bind to receptors on membrane of post-synaptic cell
○ Where axon of one neuron meets another
neuron (at the dendrites), they release
neurotransmitters
○ If neuron membrane is depolarised, new
action potential is created which travels
down the axon to continue passing on the
message
○ EEG does not reflect immediately when a
neuron fires, but it actually reflects the negative
potentials adding up at the dendrites, which are
the potentials that arise when neurotransmitters
bind to receptors because they have incoming
information for that neuron
○ Negative pole at dendrites and positive at cell
body
○ Causes ion channels to open or close leading
to graded changes in potential across the
membrane
○ This is understood as a small ‘dipole’
• Signals from single cells not strong enough to be recorded outside of the head, but if many neurons spatially align, their summed potentials add up and create signals we can record
• This pooled activity from groups of similarly oriented neurons mostly comes from large cortical pyramid cells
• Orientation of neurons determines sign of the recorded potentials (+ or -)
• Some orientations lead to signals which cannot be recorded
○ Eg if positive is facing sideways and negative is sideways, so no sign is pointing toward skull you get no signal recorded, or if they are facing against each other so the signs cancel each other out/if one layer is below another
• EEG signals do not reflect all activity in brain
What is the functional unit for an EEG?
• Functional unit is >10 000 simultaneously activated neurons
○ Meaning if you get more than 10 000 neurons in one place with the same orientation, then you can pick up signal with EEG
What are limitations of EEG
○ EEG is biased to signals generated in superficial layers of cerebral cortex on the gyri (ridges/peaks in brain formation) directly bordering skull
○ Signals from sulci (dips in brain formation) are harder to detect than from gyri and may additionally be masked by signals from the gyri
○ Meninges, cerebrospinal fluid (CSF) and skull smear EEG signal making it difficult to localise source
- Known as the inverse problem
□ If sources are known, resulting scalp reconfiguration of signals can be reconstructed, but reverse is not true - one scalp configuration of signals can have multiple dipole solutions
How are EEG signals measured?
-From scalp in relation to reference electrode
- To do this, can take average of all electrodes as
reference
□ However because brain has many dips and isn’t perfectly round, so bit noisy to use average of everything
○ Reference should be a neutral point (eg tip of nose/behind ear) but some people reference to the average of all scalp electrodes
What is the typical amplitude of an EEG signal?
10-100 microvolts
- Tiny signals so need to be amplified typically by a
factor of 1 000 to 100 000
- Signal then digitalised (sample frequency is typically
256-1024 Hz but can be >4000Hz)
How is EEG signal filtered?
○ Signal is band-pass filtered to remove the low (<0.5-1Hz) and high (>35-70 Hz) because cannot reflect brain activity
○ Signal is notch-filtered (at 50-60Hz) to remove line noise which is also not brain activity
- Attempting to remove those signals that are not from the brain
What are artefacts?
- Signals that are not brain signals
○ Eye blinks and movements have a strong impact on EEG signal because eye can be regarded as a dipole itself
○ Signals originating from the eye will contaminate the signal of interest - will be much larger
○ These signals can be recorded by placing electrodes next to and under the eye to capture vertical and horizontal eye movements
○ Eye-related signals can then be removed by excluding contaminated trials, or mathematical algorithms, such as independent component analysis (ICA)
What are the limitations of single-trial EEGs?
-Not likely to be very good and are very noisy - not very useful when you need to find brain activity that is reliably related to cognitive processes of interest
○ Need to collect a large number of trials (eg 20)
and average all of them to get a clearer pattern
• There Is a lot of variance between sessions from same participants and between participants - even with averaging
How to you obtain the Event Related Potential (ERP)?
• Average across all like events in the trials (eg in this example there were expected events and unexpected events - so need to average cognitive brain activity across all expected events and then separately average the activity in the unexpected events – now you can compare the two)
○ Isolate where modulation of your brain activity occurs when comparing expected vs unexpected events
How do you measure ERPs?
