I think the best way to explain this is to start with presence detection--where is there a train? The NS main line has automatic signaling, so we can stick to that for now. A "track circuit" consists of a low-voltage current in one of the two rails and returning in the other, being connected at both ends of a section by a wire passing through a current source at one end of the section and a relay at the other to detect that the circuit is complete. If a train or engine is in the section the current will be diverted--"shunted"--through the wheels and axles of the cars, preventing it from reaching the relay, resulting in a binary function; i.e., 1 = circuit complete, no train present, or 0 = circuit not complete, train--or something--present in that section. This constitutes a building block on which all information pertaining to the presence of trains is based.
Where automatic signaling is in effect, as here, signals fall into two categories: fully automatic, and controlled. A fully automatic signal automatically displays whatever indication is appropriate to track occupancy ahead, as determined by the track circuits. Controlled signals are nowadays electronically controlled by a dispatcher. Historically, there were "block operators" at towers along the line, but today all significant main lines are under the control of a dispatcher who directs movement over hundreds of miles of route--somebody on here knows where the dispatcher for that part of NS is sitting. When the dispatcher desires to line a route for an approaching train, he "clears" the signal at the entrance to the appropriate route--the switches line up automatically and the signal then displays the most favorable indication consistent with that route and the distance from any preceding train.
As to the second question, when a train passes a signal the front wheels and axle immediately shunt the circuit, automatically causing the signal to change to red.
To the third question, hand-thrown switches are normally wired in with the track circuits so that when a switch is unlocked by hand prior to throwing it the track circuit is shunted just as though a train were there. There are rules and special instructions in effect prescribing exactly how this is to be done to avoid conflicting with a movement which has already been set up.
On most railroads a fully automatic signal has a number plate on the mast, usually 3 to 5 numerals and sometimes a capital letter, while the absence of a number plate identifies a "controlled home signal" controlled by the dispatcher. Any yellow or green on an automatic signal simply indicates the conditions ahead, but on a controlled signal indicates that the dispatcher is expecting a train. (A totally red controlled signal, however, may also mean that a train is expected but the needed route is at that moment occupied by another movement.)
I think the signal rules in effect on that part of NS (although they may have changed following the split of Conrail) are those of the Northeast Operating Rules Advisory Committee (NORAC). One website that shows them is
" onclick="window.open(this.href);return false;
with the signal aspects and indications beginning on page 76. They can help visualize what's happening on site.