Yesterday, I responded to someone in a Facebook group regarding the reliability of memory and its implications on historical apologetics.
From my reading of the literature I've gathered it is misleading to lump all kinds of memory into one category. What I mean by this is that there are different kinds of memory and there are certain conditions for when memory is very reliable. For example, unique and/or important events are most likely to be remembered alongside the gist (with some variation). Frequent rehearsal of a particular event also strengthens the stability of memory. I also would like to point out a couple things about laboratory studies. 1.) There is a deliberate focus on the failure of memory in these studies which means that failure is more interesting but not necessarily common. 2.) the environment of these experiments do not capture the natural environment in which actual remembering occurs. For example, the subjects chosen for these experiments often don't know each and/or have no social connection to the material being remembered. 3.) These experiments usually utilize single-chain models of transmission as well as testing semantic memory when actual oral tradition and remembering usually take the form of communal memory with multiple chains (aka net transmission). One last thing I would point out is that oral tradition is specifically designed to counter fragilities in memory by utilizing various cues such as rhythm, meaning, imagery and sound (a favorite song for example can be repeated hundreds of times). I'd recommend the work of Richard Bauckham, Craig Keener, Alan Kirk, and David C. Rubin on this topic.