Figure 3. Pain-stimulus associated brain activations where the contrast pain > smell was significant (conjunction of second-level t-contrasts 1 −1 0 and 1 0 −1 denoting CO₂, H₂S and vanillin stimuli associated responses respectively).

Statistically significantly activated voxels (p<0.05 FWE-corrected; t>5.14) are presented reflecting a 12-subject group analysis. Al the bottom line, the significance level was increased to FWE p<0.01. Effects of different masking of the pain > smell activations with different activations associated with pain (CO₂) or smell stimuli (the respective masking is indicated at the left side of the glass brains) at various FWE corrected significance levels of the mask, and the effect of increasing the FWE corrected significance level of the pain > smell activation to p<0.01. At the top of the figure, the pain > smell contrast is unmasked. In the glass brains following below, this contrast is masked inclusively or exclusively with regions significantly activated following CO₂ or H₂S stimuli. As the regions showing activations associated with H₂S stimuli completely covered those showing activations associated with vanillin stimuli, the first were taken as activations associated with smell. As expected, the activations pain > smell were within activations following CO₂ stimulation (inclusive masked) and not outside them and therefore disappeared when exclusively masked with the CO₂ associated activations. Unexpectedly, the activations pain > smell were also within activations following olfactory (H₂S) stimulation (inclusively masked) and disappeared when exclusively masked with H₂S associated activations, i.e., they were not outside the regions significantly activated by olfactory stimuli.