There are different aspects of the ERP component of interest that can be analysed
- Peak amplitude (used in 70% of studies)
- Area under the curve (because sometimes it is of
interest that although a peak may be higher, it is also
much narrower) (used in 20%)
□ Narrow peak could represent that cognitive
process finishes very fast, and wider
represents a more ongoing process
- Peak-to-peak (distance between negative and
positive peaks) (used in 10%)
□ The significance is in the rise of the peak, not
the highest height
○ No clear rule - results may differ between measures
○ Another option is to determine the onset of a component
- Looking at when a component starts to rise
-Difficult to determine exactly when it is rising
and when it is fluctuating
Why use ERPs?
○ One reason is that many components are very well studied
- Finding that a specific component is modulated by
the experimental task might shed light on what
cognitive process is involved
What did Woodman and Luck’s study on attention show/do?
- Used N2pc component - known to index attention: strongest over the posterior cortex contralateral (opposite side) to where observer is attending (shifting attention without moving eyes)
- Found that if you are attending to the left hemisphere/field, there will be a stronger (more negative) N2pc component in right hemisphere
- Interested in difference between N2pc for right and left
- Experiment:
□ Asked participants to search for a target (coloured
square open to the left)
□ Need to ignore all distractors
□ Can you do this by taking in an entire scene at once
and finding it (parallel search - don’t have to shift
attention around) or do you need to shift attention
around (serial search)
□ Theory using N2pc - attention should only shift if
need to serial search
- If search is parallel nothing should change
□ To get people to attend to one hemifield first -
manipulated probability that a specific colour was
the target
- One had 75% one had 25%
□ Prompted participants to quickly attend to the most
likely colour first and researchers could monitor
attention while participants were scanning field
□ Tested what happened on trials where there was no
target
□ Found:
- When no targets: found that when the two
colours where in different hemifields of the
screen, N2pc shifted - attention shifted to
field with more likely colour then to less
likely colour
◊ Serial search
- Not much shift was shown when there were
no targets and stimuli in same hemifield
- When more likely stimulus was target and
in different hemifield as other stimulus - no
attention shift
- When less likely stimulus was target and
stimuli in different hemifields - attention
shift
□ Therefore attention only shifts when serial searching is used
What did Gehring et al’s study on error-related negativity show/do?
□ Asked whether there is a cognitive mechanism for detection of and compensation for errors
□ Asked participants to emphasise either accuracy or speed in a Flanker-task
□ Incongruent displays should lead to more errors and error detection should only matter in accuracy condition
□ Found:
- ERN was strongest when emphasising accuracy,
weakest for speed
□ Is ERN indicative for compensating for errors?
- It it did we would expect that ERN should also reflect
attempt to break the error
- Greater ERN lower response force (force at which
they press button) - trying to correct error
- Greater ERN, higher probability to get it right on next
trial
◊ Successful learning from errors
- Greater ERN, slower response on next trial
What is fMRI?
- Functional Magnetic Resonance Imaging
- Creates an image - EEG and TMS don’t
- Good spatial resolution
- Measure signal - BOLD signal
- Use reverse inference to draw conclusions about cognitive processes from the presence of activation
- We usually want to know something about cognitve processes
- Know cognition happens in the brain
- How much can we really learn about it?
- Temporal resolution of fMRI is poor - takes a long time to record whole brain
why use MRI?
• One method used to scan the brain was Positron Emission Tomography (PET) - involves administering a radioactive isotope to the patient
○ Exposes patient to significant amount of
ionising radiation
• fMRI - do not work with radioactive substances - super harmless
How does MRI work (practically)?
• Creates a magnetic field
○ 1.5-9 Tesla (usually 3T for most functional imaging)
○ Earth magnetic field is 65microT - fMRI is a very
strong magnetic field
○ Cannot feel magnetic field - harmless
• Patient placed on the bed and moved to centre of magnetic field
• Head coil - like a helmet
• Experiments can be controlled from outside scanner room - tests can be given to patients while in the fMRI machine
• Can really investigate cognitive processes as you would in a lab
• Cannot have any metal in the lab
• Patient has a maximum of 1 mm movement - cannot move
• Participants can see a projection (usually computer-controlled experiment) via mirrors mounted on the head coil
• Responses can be given via scanner-compatible keys, joysticks, or a touchpad
• Head position is fixed to avoid any movement
Why does MRI work (in terms of physics)?
