Raw Data
Your question is really good. In my view there are very great similarities - as some of the recent contributors have said - and great differences.
The reason I reacted sensitively to MOR was not because of ego, but because I am sensitive to the dangers in the belief that there is nothing to learn from people who are not in the aviation industry, and, even worse, that there is nothing sensible to be said about flying planes by people who don't fly planes. (I got my CPL and IR not because I was a pilot wannabe, but so that I could at least part way get past that). It is really pleasing to me to see that this thread is showing that there are more and more people who realise that these things are not so.
Let me give an example. I am currently working for a European national rail system where the challenge is to get more trains running on existing tracks without speed increases. The answer they are developing is a combination of GPS, computerised train control, and Eurofighter adapted HUD's in the cabs. This reverses 50 years of shifting towards the dream of 'automatic' driverless trains, back to placing most of the judgements with the driver.
In other words in this case increasingly sophisticated technologies are requiring INCREASES in driver skills and responsibilities. The railway company in this case is learning from aviation technology, but I believe that the human factors and competency challenges they are struggling with have a lot to offer to aviation safety.
In response to your main question - the things that cannot be transported across settings are such things as the inability to stop an aeroplane immediately you have a problem, the fact that when you stop a train or a truck there is always the risk of something going up your backside, and the slow response time of ships coupled with their inability to fly above the weather.
Where they are most similar is that basic human physiology and psychology is the same whatever sector you are in, that costs and operational pressures are always in tension with optimally safe practice, and that the emotional and the cultural always have startling ways of creating unintended consequences when new technolgies are introduced.
The trick in cross-sector learning is to apply the high level learnings and translate them into the specific industry context.