8.1 fMRI experimental design

Question 1

What is the point of optimising the experimental design of your study?

Answer 1

so you can later analyse the collected data in the best way possible

Question 2

What generally are the two types of MRI?

Answer 2

structural (T1, T2 etc)
functional (fMRI, fcMRI)

Question 3

Does fMRI investigate brains on a neuronal level?

Answer 3

no! fMRI measures brain activity many layers above that

Question 4

What is the purpose of including experimental controls?

Answer 4

to remove the effect of confounding variables

Question 5

What are the different types of experimental designs used in MRI?
What sort of statistical test would each of these study designs use?

Answer 5

-between group comparisons (independent t-test)
-within-group comparisons (paired t-test)
-correlations (regressions)
-mixed designs

Question 6

What is a mixed design study?

Answer 6

a study with ie multiple different groups and multiple different conditions

Question 7

What is the control group when investigating patients with carriers of an allele of interest with fMRI?

Answer 7

non-carriers of the allele of interest

Question 8

Can you have fMRI experiments without assigning a task for the participants?

Answer 8

yes! these are fMRI task-free designs

Question 9

What happens when our brains are ‘at rest’?

Answer 9

never doing nothing: often we engage in forms of introspective and abstract forms of cognition

Question 10

What happens during functional connectivity of brain networks?

Answer 10

neural networks activate and deactivate together in synchrony

Question 11

What is the default mode network (DMN)?
Which disease has defects in the DMN?

Answer 11

four regions of the brain which are correlated with functional connectivity during resting state. these regions activate simultaneously during rest

Question 12

What are spurious correlations?
How to avoid spurious correlations?

Answer 12

strong correlations between two variable which happens by chance and there is no underlying causality
correlation does not equal causation
-have a strong apriori hypothesis in which you believe the experimental variables are correlated

Question 13

What are the advantages of using MRI?

Answer 13

some are ‘in relative’ terms
-versatile (structural and functional MRI)
-in vivo methodology
-longitudinal designs
-simple statistics
-‘safe and cheap’
-‘translational neuroscience’

Question 14

What are the disadvantages of using MRI?

Answer 14

-cant have any metal in your body (safety issue as participations are not always clearly written)
-MRI is rather devoid of biological information
-temporal resolution is ‘suboptimal’ as compared to other neuro-techniques (EEG)

Question 15

What generally are the two types of MRI?
What are the two main areas which structural MRI experiments investigate?
ditto functional MRI?

Answer 15

-structural and functional
-structural: brain volumes and cortical thickness
-functional: resting-state fMRI and task-based fMRI

Question 16

What is real-time fMRI?
Why is it an improvement from regular MRI?

Question 17

What are the applications of real-time fMRI?

Answer 17

-novel brain-machine interfaec
-planning neurosurgical interventions (eg. neurosurgeon uses rt fMRI to know which area of brain to remove to treat epilepsy)
-coupling of conscious experience and brain activation
-neural feedback (show images to participant and tell participants do to certain things to improve activation in certain area in the for the next image acquirement)

Question 18

What are the advantages of real-time fMRI over other MRI feedback methods?

Answer 18

-can directly target specific brain regions with superior spatial resolution due to rt fMRI
-overcomes limitations of other MRI feedback methods (EEG)

Question 19

How did experimenters show that neural feedback was possible with real time rt fMRI?

Answer 19

used a sham and actual rt feedback. actual rt showed an increase in activation in a particular brain region as actual rt was given to participant whereas sham showed no correlation

Question 20

What does MVPA stand for?

Answer 20

multivariate pattern analysis = looking at activation distributions of two or more voxels (under different conditions for each voxel) with two different channels

Question 21

What is the advantage of doing MVPA instead of univariate?
What does univariate investigate?

Answer 21

looking at the conjunction of the two recording channels shows when both voxels are active together whereas univariate cannot. you can get more detailed information about the relationship between two voxels from a multivariate than univariate.

univariate investigates the activity pattern of a single voxel independently

Question 22

What is a representational dissimilarity matrix (RDM) show?

Answer 22

a correlation matrix which visualises pairwise dissimilarities between the neural activity patterns due to different stimuli/conditions

Question 23

Kriegeskortes et al. 2008
Which pair of stimulis had the lowest dissimilarity in the RDM? (between human faces, artificial objects, animals etc)
How does the pattern of the RDM in from early visual cortex to a whole human brain?

Answer 23

pairwise stimuli of human faces showed the least dissimilarity in neural activity patterns

-fuzzy pattern showing no categorisation between the same stimuli in early visual ctx -> ev ctx not interested in categorisation of stimuli just the shape/aesthetics -> things are group together if they look similar
whole human brain has pairwise patterns of dissimilarity based on conceptual meaning of stimuli

Question 24

Did the monkey brain RDM have a pattern more like the human or the early visual cortex RDM?

Answer 24

more like human RDM because monkey reacts to visual stimuli on a conceptual basis and categorises based on concept

Question 25

What cognitive process (+brain location) is impaired in alzheimers?

Answer 25

memory functions of the hippocampus are impaired

Question 25

What was a confounding result of activation from people genetically at higher risk of alzheimers disease in astudy by Bassett et al. 2006?

Answer 25

Increased activation at medial temporal lobe for high-risk individuals comparing to low-risk controls during encoding but decreased activation during recall (Bassett et al., 2006). controls=not high-risk for AD

Question 26

Why was using MVPA with RDM a good way to show that people at high-risk of developing AD have worse memory impairments than healthy controls

Answer 26

more complex analysis than Bassett et al.: used stimulus specific information using MVPA instead of overall activation -> showed that there was worse ability to encode information than controls in parahippocampus