This is a really interesting discussion - thanks everyone for your thoughtful replies.
It’s clear that the effect (real or not) is maximised when the braking effect is maintained right up to the moment that the vehicle comes to a stop, and that it can be mitigated by progressively easing up on the brakes. By the same token, it would be most noticeable in an emergency braking situation, where passengers would be thrown forwards at the sudden onset of maximum deceleration, and (apparently) thrown backwards when the deceleration suddenly stops.
Conductors are the guys walking around inside the train, and have no control over braking. (This mistaken usage has become extremely widespread in the US and I’m not sure why.)
The train’s driver (“engineer” in American English) operates the brakes.
I wondered about “conductor” as well but thought it must be a local Japanese usage. In New Zealand the conductor is called a train manager. The driver is called a locomotive engineer, often shortened to “LE” in the business, but “driver” is also used casually.
I think it’s slightly more complicated than that. As I understand it, the locomotive’s brakes can be applied or released separately from the wagon’s, helping with “stretch braking” where tension is maintained on the couplings during braking.
See Post 8, they train them to do that. Local trains also need to stop on the stop lines, but they are on a much tighter schedule so you see them clamp the brakes occasionally. I have also seem the overshoot the line and need to back up. That’s embarrassing for the driver.
Or sometimes, (around here) doesn’t backup, and doesn’t open the doors. I’m sure that the rules here have changed over the years: back when the passengers controlled the doors, they would back up, or else risk passengers jumping out.
I’ve never seen a passenger train here settle back after coming to a stop.
It’s “engineer” or “operator,” but never “conductor” in Canadian English. A friend of mine proudly called himself an operator of Canadian Pacific trains, and he would tolerate being called an engineer (“Engineers design bridges, but okay, fine”), but he would always point out that train conductors, in Canada anyway, had an entirely different job, and most could not operate an engine at all.
Here the Wellington metro train doors are controlled by the passengers but only after the Train Manager has released the appropriate doors. I think the drivers can also override doors to keep them shut. So the train stops at the appropriate mark for the length of train, the train manager releases the appropriate doors (not the ones that open onto the adjacent track!) and only then may the passengers open the doors themselves. If the driver misses the mark by a little bit, the train manager can prevent passengers from opening doors that don’t open to the platform.
If the train does need to set back then it just needs verbal authorisation from the Train Controller. The Train Controller is in charge of all movements on their part of the rail network, they do signalling, verbal authorisations for setting back, arrange crossing of trains in single track areas, that sort of thing.
In the 20 plus years of riding Tokyo area trains, it only happened a few times. I wonder if they have to get special authorization as well. They really punish drivers who make mistakes and that punishment and punitive retraining system lead to a really deadly train accident after the driver needed to back up.
This accident occurred while I was in Japan and they had a special on the punishment system for the drivers, and how that humiliation was seen as a likely cause of the accident. From the same linked article.
They have that culture of never being late, but this was the unfortunate outcome.
The effect you are noticing is the transition between dynamic and static friction (the little jerk). The locomotive rolling back a couple of inches is the delay in releasing the brakes and the elasticity of the train (after the whole train has stopped, the brakes are released (last car last) causing the locomotive to be “pulled back” by the last car: The whole train is stretched out a little bit during braking and goes back to normal when all brakes are released with the brakes on the last car the latest to release. If the brakes all release at the same time the locomotive only goes back 1/2 the distance (depending on the center of gravity of the train).
In post 8 you were speaking of deceleration, i.e… the time rate-of-change of velocity. The detail I was referring to (what Shinkansen operators are good at minimizing) is jerk, i.e. the time rate-of-change of acceleration. You can comfortably subject passengers to pretty high deceleration, provided you ramp up/down to/from that level in a gradual fashion. Conversely, even mild deceleration can create unpleasantly high instantaneous jerk values if that decel isn’t gradually tapered to zero as velocity also approaches zero.
As kids we used to watch the old Ripcord TV series. For may years afterwards my friend refused to believe that skydivers didn’t go back upwards a bit after the 'chute inflates.
The illusion, of course, being created because the cameraman continued to fall after the skydivers parachute deployed and slowed them.