It's like long-term health. You can't determine whether cigarettes are good or bad for your longevity by any self-experiment. You can only look at historical data on others.
Likewise, deontologists look at other people's lives to decide which virtues seem worth adopting. It's very time-consuming, because people don't reliably self-report their own virtues. The only way to know if someone is loyal or not is by observing, over a big chunk of their life, how they treat people around them. You have to know a lot of people quite well for a long time to make reliable observations on how virtues affect people's lives.
Since you have to adopt virtues fairly young to get the benefit of them, you don't have time for all this observation. The best you can do is to read the books of wise elders and adopt the virtues they agree on.
Discovering new facts that throw new light on the circumstances of old actions doesn't mean e.g. an evil action became good, but that it was revealed as having been such all along.
On practice, nobody is completely either one of those.
How deontologists learn to change their mind (adapt to change) if they prevent themselves from experiment to see the consequences?