○ Human brain is 70% water
○ Hydrogen atoms can be thought of as small bar magnets “processing” (rotating) like a spinning top about an axis
○ Random spin directions of protons can be aligned parallel or anti-parallel to an externally applied very strong magnetic field in the MRI scanner - because they are a little bit like magnets
○ Align with the magnetic fields of the MRI but still processing about an axis - instead of doing it in random direction, now do it in systematic direction
○ Not perfectly aligned - also not static - keep processing in a random fashion
○ Some align with the field and some against, but most of them align with the field
○ Precession frequency of protons depends on strength of magnetic field
○ Axis along which the magnetisation is built up in the scanner = z-axis
Magnetisation along z-axis cannot be measured
A radiation frequency pulse (RF) is applied perpendicular to the magnetic field (delivered through head coil)
Also harmless
Frequency of the RF matches the precession frequency (frequency at which the protons ‘process’ about their axis)
○ Matching the frequency will cause the protons to absorb energy - has two additional effects
- It tilts the magnetisation vector into the
traversal plane (x-y plane)
- Aligns precession of the spins - protons’
rotations are in phase
○ The transversely rotating magnetisation vector can then be recorded as a signal
- The head coil is used to send the RF pulses
but it is also the receiver
○ Trick is to now switch off the RF pulse
- After switching it off, transversal magnetisation
decays - protons emit excess energy
- Lose phase coherence - every proton does
their own thing - still aligned with magnetic
field by processing at different stages
- Magnetisation along z-axis returns and
transversal magnetisation disappears -
processes are called relaxation - signal goes
away
- Independent
○ We are interested in how fast the signal goes away during relaxation
- Summed effect of many protons undergoing
relaxation is measured
○ Transversal magnetisation decays at different rates depending on the brain tissue (ie where in the brain)
- One reason is because of proton density - lose
coherence because they will be influenced by
other protons in their environment
- Different tissues have different number of
protons
○ The signals from different protons will get out of phase with each other and begin cancelling each other out
- Structural brain image depends on when signal is recorded during this process
How do you reconstruct brain images?
• First step: divide brain into slices
○ Can now vary the gradient field along the z-axis and know that the different slices were exposed to different field strengths
○ Slice-selecting gradient
○ Thus if different protons are in different magnetic
fields, precession frequencies will be different
○ Only one slice of the brain will be excited at a
time using specific RF pulse because the
precession frequency will not be matched for
the others
○ By exciting one slice at a time we get the z-
coordinate of all resulting signals
• Now can use second gradient to change magnetic field within this slice - vary along the y-axis
○ Protons in each slice also have different precession
frequencies
○ Gradient is called frequency encoding
gradient
○ Gives us the x-coordinate of the measured
signal
• Finally, very briefly using a gradient along the y-axis causes protons to speed up their precession according to the strength of the magnetic field for a short time
○ When switching off this gradient, all protons are
back to the same precessing frequency but are out
of phase with each other
○ Phase encoding gradient
• Allows you to specify the exact place within the brain after all three gradients because there is only one spatial position that could have a particular position/frequency
• Now we know precisely what we have done to the protons at each location in space - can use technique called Fourier transformation to reconstruct entire space
• Can measure slices in ascending, descending, or interleaved order to get 3D image of brain
What is a disadvantage of fMRI/why does it have poor temporal resolution?
- This whole process takes a lot of time to measure the brain just once
• Usually measuring one full 3D image takes 1-3 seconds (1.5-2s is standard)
What is difference between fMRI and MRI?
fMRI goes beyond MRI - measures brain activity
Images of the functioning brain
how does fMRI work?
