Christine,
Both methods of yours are overly conservative. The first three steps in method #2 should be good enough:
1. Voxel-wise thresholding for statistical test #1, followed by cluster-wise thresholding
2. Voxel-wise thresholding for statistical test #2, followed by cluster-wise thresholding
3. Identifying spatial overlap among above-threshold voxels for tests 1 and 2.
In other words, as long as each individual cluster is corrected for multiple testing, you don't need any further correction after the conjunction analysis. The rationale is discussed in the following paper:
TE Nichols, M Brett, J Andersson, T Wager and J-B Poline. Valid conjunction inference
with the minimum statistic. NeuroImage, 25:653–660, 2005.
Gang