Skip to main content

Imaginary Presidents and Imaginary Gods: The Real "Empty Chair" Effect

This article was published in Scientific American’s former blog network and reflects the views of the author, not necessarily those of Scientific American


If you were to have told me just last week that one of my psychology experiments would soon be brought to life on stage by none other than Clint Eastwood at the Republican National Convention, all to the fêted laughter and applause of tens of millions of people who, in the true spirit of literalism for which so many of them are accustomed to thinking, would fail to see the irony of their own rapturous enjoyment of the scene before them, I’d have thought you were insane. But there it was, the old bleary-eyed star having a feverish, bizarre conversation with an empty chair beside him, a chair in which throngs of delighted viewers—viewers who happen to know a thing or two about having feverish, bizarre conversations with imaginary authority figures—were playfully led to believe sat the invisible president of our country.

The connection may not be immediately apparent to those uninitiated into my research area of the cognitive science of religion, but take my hand and allow me to walk you through this theoretical briar patch.

In an article titled “Princess Alice is Watching You: Children’s Belief in an Invisible Person Inhibits Cheating,” published last year in the Journal of Experimental Child Psychology, I, along with my former PhD students, Jared Piazza and Gordon Ingram, used the invisible-person-in-chair scenario to test a simple but important hypothesis: When no other actual person is around, and when we’re tempted to do something we know we shouldn’t, the illusion of a supernatural watcher should meaningfully influence our behavioral decision-making. The key to our experiment was creating a fun but competitive game in which, on the one hand, children were explicitly told that cheating was not allowed. On the other hand, however, they would find cheating very appealing if they thought they could get away with it. In other words, we created a laboratory condition that, at least in a very general sense, reflected the temptations that both children and adults face every day.


On supporting science journalism

If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.


There was also an empty chair in the room, facing the child. I’ll get to that soon.

We tested 68 kids between the ages of 5 to 9 for this particular study. The basic rules of the game—a target board using Velcro balls—were straightforward enough for even the youngest children. After being told that the goal was to get the bull’s-eye and that lesser points were awarded for throws that connected to the board further away from the center, each child was given three additional rules. First, they had to throw the balls with their back to the target (looking over one’s shoulder was permitted, however). Second, they had to throw the balls using only their non-dominant hand (a wristband was placed on their right or left wrist as a reminder of which hand they could use). Finally, they had to stand at a distance of approximately 6 feet from the target. This line was marked clearly with tape on the floor and children were instructed not to cross it.

In essence, then, we created a nearly impossible task. But that didn’t dilute the children's efforts. Only after each child displayed six successive throws without breaking any of these rules (indicating their comprehension) did we proceed. Each child was also told that they were competing against other kids, and that whoever got the highest score would win a special prize. A bit cruel, yes, but we had to create an incentive—and all children in the end got a decent consolation prize.

These foregoing rules, the basic setup, were the same for every child. But here’s the trick: each child had also been randomly assigned to one of three different conditions. For those who’d been assigned to the “no supervision” condition, the experimenter feigned having to do something and left the child alone in the room to play the game on his or her own, saying that the points would be added up when he returned. For those who’d been randomly assigned to the “adult supervision” condition, by contrast, the experimenter again left the room for the child to play on his or her own, but a second adult, friendly enough but neutral, sat in the chair and watched the child play the game without interacting. Finally, those children who’d been randomly assigned to the “invisible agent” condition were introduced to “Princess Alice.” They were told that Princess Alice was a “friendly magical princess … that could make herself invisible … and is in the room with us right now and is sitting in that chair.” Before exiting the room, the experimenter reminded this group of children that they wouldn’t be alone, but rather Princess Alice was there too and would keep them company. So in other words, this last condition was physically the exact same as the “no supervision” condition (alone in the room with an empty chair), but it included a psychological manipulation: children in this condition were told about a supernatural presence occupying the chair.

In reality, the experimenter simply went to an adjacent observation room. There, together with the child’s (usually giggling) mother or father, he recorded the child’s behavior through a hidden camera for a period of three minutes before returning.

Although we examined a host of other factors as well (including age, gender, and individual differences in children’s temperament), the most important question for us, of course, was the extent to which children would be more or less likely to cheat as the result of their being randomly assigned to one of these three different room conditions. Remember, we stacked the deck methodologically so that it was possible for any given child to cheat in one of three ways: facing the target when throwing the ball, using their dominant hand, or crossing the line to get a closer range. Any of these violations, let alone all three, would put someone at an unfair advantage over those who obediently followed the rules of the game. Every child knew this going in.

So what did we find? Just as we expected, basically, and probably what you’d have as well. Those children who were randomly assigned to the “no supervision” condition were the most likely of the bunch to cheat when the experimenter exited the room. Nearly half of these kids, in fact, were so brazen in their transgressions that they simply walked right up to the wall and manually placed the ball on the target (usually just shy of the bulls-eye to cleverly simulate some relative degree of marginal error), thereby breaking all three rules at once! Those in the “invisible agent” condition, by contrast, were just as well-behaved when they thought Princess Alice was in the room as were those kids being watched by an actual, flesh and blood person sitting in the chair before them and supervising their behavior.

