A while back I had game AI expert Dave Mark on my podcast. One of the things he talked about in the context of using psychology to shape player behavior was the Kuleshov effect. I wanted to look at this topic in detail.

The effect is named after a Soviet filmmaker of the the same name, who created mini-montages and showed that the content of one shot could change the viewer’s perception of another. Imagine you have four small clips of film. One shows a man staring blankly into the camera. Another shows a bowl of soup. A third shows a sexy lady. The last shows a dead child. Actually, you don’t have to imagine. Here are clips similar to what Kuleshov used, all strung together:

Now, take three groups of people. Show one group just the shot of the soup followed by the shot of the man staring into the camera. Show another group just the sexy lady followed by the shot of the man. Finally, show the third group just dead child followed by the man.

Now ask everyone to describe what the man in the clip is thinking.

What researchers have found1 is that people’s perceptions of the man in the shot are greatly influenced by the context created by the other clip. Supposedly, those who saw the soup were more likely to say that the man looked hungry. Those who saw the woman reported a lascivious look in the man’s eye. And those who saw the dead child reported seeing that the man’s expression bent to sorrow. When faced with an ambiguous situation or stimulus we immediately lean on contextual information for guidance on what information to attend to, how to interpret it, and even what we remember. In my mind this also relates to what’s called the fundamental attribution error, which describes how we tend to blame internal states for the behaviors of others, as opposed to external factors. For example, “That person is crossing her arms because she’s unfriendly; we’re crossing our arms because we’re cold.” When contextual information supports this bias, it’s even more likely to be used to make sense of an ambiguous situation.

As in film, the Kuleshov effect and the fundamental attribution error can be leveraged by game designers to add a little depth to their world. And the best part is that it can often be the players themselves who are creating the backstories, justifications, and narratives. If you’re playing an open world game and you come across two NPCs having a chat in town, that’s just background decoration. But find the same NPCs with the same animations and chatter while exploring the woods outside of town? What’s up with that? They up to something nefarious? Scandalous? The answer could emerge out of the player’s situation –whether she found them there at night, while there are enemies about, or while on a mission to hunt animals. It could be different every time.

We also see this come up a lot in games like those in the Dark Souls series. Why are these creatures here at this particular location, in possession of these particular items? Little details create bigger contexts, which impel players to make up their own minds. And sometimes that’s even more fun.

And sometimes it’s hilarious. I recently played Batman: Arkham Knight and noticed the Kuleshov effect in action. At several points when Batman has a conversation with an NP, he takes on an expression much like the man in Kuleshov’s clips above. The Bats is just dead eyed, expressionless, and stares into the middle distance while the firefighter he just rescued earnestly thanks him for saving the day. This happens over and over, and I soon started believing my own narrative, that Batman was just a hair’s breadth from flipping out. He was seconds away from buckling under all the pressure, all the Scarecrow toxin, and the general, overwhelming hell that is his life. He was just ready to snap and break that firefighter in half at any second. This was aided by the context of Arkham Knight’s storyline, and that situation with Batman and the firefighter suddenly got a lot more tense.

All because of a blank stare.