Of course, if you also assigned a probability of zero to the event you just observed, now you have 0⁄0 error, which is more awkward to deal with. The case of having a posterior probability of zero in contradiction to the evidence is not particularly problematic for the agent’s thinking, it just isn’t very useful. But a true 0⁄0 event might well cause serious issues.
Of course, if you also assigned a probability of zero to the event you just observed, now you have 0⁄0 error, which is more awkward to deal with. The case of having a posterior probability of zero in contradiction to the evidence is not particularly problematic for the agent’s thinking, it just isn’t very useful. But a true 0⁄0 event might well cause serious issues.
In practice, you conclude you hallucinated the event.