• Oxygenated blood, oxyhaemoglobin, is diamagnetic, enhancing the signal (signal gets better)
• Deoxygenated blood, deoxyhaemoglobin, is paramagnetic - introduces field distortions (susceptibility artefacts) decreasing signal
• Neural activity is accompanied by a local increase in blood oxygenation - brlood is pumped to that region
○ Oxygenated blood is needed for glucose
metabolism
• Neural activity is also accompanied by local oversupply in oxygenated blood and therefore gives a better Blood Oxygen-Level Dependent (BOLD) signal
- large blood vessels cause ‘brighter areas’ (better signals) in the scans
• This technique makes use of the fact that all neurons need oxygen supplied from the blood
Means we can see local activity directly in the scans because the signal is slightly better
• In a typical fMRI experiment, BOLD signal within a region is measured while participants engage in a cognitive task
• Need to compare the activity against a control
• Differences in BOLD signals tell us about whether a brain region is involved in a task
Does fMRI measure brain activity?
No:
measures BOLD signals - an indirect consequence of neural activity
What is Statistical Parametric Mapping?
The analysis of fMRI results:
• Areas of enhanced activity can be mapped onto structural image of the brain
○ For the analysis, a General Linear Model is usually fitted to brain activity at each measurement point (voxel): significantly stronger activation in region X for task A compared to task B is interpreted as involvement of the region in task A
Why is repeated measurement of the brain needed for an fMRI?
• Repeated measurement of brain activity is required for the whole brain while performing experimental tasks because signal is very noisy
○ Like EEG - need to average the activity after
all the trials to get an accurate view of what
brain regions are active
What are the limitations of the BOLD signal due to biological foundations?
○ we have to be careful when interpreting differences in BOLD signal
- If all these complex processes are driving the
signal, it’s not a one-to-one mapping onto neural
activity
○ There is a substantial temporal lag between neural activity and the peak of the BOLD response - in the order of 8 seconds (signal reaches peak about 8 seconds after neural activity)
○ BOLD signal further needs up to 16 seconds before returning to baseline again
• Not valid to compare signals between different regions of the brain because the signal change is different
○ Can only compare the BOLD signal in one
region and then the BOLD signal in the same
region for a different tasks
What is the Haemodynamic Response Function (HRF)?
HRF is the measured response of different regions during fMRI (which look very similar in terms of shape)
What are the neural processes driving the BOLD signal?
• BOLD signal does not represent ‘neurons firing’
• Logothetis describes local cortical excitation-inhibition networks (EIN), small and highly interconnected functional microunits, which show massive recurrent feedback
○ Like when you and your group of friends are talking about going to the movie - lots of excitement with some very keen to, others not so keen to - doesn’t matter if you reach consensus so long as there was excitement in your group
○ Not neurons firing or causing anything to happen, you can only see the interactions/excitement in the brain
• Feedback processing within EINs (which determine what the output of the microunit will be) might account for most activity measured by the BOLD fMRI
What are the limitations of the fMRI?
• Problem 1:
○ Because we see blobs in the brain, the organisation of the brain’s architecture must be ‘modular’ (brain has one functional unit in charge of each activity)?
- This is not true - often we do not see the
full networks involved
- fMRI only shows ‘tip of the iceberg’
- Can never conclude that nothing else is
going on than shown by BOLD signals
- fMRI might not always map the functional
units that matter
- Just because it’s what we see in our scans
does not mean it is exclusively what is
happening
• Problem 2:
○ fMRI has poor temporal resolution (takes time)
- Given that HRF is slow, very fast processes are
very difficult to image
- Because it takes about 2 seconds to measure
the brain at once this does not allow exploring
any processes that take place within these 2
seconds
- We need to use some tricks in our
experimental designs to separate events of
interest if we don’t wait for HRF to reach
baseline every time
• Problem 3:
○ Spatial resolution for fMRI is good but
- It’s also not great - smallest measurement
unit is a voxel - a 3D pixel
- Standard voxel size is 3x3x3 mm and we do
not learn anything that happens within a voxel
- But one voxel still contains >100 000 neurons
• Problem 4:
○ Multiple comparisons problem
- For such a complicated method only run a t-
test for every voxel in the brain for comparison
of condition A vs B
□ Two values per voxel
- However we have more than 50 000 voxels in the brain and we somehow need to correct for false positive findings (if it was caused by chance even if there is only 5% that it was) which are very likely to occur with so many tests
□ 5% risk of making a false judgement
with every voxel - this really sums up
if we do it 50 000 times
- Would expect 5000 false
positives
- Strictest correction of this is Bonferroni-correction - divide significance level by number of tests (eg voxels) and use this new significance level for each test - therefore overall risk of false positive will be 5 (or whatever the normal value is for p)
what did Kanwischer’s fMRI study of the brain’s recognition of faces show?
