Running outside or Threadmill

Let's look again at your phrase 'pulled backwards'.  _Is_ the foot
'pulled backwards' in either situation?

  Run along the earth.  When you put your foot down, it stays exactly
where it was, in the earth-coordinate frame.  Any 'pulled backwards'
is an illusion from thinking of your center of mass as a stationary
reference point (thence giving the notion of the foot being 'pulled back').
  The work done by the earth against your foot is F*d -- but d is exactly zero, so the work is also exactly zero.  This is true regardless of the speed your foot is moving at touchdown relative to your center of mass -- once down, it stays put.
  Now run on a treadmill.  (Better, watch someone else run on a
treadmill.) Your foot lands squarely under your hip (center of mass).   When you lift it off the treadmill surface, however, it is somewhere
behind your hip (same relative position as on land).  So d is distinctly
_not_ zero.
  _If_ you're running with the foot speed equal to the treadmill speed, then the treadmill applies no force to your foot (no need to accelerate it to running speed, F = ma, but a is zero).  So, again, the work is zero.
  _But_ if your foot is moving too slowly, the treadmill accelerates
it to running speed.  F is not zero, now.  d is never zero on treadmill
running.  So, in this case, the treadmill does F*d work on your leg.  
This is work that you don't have to do yourself, so the treadmill is
saving you a bit.
  If you're running properly, the treadmill belt never accelerates
your foot, so does no work and we're back to the ideal situation of
having minimal difference between treadmill and road.  (Wind resistence, surface flexibility, etc. still differ, but can be compensated for and/or are small effects.)  But if you're running improperly, it is
indeed the case that the treadmill does work for you that you otherwise do on the road, making the road slower running.