After years of anthropological work in criminal circumstances, I disagree that memory is *easily overridden" by circumstances. I doubt we'd survive as a species if true.
But, if there are external threats and pressures, yes, the memory does change (and in domestic situations, that's rather common). This is not one of those situations, IMO, so far as we know.
Certain details of memory are not as malleable - I could write a dissertation on that topic. People rarely forget, for example, whether an assailant is a man, woman, boy or girl. People may forget/fudge some things - but a sensitive interviewer finds that out rapidly.
We can't interview missing people, of course.
But I disagree that eyewitnesses are one of the "most unreliable forms" of evidence - and in fact, I also assert that the law agrees with me. Eyewitness evidence is pretty much the only thing that's not circumstantial. Eyewitnesses are revered throughout history. Sure, they can miss on some small detail - but perhaps you can list examples where eyewitnesses have been very wrong?
Or any other citation to back up this rather extraordinary statement. Just look through one or two day's worth of WS threads - the eyewitnesses are rarely wrong. Erika Lloyd's car was found vandalized and abandoned. Her camp was found deserted. By eyewitnesses. We could go on - but truly, I'd like to see the research on which you base this assertion, because without eyewitnesses, most criminal cases are dead in the water. So they're valuable, IMO.