Putting this together, it seems there must be a small calculable amount of IM coming back to earth, though most of it escapes...which is the impression I had before I became aware of these details. Loosely speaking then, a small amount of IM is "reflected" by CO2.
Here is what happens: 1) A CO2 molecule absorbs a photon of light in the infrared (IR) region of the spectrum. 2) A CO2 molecule can vibrate in a number of different directions (or modes.) The absorbed IR photon causes the molecule to jump to a more energetic vibrational mode 3) The excited vibrational state is "unstable," because there is a lower state (the one the CO2 molecule was in until a moment ago) so after a brief period of time, the molecule re-emits the photon. 3a) IF during this brief period of time, the CO2 molecule collides with another atom or molecule, the energy can be transmitted duing the collision, instead of re-emitting as a photon.
If (3) happens, you have "reflection." It is what happens "most" of the time, and there is no heat transferred to the atmosphere; it just goes back out into space. If (3a) happens, the colliding molecules pick up speed, and there is heat.
Here is a decent explanation of what would be covered in a freshman general chemistry class. http://www.wag.caltech.edu/home/jang/genchem/infrared.htm