If you watch many movies you will see a disproportionate number of conversations set in elevators and cars. Of course people converse in such settings sometimes in real life, but why is it so common in the movies?
Both cinematically and in terms of acting, the ordinary way people talk to each other — face to face — often isn’t ideal for film. Audiences don’t want to look at the back of an actor’s head, so the director usually cuts back and forth between the two actors or shoots the two actors in profile. The former approach doesn’t let the audience see the reactions of both characters at the same time. The latter gives the audience more of this, but if the characters are far apart it can be distracting for the audience to switch attention back and forth as in a tennis match, and even worse when such conversations are shown on television with pan and scan rather than keeping the original aspect ratio, sometimes the back of each character’s head gets cut off to fit everything in the same shot.
Directors try to work around this by having both actors face the camera, particularly during critical conversations. Otto Preminger and his frequent cinematographer Joseph LaShelle did many such shots. Here is one from Where the Sidewalk Ends. This can work to a point, but risks take the audience out of a scene if they start to think “If someone was facing my back and proposing marriage/confessing to murder/revealing the secret of King Tut’s tomb, wouldn’t I, you know, turn around and look at them?”. In the generally good film noir The Big Combo for example, Joseph Lewis staged so many of the film’s key conversations with both actors looking at the camera that the artificiality worked against the grittiness of the story.
The easiest way to resolve this problem is to shoot dialogue scenes in the few places where human beings have conversations while they face the same direction in real life: when they are driving, seat belted in and looking at the road, and, when they are in an elevator looking at the door/floor indicator. The tight physical location means that it’s easy for the viewer to take in each participant and all their actorly non-verbals.