Basically when you work with audio through speakers you have a point where the audio is 100%. If you go above that, you don't get higher volume, but instead you get distortion. So instead of working with absolute volume, you have to work with relative volume. The volume of the engine, for instance, can't be constant throughout the entire time, or there won't be enough space left for the other sounds. So when driving alone in low speeds, the engine sounds would be the dominant, perhaps occupying 70% of total volume available. Now, if another car catches up with you, and that engine sound would have to occupy 50% of total volume, then you can't have 120% volume because that would be distortion. Instead you would have to reduce the volume of both sounds to fit withint the 100% range, only keeping the relation between them. So your engine sound would then go from 70% of total volume to 7/12th of total volume (about 57%), while the other car's engine sound would go from 50% to 5/12th, or about 43% volume.
Another trick that's common when you work with sudden, loud noise, is to, instead of inserting an incredibly loud BANG! (that wouldn't be louder than a bird's tweet, scince all you've got to work with is those 100%) you cut the volume drastically, almost muting, to give an impression of a loud noise that blends all other sounds out.
It doesn't always work great, and sometimes the balance is not right, but that's the theory anyway.