It is truly because you are reading the sentence and not just scanning for F’s. An F subvocalized as a V is ignored.
Try looking at individual letters, starting at the end of the sentence, and you will pick up all F’s.
The first time I saw this, it was handed to me on a little card. I turned it upside down so I was less likely to read it and counted the Fs. Got it right the first time.
I think it’s something like this. The little words don’t really register in people’s minds when scanning for a spelling mistake. It is as if they were an atomic glyph in and of themselves.
I’ll bet you this trick wouldn’t work in lowercase, and I think it may even be a trick of upper case. The three that people get are all FI, and the three that they miss are all OF.
I agree that it is because “OF” is perceived as an individual entity. The word is so short and common that the brain can’t help but take it in as a whole.
It would be interesting to repeat the experiment with people who had no knowledge of English. I’d wager they would spot all the Fs.
The trouble with that experiment is that it wouldn’t distinguish between the F’s pronounced as V theory and the “of” is thought of as a unit itself theory.
I’m thinking it’s because people just skim the sentence. Show this to a small child that is still learning to read or doesn’t have a lot of experience reading and I bet they get them all because they’re concentrating harder on actually reading the sentence.