Lecture 4- analysis and experimental design

Question 1

Typical preprocessing steps for fMRI analysis

Answer 1

Spatial smoothing
Motion correction
Temporal filtering

Question 2

Statistical maps in fMRI show the probability of…

Answer 2

Type 1 error

Question 3

To support robust statistical inferences which can be extended to a wider population, multi-subject fMRI experiments typically employ…

Answer 3

random/mixed-effects, within-subjects design

Question 4

Why are voxel’s activity modelled by combining predicted time series for each distinct class of event (or block) which are fitted to the fMRI data?

Answer 4

to minimise residual variation

Question 5

If BOLD responses to individual experimental trials of the same task are to be compared (e.g., correct versus incorrect trials) in an fMRI study, it is essential to use:

Answer 5

an event related design

Question 6

Advantage of block design

Answer 6

High signal to noise

Question 7

In an fMRI experiment, what is the IV and the DV?

Answer 7

IV= task 
DV= BOLD signal

Question 8

What is cognitive subtraction?

Answer 8

Choose 2 tasks which differ in one critical aspect

Then compare BOLD signal in one from the other

Question 9

How can confounding factors be controlled for?

Answer 9

Randomisation and counterbalancing

Question 10

Why should fMRI experiments be within-subjects?

Answer 10

To reduce individual variation in BOLD signal and differences in anatomy

Question 11

What is block design?

Answer 11

Conditions are divided into distinct blocks
Typically several trials within each block (5-30 secs)
IV = constant within a block

Question 12

What does the transition between blocks represent?

Answer 12

Changes in the level of the IV

Question 13

Advantage of longer interblock intervals

Answer 13

the hemodynamic response can return to baseline during the control condition = more signal change

Question 14

Advantage and disadvantage of shorter interblock intervals

Answer 14

Allows for more repetitions
BUT
The sluggish nature of hemodynamic response means less signal change

Question 15

Disadvantage of block design

Answer 15

Response to individual trials cannot be distinguished

Question 16

Why is there a high signal:noise ratio in block design?

Answer 16

Because of plateau within a block, there’s excellent power to detect active voxels

Question 17

What is an event related design?

Answer 17

Each event (trial) is modelled separately

Question 18

How can responses be estimated for different classes of trial in an event related design?

Answer 18

Carefully spaced and time-locked trials

Question 19

Disadvantage of event related design

Answer 19

Reduced detection power (smaller signal change)

Question 20

Advantage of event related design

Answer 20

Data can be analysed according to PS response

Time course of BOLD response can be accurately estimated

Question 21

How is motion corrected?

Answer 21

Successive images are aligned with one another

Question 22

What is spatial smoothing?

Answer 22

Blurring by averaging a few voxels together in space = increase signal to noise

Question 23

What is high-pass temporal filtering?

Answer 23

Gradual changes in signal intensity removed

often caused by breathing and heart beat

Question 24

In the GLM, what do beta weights tell you?

Answer 24

tells you how much a voxels signal increases in response to particular stimuli/task

Question 25

BOLD measures the concentration of WHAT?

Answer 25

deoxyhaemaglobin

Question 26

In fMRI analysis using GLM, what do we KNOW (input) and what do we WANT to know (output)

Answer 26

INPUT: a voxels true BOLD signal and our predictors OUTPUT: beta weights that combine with the predictors to give us the best approximation of the true signal

Question 27

What are some predictors in the GLM?

Answer 27

Baseline activity Response to task condition Response to control condition

Question 28

What do we combine with the predictors to best estimate the true signal from a voxel?

Answer 28

beta weights | error signal

Question 29

What mathematical tranformation do you apply to predicted time-series of activation?

Answer 29

Haemodynamic Response Function (HRF)

Question 30

In GLM analysis, the aim is to find the __-__ that best appoximate a voxel's signal time-series. The best approximation is the one with the ___ ___ Then, a comparison is made between the ___-___ for the task condition to the ___-___ to control condition.

