Narrative as data compression

The ability of our working memory to hold different pieces of information is limited. We can usually remember something like six random digits, but not much beyond that. Some people manage to remember much more but they usually do it by using some kind of memory trick. Rather than remembering six-three-six-five-six-seven, they remember sixty three - sixty five - sixty seven. They compress the data from six separate pieces to three. I believe we do this with almost all data that we have. We combine it into narratives.

E. M. Forester pointed out the difference between "The king died and then the queen died.", where we just have a sequence of events, and "The king died and then the queen died of grief.", where we have have given the sequence of events a casual connection and thus have a plot. We add these casual connections to our data to gather them together into a narrative to compress them. For it is much easier to remember the narrative than to remember the separate pieces of information.

The problem arises because this data compression isn't lossless. Things that then don't fit our narrative explanation can get left out. This gives us the confirmation bias. We can also go in the other direction and start adding other information to the narrative just because the fit, even though there is nothing connecting them to the events in question. Do this enough and you get conspiracy theories.