○ Presented participants with images of faces and contrasted BOLD signals to when participants saw objects (some faces some not faces)
○ Why is the brain so good at processing faces?
○ Found that a region located in the fusiform gyrus (in temporal lobe) responds more strongly to faces than to objects
○ They could show this result reliably in most of their participants, and could replicate it with different participants
○ Does that mean this area is specialised in the processing of faces?
○ To rule out this result being imply due to using objects as a control category, they replicated the study with a different control category: faces vs scrambled faces
- Taking elements from the face and rearranging
them so they no longer look like a face
○ The same brain area was strongly activated for faces but not for scrambled faces
○ Maybe it’s the same for other body parts, not just faces
- Contrasted faces with hands
- Same brain region was activated with faces rather
than hands
○ Led researchers to name the region ‘fusiform face area (FFA)’ - so sure this is the brain’s module for face processing
What are other brain ‘modules’ in the visual brain?
○ Fusiform Face Area (FFA)
○ Parahippocampal Place Area (PPA) - houses and places
○ Extrastriate Body Part Area (EBPA)
What are some criticisms of FFA and its implications of brain modularity?
○ 1: If our brain has one are specifically designated just to faces or places, what does that mean about other millions of objects we recognise? Not enough brain space
- Could it be that it only looks like this region is
specialised in faces
- Gauthier et al. Greeble study:
□ Asked participants to distinguish
between different ‘Greebles’ - made
up faceless figures which no
participant had ever seen before
□ During the experiment, participants
learned the family structures and
became experts for Greebles
□ First, when they did not know much
about Greebles, FFA responded
strongly to faces, but not to Greebles
as predicted
□ However, after learning to
distinguish really well between
individual Greeble families, the FFA
also responded for Greebles
□ Therefore FFA is not a face area, it is
an expertise area
○ 2: visual system might not be organised by specific object categories, but by where in our visual field objects are usually encountered
- Organisation in ventral visual cortex might follow
cortical topography (eccentricity mapping) - where
do you usually encounter things
- Coding is driven my resolution needs - FFA is good
for everything that usually requires ‘high’ resolution,
faces just happen to be centre of our vision
- Face area corresponds to brain regions that would
also activate when you focus on a very specific
point
- We encounter faces in the centre of our vision not
the periphery
- We usually need a high resolution to recognise
faces, simply because we need to see the details
- Might not be a matter of what it is but where it is
- The module for places/houses is in reality better suited for processing the periphery - that’s where house are usually encountered
What does the evidence of the FFA and its criticisms suggest?
• Researchers have found evidence for all three theories - suggested all might be true to some extent
• The fMRI signal might therefore reflect a mixture of all three coding schemes
○ Can’t tell from the BOLD signal that it’s all true
§ Might be overlooking evidence that is equally likely to produce the results - ‘tip of the iceberg’ - does not represent the only thing that is going on
What was Haxby’s argument against brain modularity?
○ Argued that in order to represent all possible objects we know, objects must be represented in a distributed fashion - meaning all objects are represented in the entire ‘object region’ in the brain
○ In their study, they showed participants many exemplars for different objects - including objects - while fMRI was recorded
- Ran correlations between distributed activation
patterns for exemplars within object categories and
between object categories
- If the entire pattern codes for an object, then there
should be high correlations for objects within
categories as opposed with between categories
- Found that correlations were higher within
categories than between categories
- This was still true when most responsive voxels for
each category were excluded - eg FFA for faces
(so basically when you don’t consider the FFA, the
correlations were still stronger when comparing
face-face response as opposed to face-object
response and the same for all of the objects)
- Therefore throw away the brain area for faces and
the brain still recognises faces
- The distributed patterns of large and small
responses, not just modules, were associated with
the object categories
What is the Problem of Reverse Inference with regards to fMRI?