Answer 30

beta-weights least errors beta-weights beta-weights

Question 31

A voxels BOLD signal at rest is 100. When the PS is reading scentences, the signal increases to 120. When reading nonwords, the signal increases to 110. When approximating the voxel's BOLD signal using our three predictors, what should the beta weights be for sentences and non words?

Answer 31

20 , 10 the beta tells you by how much the signal increases from baseline!

Question 32

What are the components of the GLM?

Answer 32

BOLD signal = design matrix (where each column is a predictor)*their beta weights + errors OR explained variance/task related activity changes + unexplained variance

Question 33

What do we give the GLM? What do we know?

Answer 33

The BOLD signal | X/The design matrix (each column is a regressor that we built ourselves)

Question 34

What does the GLM give us? What do we want to find? How does it find it?

Answer 34

vector of beta weights (one per regressor in X) that gives the best approximation of the BOLD signal By minimising the sum of squared errors.

Question 35

Two reasons why we square the errors?

Answer 35

To get rid of negatives It gives a big weight to big errors and a smaller weight to small errors (because squaring a big number disproportionately increases it compared to squaring a small number)

Question 36

A comparison of beta-weights is called a ____.

Answer 36

contrast

Question 37

What is a beta-weight?

Answer 37

A weight indicating how much a predictor contributes to the true BOLD signal. e.g. how much the signal changes due to the hypothesised process, holding everything else constant.

Question 38

The error can be defined as the difference between....?

Answer 38

The difference between the true BOLD signal and the combination of predictors that best approximates it. These are changes (variations) in the signal we cannot explain with the predictors

Question 39

Contrast is the difference between...?

Answer 39

The difference between two (groups of) betas.

Question 40

the contast t-value is the number resulting from dividing the ____ ____ by a measure of ____.

Answer 40

contrast value (e.g. beta for task - beta for control) DIVIDED BY Error

Question 41

What do statistical maps indicate and how do they represent it graphically?

Answer 41

Indicate the probablility of a type-1 error Display a t/z statistic that is converted to p-value The lower the probability (p-value) the brighter/hotter the colour They are overlain on a anatomical image Threshold is used to exclude higher p-values

Question 42

What is bonferroni correction?

Answer 42

Takes into account the number of comparisons

Question 43

What is FWE voxel correction?

Answer 43

Takes into account the number of independent comparisons

Question 44

What is cluster-size correction?

Answer 44

Takes into account the contiguity of active voxels

Question 45

What is FDR correction?

Answer 45

Controls for the false-discovery rate

Question 46

Problem with whole brain, voxel-by-voxel comparison?

Answer 46

Performing more than 50,000 statistical tests so risk of type 1 error!

Question 47

Advantages of Region-of-Interest analysis?

Answer 47

Avoid multiple comparison problem by restricting analysis to particular region, identified in advance. Greater statistical power.

Question 48

Disadvantage of Region-of--Interest analysis?

Answer 48

May miss potentially important regions outside of ROI. Danger of biased post-hoc selection of ROI

Question 49

When performing group-level analysis, what two ways can you combine data across subjects?

Answer 49

Fixed effects Mixed/random effects

Question 50

What does 'fixed effects' analysis entail?

Answer 50

- assumes the effect of the exp. condition is fixed (constant) across subjects - the data from all PS is combined (as if it came from the same PS) and the stat. tests are performed

Question 51

What does 'mixed/random effects' entail?

Answer 51

- Assume that the experimental condition could have a different effect on different PS - a stat. test is run on each PS (first level), followed by a (second-level) test to determine whether the first level analysis is consistent across PS

Question 52

Does modern research utilise 'fixed effects' or 'mixed/random effects'

Answer 52

Mixed/random effects

Question 53

When performing whole-brain analysis, different PSs' brains need to be aligned with each other in the same space. What is this called?

Answer 53

Coregistration or Normalisation.