In this particular case I don't see how this is not causation. Unless they made a huge blunder in the setup, like showing sexy pictures with some cars but not others. The sound of the cars is a constant. It's not influenced by the subjects nor by external factors. So correlation in this case pretty much indicates causation.

I'm referring to the increase in testosterone being used to indicate an increase in arousal. Testosterone can be increased for other reasons.