Specifying complex designs in events.tsv with multiple columns or overlapping events?

Hi all,
I’m wondering what would be considered more BIDS spec in the events.tsv for more complex designs. For instance, we have designs where we want to check univariate effects for classes of stimuli but then, separately, estimate individual trials for bringing forward to a different analysis. Or we also have state-item designs where we want to simultaneously model blocks and events.

Reading through past responses, it seems like one route is to specify everything in the standard “trial_type” column, though it gets messy as we double or triple the ways of describing individual trials. A cleaner option would be to have separate columns corresponding to the event specifications for different designs (all described appropriately in the json sidecar). However, the latter seems a bit less BIDS standard, or at least the spec doesn’t explicitly suggest it as an option. Anyone in the know have any thoughts as to which is more likely to be preferred down the road? (i.e., for data sharing, bids apps, etc.).