In the 1920s, a Russian journalist was being told off by his editor for not taking notes on the assignments he was given. The journalist, SV Shereshevski, repeated back the assignment word-for-word. Unaware that such memory feats were unusual, Shereshevski was found to have a perfect memory by a neurologist called AR Luria.
Luria found that S could remember any length of word lists, even 16 years after originally learning them.
Richard Cytowic's book "The Man Who Tasted Shapes" documents a condition called Synesthesia, a bit like the concept in NLP, except that the condition is involuntary, and the correlating synesthetic representations are consistently the same eg. the taste of mint always feels like glass columns.
Now, compare human and computer storage and retrieval. Computers are obviously a lot better at it than humans and yet they still use redundancy. Memory and data transmission have parity bits. Many programs utilise CRC.
This is so that in the event of potentially incorrect retrieval, the computer has two or more ways of checking.
Human memory works pretty much the same way. We know that a single representation is not that reliable, so we make simultaneous ones, usually in different sensory systems.
So you might know you ate chicken yesterday, not just because you cooked/ordered it, but because you saw/smelt and tasted it as well. You may even remember digesting it.
The same with spelling. I know how to spell extraordinary, not just because I know what it's supposed to look like, but I often over-emphasize the second syllable: exTRAordinary.
Whilst S was supposed to have a perfect memory, Luria found that he naturally used a strategy whereby things he experienced would induce simultaneous vivid, distinctive and consistent associations.
Further proof of the myth of the Photographic Memory.
"Rather, we found that superior memorisers used a spatial learning strategy, engaging brain regions such as the hippocampus which are critical for memory and for spatial memory in particular."