I'm confused about our application of the definition of "support".

We said A supports B iff A increases the likelihood of B. But in probability theory, this relation is always symmetric: if A supports B, then B supports A (by the same ratio, according to Bayes' theorem).

In the tiger example, the conclusion does actually support the premise: to see this, note that if all mammals were suitable pets, it would be pretty unlikely for tigers to be maiming humans.

So I'm wondering how to reconcile this with the directional arrows between premise and conclusion, the different words used, etc.

I'm confused about our application of the definition of "support".

We said A supports B iff A increases the likelihood of B. But in probability theory, this relation is always symmetric: if A supports B, then B supports A (by the same ratio, according to Bayes' theorem).

In the tiger example, the conclusion does actually support the premise: to see this, note that if all mammals were suitable pets, it would be pretty unlikely for tigers to be maiming humans.

So I'm wondering how to reconcile this with the directional arrows between premise and conclusion, the different words used, etc.