But there’s an important caveat, too. This rather astonishing Princess Alice effect only panned out statistically for those children who said that they believed that she was real. The more sceptical children in the “invisible agent” condition, by contrast, were just as likely to cheat when left alone as those in the “no supervision” condition. Yet even those who adamantly denied that Princess Alice was real during their initial introduction to her, when left alone in the room, seemed to display some curious signs of ambivalence about her. In fact, for those kids in the “invisible agent” condition that did cheat, the majority only did so after “disconfirming” her non-existence by running their hand across the chair. Some even "Eastwooded" her by speaking to her.

We’re all susceptible to tales of the supernatural, in other words, but genuine belief matters. For those who truly believed in this laboratory concocted spiritual entity, she was real enough in their minds, anyway, to affect their behavior in an empirically demonstrable way. As I’ve been arguing for the past several years now, from an evolutionary perspective, it doesn’t make any difference whatsoever if the belief itself is true or patently delusional. Natural selection isn’t a mind reader and isn’t concerned about ontological reality; rather, evolution operates on the organism’s behavioral output, and if that behavioral output consistently leads to reproductive success, then the mental processes responsible for it are going to undergo selective pressure. If our ancestors thought that they were alone and/or could get away with something, but in fact were underestimating other people’s finding out, then the illusion of a concerned “invisible agent” would have helped them to inhibit selfish, impulsive decisions that could have seriously compromised their reputations, and hence their genetic interests.

Sure, Princess Alice probably wasn’t haunting the African savannas in which our ancient relatives were having their neural systems pruned by evolutionary forces tens of thousands of years ago, but there were almost certainly other fabricated creatures just like her (e.g., the spirits of dead loved ones, supernatural deities). And incidentally, we found the same general effect in a separate study with adult participants several years earlier (Study 3). College students who were told that a ghost had recently been spotted in an empty chair in the corner of a room were less likely to cheat than other subjects at a competitive task when left alone.

Just as we implanted the thought of Princess Alice in the minds of young children in our lab, so too has the idea of every single god or spirit been transmitted from adult to child in the past. The difference, of course, is that whereas we took great care to debrief the children in our study by telling them that Princess Alice was only make-believe and part of our silly little experiment, children in the past were never debriefed about the fictitious gods that they grew up with. The original fabulists died off without telling them that it was all made up. So when these children became adults and communicated the very same stories to their own children, they did so with all the potent conviction of true believers. (For more on this, see psychologist Paul Harris’s excellent new book, Trusting What You’re Told.) Fast forward eons and gloss over infinite complexity and conceptual nuance, but the result is all the same: Today the earth brims with the descendants of those children who were never disabused of such false information.

The children in our study who believed ardently that Princess Alice sat watching them throw Velcro balls against a wall bring us back full circle to Clint Eastwood’s performance at the GOP convention last Thursday. The actor stood before an audience buzzing with religious brains just as equally convinced that Jesus knows (and cares) what they do in secret and will reward them, not with stickers and stuffed animals, but with an eternity in heaven for following his rules. And they laughed hysterically at a man having a conversation with an invisible person sitting in a chair.

The fact that Clint Eastwood’s empty wooden chair at the RNC held the missing Commander-in-Chief is poetical to me in another way as well. Much like the God of biblical lore, but without omniscient supernatural abilities, the President represents the ultimate “Big Brother” in social regulatory terms, with a panoptical view into our private lives. Here’s what I wrote in The Belief Instinct:

The philosopher Voltaire famously said, “If God did not exist, it would be necessary to invent him.” That was sound logic at the time. But remember, Voltaire wrote this in 1768 during the French Enlightenment. Things have changed since then, to say the least. With today’s social-tracking technology (Social Security numbers, the Internet, hidden cameras, caller ID, fingerprints, voice recognition software, “lie detectors,” facial expression, DNA and handwriting analysis, to name just a few particularly effective behavior-regulating devices presently in place in the modern world), Voltaire’s declaration doesn’t really pertain anymore—at least, not for large-scale, developed nations. Who needs Voltaire’s “eye in the sky” when today we’ve got millions of virtual superhuman eyes trained on us from every possible angle, lodged discreetly in every pore of our lives? Human [brain] evolution hasn’t quite caught up with human technology, however, and the adaptive illusion of God is likely to survive so long as … our species’ cognitive blueprint [remains].

In any event, RNC shenanigans aside, I do take some comfort in knowing that another famous actor, Morgan Freeman, the very voice of God, in fact, has a lucid understanding of what goes on inside of human minds when contemplating invisible people in chairs. Click below to watch a re-enactment of the Princess Alice study, from the latest episode of Through the Wormhole with Morgan Freeman.