Here, Holmes hits upon several threads that form some of the backbone of social psychology: we tend to behave quite differently when we expect to be observed than when we don’t and we are acutely responsive to prevailing social mores and social norms.
When we decide to do something, should it matter to us whether or not someone else is watching? While theoretically, it’s easy to argue that it shouldn’t, that the same behavioral norms apply no matter what, in practice, it usually does. This goes for minor behaviors (Will you pick your nose in public? What about if you’re pretty sure no one is watching you?) as well as much more important ones (Will you hurt someone, be it physically or otherwise, if others are observing your interaction? What about if you’re fairly certain the misdeed will never go beyond the two of you?).
Many studies have shown that the answers to those questions differ significantly. For instance, people are much more likely to cheat (on an exam, on taxes, on each other) if they don’t think they’ll be caught. They are more likely to steal in a corporate environment that fosters a feeling of anonymity than in a more personal setting. And on the flipside, individuals tend to be much more generous and altruistic when someone else is looking. In game settings, divisions become much more fair, cooperation more common than any theory of simple value-optimizing behavior would predict. In the real world, philanthropic donations rise and helping and law-abiding behaviors increase.