In the output file output_contrast+tlrc, the sub-brick s(TENT):pain contains the \chi^2(2) values for the interaction between pain and stimulus. Meanwhile, the sub-brick s(TENT) provides the \chi^2(2) values for the stimulus contrast between taste and tasteless, with both groups combined.
For reporting results, I recommend a "highlight, but don't hide" approach, as described in this blog post. If subcortical structures emerge in the findings, you could treat them as providing some level of statistical evidence.
To visualize the group-level HRFs, you can extract the relevant sub-bricks (point estimates and standard errors) from the second half of output_contrast+tlrc and view them using the Graph button on the AFNI GUI. Figure 6 in Chen et al. (2023) was created using R for a specific voxel. If you’d like to illustrate the HRF for an ROI, you can average HRFs across voxels and plot them in software like R.
You may also add covariates to your model, but ensure their inclusion is justified. For guidance, refer to this blog post on covariate selection.
Gang Chen