○ Usually apply following logic:
- In this study, when task A is done, then region Z is
active
- In other studies, when cognitive process X
happens, then brain region Z is active
- Therefore, in this study, activity in Z –>
engagement of cognitive process X
○ One problem is that the second point is not exclusive - brain region Z could be active for many tasks
- Brain regions are very flexible
○ Problem is if we find activation in a region which is part of this multiple-demand network we still don’t really know what the region is doing
○ If a brain region is activated by many cognitive functions, we learn very little from observing activation in those areas
○ How good is task A at manipulating the cognitive process X?
- If the task measures more than one cognitive
function, we also don’t learn much
○ The probability of that we really learn from fMRI results that cognitive process X is involved depends on:
- The quality of the task to measure the cognitive process
- The specificity of region for this cognitive process
How does the difficulty to understand the prefrontal cortex show an example of a region being active for multiple tasks?
- Some researchers have concluded from looking at results from many studies that many anterior regions (towards front of brain) represent more abstract information, and more posterior regions represent more specific information
- Others concluded that most regions in the frontal cortex can actually be found to be activated in many different tasks
- Duncan argued that the frontal cortex shows relative, but not absolute specialisation
□ Means prefrontal regions might just be recruited ‘more strongly’ if the task at hand becomes more difficult
□ True for other regions too
Explain the other issue with fMRI (being the overinterpretation of null results)?
○ What does it mean if you find that no region was significantly stronger activated for task A vs task B?
- We don’t really know
○ Our statistical tests are designed to make it difficult for the H1, not the H0, meaning we can’t really interpret null results
○ Also don’t know whether method might just not be sensitive enough to detect small differences (if just some neurons within each voxel might show differences, but not many, or there are neurons coding for both/many conditions within a voxel)
- Should always avoid concluding that brain
regions are not involved in a cognitive process
What are nucleotides?
The building blocks of the genetic code
What are the four different bases in DNA?
§ Adenine (A)
§ Cytosine ( c)
§ Guanine (G)
§ Thymine (T)
What are amino acids?
the building blocks of proteins
- A specific sequence of three bases constitutes
genetic code for a particular amino acid
How many bases are in the human genome?
○ 3 billion bases in the whole human genome
○ 20-25 thousand genes that code for proteins
What is the structure of DNA?
○ DNA helix is double stranded
- Two strands carry redundant information
- Each base pair has a partner on the other side
□ C-G
□ A-T
- So if you know the base on one side you
automatically know the base on the other
○ DNA bundled in chromosomes
- Human karyotype comprises 46 chromosomes
□ 22 pairs of autosomal chromosomes
(1-22)
□ Two sex chromosomes (XX/XY)
What is a codon?
Sequence of 3 bases representing an amino acid
How do genetic variants occur?
○ Function of a protein is determined by its structure
○ Structure of a protein is determined by its sequence of amino acids
○ A change to a single base can change the amino acid (not always)
- Changing the amino acid can change the
structure and function of the protein
○ A single-nucleotide polymorphism (SNP) is a position on the genome at which the base (nucleotide) differs between individuals
- EG some may have a G where others have a T
- The two alleles of a SNP are the alternative
bases
□ In this example, T is the major (most
common) allele, and G is the minor
- An individuals genotype at a SNP is
determined by the two alleles on the copies of
the chromosome
- An individuals phenotype is the presence,
absence or value of a trait of interest
□ Psychological diagnosis (binary
phenotype)
□ Parenting style (categorical
phenotype)
□ IQ (quantitative phenotype)
What are possible genetic variants?
○ Insertion-deletion variant
- Bases added or missing (where one might have
GCG, that is not in the other one)
○ Block-substitution variant
- Multiple bases substituted
○ Inversion variant
- Bases are replaced with reverse sequence of the
other strand (so if the sequence on one is AATCG,
the other one would have the backwards-version
of the base pair: CGATT)
○ Copy-number variant
- Sequence of bases repeated multiple times
Is a mutation rare or common?
