Chapter 5.3
What is estimation in the context of study design?
"What is the precise shape of the hemodynamic response associated with a task?"
Draw the Hemodynamic Response Function with labels.
"What is ..."
Fig.1
True or False: Multivariate pattern analysis (MVPA) focuses looks at patterns of activity at each point in isolation.
"What is False?"
This common baseline condition was found to be problematic as it showed higher brain activity than other tasks.
"What is rest?"
What is fMRI preprocessing?
"What is cleaning and standardizing the acquired data before statistical analysis?"
What is detection in the context of study design?
BONUS: what is the example used in the chapter to illustrate the difference between detection and estimation?
"What is detecting and estimating the magnitude of activation?"
BONUS: "What is you have a pile of sand on the floor, if you get closer you can see the shapes of the granules on the outside of the pile but not in the middle. If you spread it out you can see all of them up close, but if step away now you can't see the "pile of sand?"
Fill in the blanks: "As a mnemonic rule, the complete process is reflected in the technique's name magnetic (_______ magnetic ____) resonance (the matching of _______ between the RF pulse and the precession of the spins) imaging (the process by which the _____ is measured by the MR scanner is spatially encoded and the computer algorithm that produces the images)
"What is: nuclear, spins, frequency, and signal?"
What are pattern estimators?
"What is a statistical method used to estimate the brain's activation pattern in response to each specific trial or stimulus within an experiment?"
True or False: Neural activity during long rest periods can reduce, eliminate, or even reverse the sign of activity during a cognitive task but not during shorter periods of rest.
"What is False?"
List 5 tools that can be used for fMRI preprocessing.
"What is: fMRI Prep, FSL, AFNI, FreeSurfer, SPM?"
This design technique starts with a blocked trial structure and then randomly shifts stimuli to break up block patterns, optimizing for both detection and estimation of neural responses.
"What is permuted block design?"
Explain what factorial design is using an experiment, and describe the main effects and interaction effects in the said example.
"What is....?"
EX If studying attention and emotion, attention might have two levels (focused vs. divided), and emotion might have two levels (positive vs. neutral).
Main Effects would look at à attention (focused vs. divided) & emotion (positive vs. neutral) separately
Interaction between attention and emotion—whether brain activity under focused attention changes differently when processing positive vs. neutral emotions
Does the pattern estimator matter? Why or why not?
"What is yes, depending on the study design certain estimators might produce more accurate, stable estimates, while others may increase error due to overlapping or correlated activations?"
True or False: Activity in the medial temporal lobe was consistently lower during rest than during other baseline tasks.
"What is false?"
Activity in the medial temporal lobe was higher during rest than during several alternative baseline conditions
Does the order in preprocessing matter? Use motion correction and slice timing correction to explain why or why not.
"What is yes because the order of the steps can have a significant impact on the quality and accuracy of the analysis that follows?"
If motion correction is done first, it gives a cleaned-up dataset where motion has already been accounted for before any other correction, this is important when the data has a lot of head movement. Mitigates risk of incorrect modeling of the hemodynamic response function (HRF)
If slice timing correction is performed first motion-induced shifts in signal intensity might lead to inaccuracies this could let motion artifacts through the times series
Describe a blocked trial with an experimental example, & and what would it look like? What's the research question of your example that would lead to that design?
"What is....
Ex:
AAAA | BBBB | AAAA | BBBB
"Which part of the brain is associated with adaptation and recognition of patterns?"
According to the article, what is the optimal suggested TR?
BONUS: why is that?
"What is ~1.5 seconds?"
BONUS:
The brain's blood flow response takes several seconds, so extremely fast TRs (< 1s) don't capture more useful signal
Shorter TR (< 1.5s): Not enough time for full signal recovery
Longer TR (> 1.5s): Might miss important temporal changes
How does temporal autocorrelation play a role in inflating Type I errors in MPVA? (also briefly explain what temporal autocorrelation is)
"What is temporal autocorrelation (consecutive data points in an fMRI time series are correlated with each other, rather than being independent)?"
"What is when brain responses from consecutive trials are correlated, MVPA could misinterpret the "spillover" as the consistent pattern of brain activity inflating false positives?"
Though not explicitly mentioned in the paper a lot of the "rest" activity is similar to....?
*hint coin
"What is the default mode network?"
concept of spontaneous cognitive processes during rest (daydreaming & self-reflection) closely align with what we know about DMN activity
What's the difference between spatial normalization and spatial coregistration?
BONUS: before spatial normalization can be done three steps have to be completed what are they?
"What is spatial coregistration is the alignment of multiple images from the same subject to make sure they correspond spatially (func to anatomical) while spatial normalization is the process of transforming individual brain images into a standard anatomical space or template?"
BONUS: "bias field correction, brain extraction, and tissue segmentation"
Draw how a jittered design can improve an experiment. Explain what problem it's solving and create a scenario in which this would be relevant.
"What is...?"
EX page 97.
What is the result of the resolution trade-off? Explain the tension between what's involved, and what has to be sacrificed in each scenario.
"What is the compromise between spatial and temporal resolution?"
For higher spatial resolution → Need more time to encode more spatial positions → Results in slower temporal resolution
For higher temporal resolution → Must sacrifice spatial detail or brain coverage → Results in lower spatial resolution or incomplete brain imaging"
We've talked a lot about experimental design, so let's test something out...what design method did Mumford et al. use for their experiment?
*hint coin
"What is factorial design (they manipulated two or more independent variables to observe both their individual and combined effects on the dependent variable?"
Independent Variables: Trial order (blocked, alternating, random) and ISI (3s, 7s, 15s).
Dependent Variable: False-positive rates in pattern estimators for MVPA.
Could the results from this study apply to non-memory-related studies? How so? Provide an example.
"What is...?"
EX yes other studies could benefit because knowing the complexity of brain activity during baseline conditions can better help us design studies with clearer results and less noise
What is structured and random noise? How does temporal and spatial smoothing mitigate these types of noise?
"What is structured noise arises from physiological processes that can correlate spatially or temporally, creating predictable patterns; random noise is typically thermal noise that has no predictable pattern and reduces the signal-to-noise ratio"
For Structured Noise
spatial smoothing averages the signal of each voxel with its neighbors, reducing structured noise and random noise by enhancing genuine brain signals and canceling out correlated noise.
Temporal Smoothing verages signals over adjacent time points, helping to mitigate fluctuations associated with structured noise, & by filtering out high-frequency random fluctuations over time, allowing legitimate signals related to brain activity to be more easily identified