I don't know if any of the following will work or not, but maybe worth an experiment.
Try a delay of about 1/2 of a millisecond, if you have the capability, between the two channels, of the mono signal. The goal is for the delay to match the amount of time it takes a sound to travel from one ear to the other. I can't remember the exact number--something like 25 or 32 frames at 44100 fps, if I remember correctly. Not sure how effective that is with speakers, but I'm able to get some very real panning effects that way, even on my tiny laptop speakers. Delays that are greater than that amount are not going to be as effective (as a panning device).
Close to the ear, there is going to be both very little roll-off of the highs and a really strong bass response. Low frequency components can travel around the head easily, but the highs will be damped slightly by the head, so I'd try rolling off the highs on the lagging signal a slight amount, but leave the lows strong.
Also, I'd make the whisper totally dry: no reverb.
Another thought, strategic contrasts to other sounds can strengthen an impression. So can you have immediately preceding sounds or context be slightly more mid-rangy, or more distant (using reverb) sneakily setting that up as a "norm" for the ears before using the whisper eq? When the contrasting whisper sound happens, the contrasting qualities should be more prominent as a result.