Mr Tullamarine and Genghis have provided a great deal of info. But I must be missing something here....
Academically speaking, if an aircraft is flying level at 250 kias and is trimmed, then there is no force required on the stick to keep it level. With slight changes to the flight path due to a variable air mass, the pilot can gently push and/or pull on the stick to maintain the altitude. The pilot can fly the aircraft without a problem.
Now the thrust is reduced slightly and the aircraft decelerates to 240 kias. Now there is a pull force required by the pilot to maintain straight and level flight. This pull force will likely vary slightly due to a variable air mass, but likely to remain a pull force (vice a push). Again, the pilot can fly the aircraft without a problem, although there is a trim system to help him which he would normally use.
Well taking your last paragraph first - if thrust is reduced slightly, any effects of change in speed are not a function of change of thrust as such - if thrust acts through the vertical CofG then there will be no change in speed. In fact, in most airliners, the thrustline lies below the CG, so reducing thrust will cause a slight nose-down pitching moment, and thus a small increase in speed.
However, for the sake of argument, you've changed thrust in an aircraft with a low thrustline and wish to adjust flightpath, either to maintain speed (pull), or to maintain level flight (more pull). Imagine that there is no stickforce gradient, then how can you correctly judge and hold the pull to maintain speed, or a bit slower to maintain level flight, and not to pull too much and bring the aircraft back to a thoroughly unwanted 150 knots? You can do it, but only by a high workload task with constant reference to instruments - whereas with a stick force you can feel where that is, maintain it, then trim out in good time. Also it's that stickforce / LSS gradient which stops the aircraft constantly drifting off the speed you've trimmed the aircraft to - otherwise speed will start drifting around whenever the pilot is not actively monitoring and controlling it.
G