So if I see a bird in the woods and it flys away then I have to disbelieve that I saw it because I can't repeat it... that's retarded.
If you saw a crow in the woods around here contextual information would suggest a high probability of you being right.
Accepting said crow would not cause any upturning of current understanding of the natural history of the area; ie, it makes little difference if I erroneously accept your statement.
I would, without further effort, give it a pretty high level of acceptance.
If, however, you saw an Ivory billed Woodpecker in Arkansas you would get only a moderately low level of acceptance without further support. The context of information that we have is far different and the consequences are vastly different (there is a large reward for finding one and proving it).
Assigning different levels of acceptance is not retarded. It is exactly what we do everyday. It is the only rational behavior.