When we get beyond pure logic and look at statistics or legal standards of proof beyond a reasonable doubt, we get into Type 1 and Type 2 errors.
Here's an example from politics. The inability to prove fraud to a legal standard does not mean the election was fair:
For an example from law, a verdict of not-guilty does not mean innocent. There are those found not guilty even though they are.
In such situations, the binary logic of Fig 1 is not an accurate model.