“Catnip Immunity and Alternatives § Optimal Catnip Alternative Selection: Solving the MDP”, Gwern2015-11-07 (, , , , )⁠:

Estimation of catnip immunity rates by country with meta-analysis and surveys, and discussion of catnip alternatives.

Bol et al 2017 tested 4 cat stimulants on a large (n > 100) sample of cats, yielding within-individual correlations of stimulant responses. This allows prediction of stimulant response conditional on observing other stimulant responses. I use this to demonstrate how to optimize stimulant selection as a sequential testing problem, yielding (for one set of realistic purchase costs) a test sequence of catnip → honeysuckle → silvervine → Valerian, which is different from the greedy policy of picking stimulants by prior probability.