OK, so let's invert the question - if the entire system is reliant on drivers obeying signals then why did we ever need tokens on single lines?
The original purpose of most signalling equipment from a system safety point of view was to try to reduce the possibility of a signaller making a fatal error that may result in deaths or injures. Although some changes have been made to also try to mitigate errors made by drivers (ATC, AWS, ATP, TPWS, DSD, DRA, ETCS/ERTMS and so on) most of these are relatively recent technologies.
The purpose of a single line token is to try to prevent the possibility of signallers letting two trains on to the same section of (single) line. Especially two trains proceeding towards one another at speed.
As MarkyT says, the multiple track circuits, the cables and relay and electrical interlocking equipment for/on long sections were expensive. A single train staff in comparison was cheap and effective, one train staff means only one train per section. A driver, when they have possession of the train staff knows that they, and only they, have authorisation to enter the single line section. And as there was only one train staff, only one signaller could hand it to a driver. It was now not possible for a head on collision to occur if the rules were followed.
However, a single train staff does put limits on operational flexibility. With some methods of working on a single line, this was relaxed slightly to allow one train to follow another in the same direction. The driver of the "first" train was only shown the token or train staff. The driver of the following train was given the train staff, but only permitted to proceed after either conformation that the first train had left the section (arrived at the next block post). Or a suitable time interval had elapsed. This system however, still allowed simple human errors or the break down of trains to result in (rear end) collisions.
The token systems with instruments that contained them and electric locks to only allow one token to be extracted at a time, but which now allowed unequal flows of traffic through a single line section both greatly increased system safety and allowed more operational flexibility.
Token systems only needed one pair of telegraph wires. A token instrument at each end. Maybe an intermediate instrument if there was a ground frame (G.F.) on the single line between block posts (signal boxes). And sets of batteries to power the token instruments. The length of the section does not affect the cost (within reason). Even twenty to twenty-five years ago, token systems for lines where speed and capacity were not important factors, were seen as cost effective.
And the system was considered foolproof until the collision at Abermule in 1921. After that, further improvements were made.
The token instrument system is effectively the equivalent to the absolute block instrument system used on pairs of unidirectional lines.