- Lecture 2 - BLOCK & EVENT related design
- BLOCK design
- All fMRI experiments normally use block designs which is where participants do alternate tasks over a long period of time.
- 1. Often group lots of trials, which means that they take an average rather that looking at individual.
- 2. Highly predictable occurrence of stimuli, participants know what is coming so will change behaviour or focus = readings skewed.
- 3. Inflexible = can’t do complex tasks (event that occurs involuntarily, you cannot change it e.g. blinking).
- 4. Ecological validity = blocks of trials may change the psychological process that person goes through whilst doing the task, won’t measure activity, hard to compare to base line?
- 5. You can’t separate trials depending on their performance as you are looking at an average, you don’t know any individual trials (gets marked as either right or wrong, no other factors taken into account).
- EVENT design
Lots of different conditions, lot of different
participants, all mixed
together to complete the task at same time.
- You can use the info you collect to predict how high the BOLD signal should be at each point in time, when the trials are close together.
- 1. You don’t need to have a BLOCK design, you can do more complex and more novel experiments.
- 2. Flexibility and randomisation: eliminate predictability, no practice effects.
- 3. Post-hoc sorting: after the experiment you can look at activity for correct vs. incorrect, remembered vs. forgotten, fast vs. slow reaction time.
- 4. Novelty: rare or unpredictable events can be measured.
- 5. Look at temporal dynamics of response.
- Two ways to pick up signal from everything that we are seeing:
- 1. Lots of trials with robust manipulations, can make an average, see if what you’re looking at is right
- 2. Cleaning data (pre-processing): attempts to remove rather than model data. fMRI data takes 2-3 seconds to collect a single volume = a strip of the brain. To reference each point in the brain, we divide the image into cubes, which are called voxels, each voxel is given a coordinate (x, y, z).
- Lots of different conditions, lot of different participants, all mixed together to complete the task at same time.
- Steps to pre-processing:
- 1. High pass filtering: we remove the low frequency data that introduce noise
- 2. Motion correction, put stages together = no fuzz.
- 3. Slice time correction: add all your slices to form one image.
- 4. Co-registration: put pixelated images into one viewable image
- 5. Normalisation: compare to other trials, can do this using the Montreal Neurological Institute, as they have 352 MRI scans of normal right-handed participants.
- 6. Spatial smoothing: multiply voxel sizes by 2, to see clearer.
- BLOCK design