§ Rare: <1% of alleles in population
Is a polymorphism rare or common?
Common: greater than or equal to 1% of alleles in population
How is an excess dosage of the X chromosome proteins in females avoided?
one copy of the X chromosome in each cell is silenced or inactivated
- This process is random in each cell
○ When there are two X chromosomes in one cell
- XIST gene produces an RNA transcript (an
intermediate step in the process of converting a
gene encoded in DNA to a protein) that coats one
chromosome, which is inactivated as a Barr body
- TSIX gene on the other chromosome produces an
RNA transcript that suppresses transcription of
XIST
- TSIX is antisense partner of XIST
□ Both are encoded by the same stretch of DNA but are transcribed in opposite directions
What is heritability?
○ Phenotypic variance (P) = variance from genes (G) + variance from environment (E) + variance from gene-environment interactions (GxE) + covariance between genes and environment (2covGE)
○ Heritability (h) is proportion of phenotypic variance due to genetic causes
§ H2=G/P
§ Local measurement: specific to a given population at a particular time
§ Varies with amount of variation there is in a population in genes and environment
How do we measure heritability?
§ Before advent of molecular genetics, used genetic epidemiology
§ Study designs in genetic epidemiology exploit the fact that related individuals share a predictable amount of genetic material
§ Eg twin studies: look at concordance in twins (if one twin has a trait or disorder does the other have it too?)
□ Monozygotic (MZ) twins share 100% genetic material, dizygotic (DZ) twins share 50% genetic material
□ Higher concordance in MZ pair than DZ pairs suggests genetic component
® Assumes they are equally similar environmentally
What are the two modes of inheritance?
§ Dominant vs recessive:
□ Dominant traits require mutation on one copy of the chromosome for expression of the phenotype
□ Recessive traits require mutation on both copies (or only copy) of the chromosome
§ Autosomal vs x-linked
□ Autosomal traits are carried on one of the first 22 chromosomes
□ X-linked traits are carried on the x chromosome
What are the rules for dominant traits?
- Dominant traits can’t skip generations
- Two unaffected parents of a dominant trait cannot have affected offspring
What are the rules for recessive traits?
Two affected parents cannot have unaffected offspring
What are the rules for autosomal traits?
Equally common in both sexes
What are the rules for x-linked traits?
- Cannot transfer father to son
-If dominant: daughter of affected father must be affected - If recessive: Father of affected daughter must be affected
◊ More common in males
What is fragile X-syndrome?
- Fragile X syndrome is monogenic disorder because we can tract its origins to a single gene
-Can cause sever mental retardation - average IQ
of 40
- Results from a copy-number variant of CGG
sequence in the 5’-untranslated region (UTR -
contains the promotor region where chemicals
bind to start the process of transcribing a gene
into a protein) of the gene FMR1 (fragile-X mental
retardation protein)
-FMR1 - critical to synaptic plasticity
-Synapses to strengthen or weaken
over time in response to
increases/decreases in their activity
- forms neural basis of learning
- Expansion of the repeated CGG sequence of
bases triggers methylation process (many
repeated CGG sequences triggers methylation
whereby chemical compound call methalyn binds
to DNA)
- Constricts the X chromosome and results
in ‘fragile’ appearance
○ Methylated promotor region prevents
transcription of the gene - promotor region
also becomes bound by methalyn meaning
it isn’t there when it’s required to allow
synaptic plasticity
What are the two types of disorders?
• Monogenic disorders
○ Can trace the origins to a single gene
§ Huntington’s disease (HTT)
§ Fragile X syndrome (FMR1)
• Polygenic disorder
○ Monogenic disorders are the exception to the rule in behavioural and psychiatric genetics
§ No single gene for schizophrenia, autism, bipolar disorder, depression or anxiety
What is the purpose of Genome-wide association studies (GWAS)?
