high cpu spikes on postgresql (rds)

I have a PG 15(.5) RDS instance, which is experiencing intermittent random(?) periods of high cpu; that do not seem to, afaict, correlate with an obvious cause: Here is one example. The queries being made before/during & after are the same; and if run with the same params once the cpu drops back down, are quick as before. I also don't see any increase in transactions: or tuples: during that period; or any large vacuum. The only graph that seems to line up, is this one: according to the pg docs, this is: > Number of times disk blocks were found already in the buffer cache, so that a read was not necessary so I'm not clear why that would be related. The instance is an db.r7g.8xlarge, with 1758 GiB of gp2; and isn't running particularly hot, as the graph shows. Is there something I can do to find out what the cpu is actually doing? Do I need to catch it in the act?