[braythwayt]

  > I found a cat that does not like milk.
  >
  > Therefore, if I see a cat, I can draw no inference
  > about the probability that they like milk.

This is not correct. It is correct to say that having found a cat that does not like milk, we know that cats do not necessarily like milk. That is valuable and relevant.

But nevertheless, we can still have some confidence that, absent other indicators, cats like milk.

p.s. Here’s another example closer to home of a “cat that did not like milk:” Roy Cohn, who carried out anti-homosexual witch hunts, and later died of AIDS.

Edit: Adding link to https://en.wikipedia.org/wiki/Roy_Cohn