best practices for calculating TENT function


In calculating the “n” to use for the TENT function, I am wondering if current best practice is to have number of TENTS to be TR-locked? I have a fast event-related design with a TR of 1.3 seconds and a stimulus duration of 1.5 seconds (with jittered ITI). I’m wondering if TENT (0,14,12) is appropriate to use or if there is an alternative? I came up with 12 based on the equation provided here: h

thank you in advance!

It is difficult to define “best practice” without knowing the ultimate research goal. Usually I would prefer to make the basis functions (knots) synchronize with TR to avoid artificial interpolation with something like (using your situation as an example) TENTzero(0,14.3,12). On the other hand, if you have strong reason to adopt TENTzero(0,14,12), that can be justified too.

Thanks, Gang. This is helpful - I will try both approaches. I should have clarified that our ultimate research goal is to be able to use this task to derive meaningful bilateral hippocampal activation on a single-subject level (i.e., clinical fMRI mapping) using a contrast of novel items vs. repeated items (in a memory encoding task). Please let me know if this changes your initial recommendation! Thanks

By “ultimate research goal” I meant the subsequent modeling steps. For example, how are you planning to perform further analysis with those estimated tent coefficients?