In the late 70s and 80s, the concept of CRM expanded enormously in response to several well publicized accidents. It remains an important theme in the culture of aviation safety to this day. In many ways, the insidious nature of confirmation bias relates to CRM. But aviation needs to now shine an intense spotlight on confirmation bias and figure out what to do about it.
Another piece of the JFK puzzle I have not heard brought up...how many clues and cues did the crew of AA 106 ignore on their taxi before crossing 4L? How many takeoffs on 4L did they have a chance to see pass before them before the DL flight? Assuming that 4R was the main arrival runway, how many landings did they have a chance to see? Was there a "string of pearls " in the sky to their right perpendicular to their taxi route? Or was the crew thoroughly busy with their checklists. Did they notice that while taxiing parallel to 31L for a mile or more, there wasn't a single takeoff on that runway?
Obviously this is not a easy problem to solve... there's simply not enough frequency time for everything to be be verified ad infinitum.