Pragmatic effects can be hard to pin down in reading time measures. I’ll talk about a garden-path-like effect resulting from eager causal inferencing in English discourses, which has been found once in eye-tracking (Hoek et al. 2021), but which I’ve failed to find in self-paced reading or Maze latencies (Duff 2023). Drawing on Clifton (2013), I’ll show new results using an alternative method, using math questions between SPR passages and comprehension questions to require long-term storage of discourse meaning, and thereby encourage rich first-pass representations. This turns out to do the trick! I’m hoping for thoughts and feedback on some ways of understanding why *this* method in particular should be helpful for encouraging incremental pragmatic enrichment in reading experiments.