• Examine the statistical association between a phenotype and many SNP markers throughout the genome
○ Typically 500 000 - 2 000 000 markers
Why do you not need huge samples in a GWAS?
• Can sample common variation sparsely (about 1 in every 1500-6000 bases is genotyped on average - median size of a human gene is 25000 bases) because linkage disequilibrium (LD) allows us to observe indirect observations
○ Chromosomes are mosaics and many variants are correlated
• In a direct association the phenotype has a functional association association with a genotype (measured) SNP
• In an indirect association, the phenotype has a functional association with a non-genotyped SNP that is in LD with a genotyped SNP
Essentially:
• We don’t have to take huge samples, because even if the sample you take does not contain the SNP you are looking for, the linkage disequilibrium (LD) means the desired SNP can have indirect effects on the SNPs near it
What is the allelic dosage model?
-Used for quantitative traits
○ Categorise everyone according to whether they are a major homozygote, heterozygus, or minor homozygote
- Plot phenotype score according to genotype
- Is there a statistical association between the phenotypic measurement and the number of copies of the minor allele?
- Could be used to determine frequency of alcohol use and IQ - (quantitative traits)
What is the allelic association model?
- used for categorical and binary (yes/no) traits
-compare the case group with the control group
○ Is one of the two alternative alleles statistically overrepresented in a phenotypic group?
○ Could be used for diagnosing bipolar disorder (because it is a binary trait - you either have it or you don’t)
What is the Manhattan plot
- Graphically summarises reuslts of all individual tests of association
- Each point represents outcome of a test for one SNP
- Physical location of the genome and within a chromosome is on horizontal axis
- Transformed p value is on vertical axis
- Lower p values are higher on the axis - indicating strongest associations
- Threshold for significance are stringent because multiple comparisons increase the likelihood of type 1 error
- threshold is p=0.00000005
- Corresponds to Bonferroni correction for approx. 1 million independent uncorrelated tests
What are imputations of Manhattan plot?
○ Predict genotypes at non-genotyped SNPs
- Relies on data from a reference panel of
individuals genotyped at high density
§ Applies patterns of LD discovered in the reference
panel - use those to predict genotypes in non-
genotyped SNPs independent of phenotype
§ Run those tests of statistical association as though
though impudent SNPS were really genotyped
What does the genetic distance and recombination rate reflect?
the frequency with which two markers are inherited together
○ Helps define the region likely to contain the functional variant
What does conservation indicate?
the extent to which a sequence is maintained across species
High conservation suggests an important function preserved during evolution
How could ‘Skyscrapers’ observed in Manhattan plots could be explained
○ Multiple SNP LD with a functional SNP
○ Or multiple functional SNPs in the same gene
What are the molecular genetics of bipolar disorder?
○ Meta-analysis of bipolar disorder GWAS found association with the ANK3 and CACNA1C genes
§ Proteins transcribed from both of these genes regulate the flow of ions in and out of neurons during an action potential
§ Both genes are down-regulated (much less of the protein is transcribed) by lithium
§ Lithium is an effective treatment for bipolar
How can GWAS provide new evidence for existing hypotheses?
○ Schizophrenia has previously been linked to abnormal dopamine signalling
○ Antipsychotic drugs block dopamine receptors - has been primary evidence in favour of the dopamine hypothesis
○ Find now from GWAS that DRD2 (a dopamine receptor gene) is associated with schizophrenia
How can GWAS raise new possibilities
○ In this case of schizophrenia, most significant association is the major histocompatibility complex (MHC)
- MHC genes code for cell-surface proteins that
allow the immune system to recognise foreign
substances
- Does acquired immunity play a role in the aetiology of schizophrenia?
How can GWAS point to environmental risk factors?
○ Variants in the CHRNA5-A3-B4 gene cluster are known to be very strongly associated with heavy smoking
§ Encode subunits of nicotinic acetylcholine receptor, chlolinergic receptors that also respond to nicotine
○ Smoking has more than 80% prevalence in people with schizophrenia
○ Association of schizophrenia with CHRNA5-A3-B4 variants suggests heavy smoking may contribute to schizophrenia risk