Regions that accumulate information across small time scales show large responses and vice versa.
Inter subject correlation-U. Hassan-Same stimulus-check the response across individuals. So one person's response acts as a 'design matrix' control for the others. Allows the usage of natural complex long-term stimuli.
At rest, responses may be different (no correlation) but while listening to the same stimulus the same areas work? So the response becomes aligned.
https://cfn.upenn.edu/stslab/wiki/lib/exe/fetch.php/fmri_club:preprocess1:smith_2004_brjrad.pdf
Understanding fMRI data.
looks like for a voxel or for a given area the activity is recorded across time. Peaks mean higher activity and lows to lower activity.
For calculating differences-first participant brain area (voxel-3D brain area) maintained the same-but response to the stimulus mean taken. Same for the other stimulus. Check for responses throughout the brain. So in the end you get a mean brain that shows different revels of responses in different regions of the story duration (averaged out) for a single stimulus. Same for the other.
Checked if there was any variation across time. Not so much
Checked for variation with synonyms. Initial regions-(auditory) showed larger responses as compared to higher order brain regions. And to make or break understanding higher order regions involved
So at some time there is less difference between the mean response to the different stimuli and some instance there is more-Caluclation of Euclidean distance
What is Euclidean distance though?
When two graphs are superpositioned on each other, the distance between two close high-points on the graph is the Euclidean distance. Lesser Euclidean distance, more similar the samples.
(18 participants)
———————————————————————————————————————————
Impact of context to the SAME story (40 participants)
Euclidean classifier set up
So 20 participants made to listen to a particular context of a story questions asked and resposes taken 19 peop;e formed a control set and the remaining person formd a test set and one person is the test sample.
Mean of set taken and compared to the test set. If similar then categorised automatically by the machine learning algorithm.