 |
 |
Dear Darren On 11th January, you asked >What is the initial threshold for each of the contrasts in the conjunction? > - I assume this is what the user chooses as the height and spatial >extent thresholds? >How close (in terms of significance) do the activations in each contrast >(as part of a conjunction) have to be for a particular voxel to show up in >a conjunction? My understand is that conjunctions show 1) voxels that are >significant based on the chosen thresholds in each contrast and 2) those >voxels whose significance doesn't differ much from one another. >For example, if a voxel had a z score of 3.5 in one contrast and 4.8 in >another, would that show up in a conjunction whose initial threshold were >(height- 3, and extent- 3 voxels)? What about 3.5 and 6? > The conjunction analysis is the sum of all the contrasts you enter where the voxels with significant interactions between these contrasts have been removed. In effect you are looking at the height and spatial extent of the overall main effect where there are no significant interactions. However, if you want to be stricter in order to identify those voxels where there is a significant effect in each of the contrasts independently, then you need to use masking. In this case, you could chose any threshold for the contrasts from 0.1 upwards. So in direct answer to your questions: >What is the initial threshold for each of the contrasts in the conjunction? There is no initial threshold for each contrasts in the conjunction, only when you are masking. >How close (in terms of significance) do the activations in each contrast >(as part of a conjunction) have to be for a particular voxel to show up in >a conjunction? My understand is that conjunctions show 1) voxels that are >significant based on the chosen thresholds in each contrast and 2) those >voxels whose significance doesn't differ much from one another. Exactly %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%