Philadelphia’s Got Talent! A novel, naturalistic paradigm for inducing socio-affective prediction errors

Human connection relies on a constant, invisible stream of guesses. In every conversation or interaction, our brains build a quick mental model of what will happen next, predicting how others will react to our words and actions. When reality matches these predictions, everything flows smoothly. But when the outcome clashes with what we expected, a gap opens up. Scientists call this a prediction error. While we often think of these errors in terms of learning facts or navigating physical spaces, they are just as powerful in our social lives. When a social outcome violates our expectations, it can trigger a sharp, self-conscious feeling. This is the territory researchers have long wanted to map: the moment our internal social map fails, and the emotional jolt that follows.

A team of researchers at Temple University and the University of Oregon has now built a new way to study this phenomenon, creating a naturalistic experiment designed to trigger these social mismatches on purpose. They call their method "Philadelphia's Got Talent." The goal was not to find the next big star, but to create a controlled environment where participants would inevitably stumble, record that stumble, and then watch it back, allowing the scientists to measure the exact moment their expectations clashed with reality. By observing how people react to their own awkwardness versus the awkwardness of others, the researchers hoped to understand the mechanics of embarrassment and how our brains update their social models when things go wrong.

The experiment unfolded in two parts, separated by a week. In the first session, participants were told they were taking part in a study about creativity. To keep them from suspecting the true nature of the task, they were asked to perform a series of one-minute creative acts in front of a camera. They had to tell a story, sing a song, or dance. Some were given strict instructions, like singing a specific nursery rhyme, while others were given more freedom to create their own version. The researchers recorded everything. The participants then rated how well they thought they did and guessed how much a partner would like their performance.

One week later, the participants returned for the second session, which was a surprise. They were told they would be watching videos, but they were not told whose videos they would see. As they sat in front of a screen, they were shown thirty-second clips of their own performances from the previous week, as well as clips of a partner's performance. To heighten the sense of social scrutiny, an experimenter sat nearby, watching silently with a neutral expression. As the videos played, participants used a slider to continuously rate whether the performance was better or worse than they had expected at that specific moment. After each clip, they answered questions about how they felt and how they thought others would judge the performance.

The results revealed a consistent and surprising pattern in how people view themselves compared to how others view them. When participants guessed how their partner would rate their performance, they were consistently too pessimistic. They expected their partner to like them less than the partner actually did. This gap between expectation and reality, known as a "liking gap," appeared whether the participants were guessing immediately after performing or after watching the video a week later. In fact, the data showed that people systematically underestimated how positively others would evaluate them. This suggests that our internal social models are biased toward negativity; we assume we are being judged more harshly than we actually are.

However, the reaction to watching oneself was different from the reaction to watching someone else. When participants viewed their own performances, they reported feeling more negative emotions, specifically embarrassment. They felt a distinct sense of discomfort seeing their own awkward movements and voices. In contrast, when they watched their partner's performance, they reported feeling amusement. They found the other person's mistakes funny rather than painful. This split in emotional response highlights a crucial difference in how the brain processes self versus other. The researchers found that the feeling of embarrassment was tied directly to the violation of social norms; it was the specific act of seeing oneself break a rule of social conduct that triggered the negative feeling, not just the act of watching a video.

The study also looked at whether these feelings changed over time. As the participants watched more and more clips, the intensity of the mismatch began to fade. The gap between what they expected and what actually happened narrowed, and the sharp sting of embarrassment lessened. This suggests that the brain is learning. With repeated exposure to the same social situation, the internal model updates, and the surprise wears off. The participants were essentially getting used to their own awkwardness, and their predictions were becoming more accurate.

Interestingly, the researchers tested whether the presence of a real person watching would change the outcome. In the first version of the study, an experimenter sat right next to the participant. In the second version, the participant was alone, told only that someone was monitoring them from another room. The results showed that the core findings held up in both scenarios, though the specific feeling of being judged by a person in the room might have made the initial reaction stronger. The study also ruled out the idea that simply watching a video of oneself is enough to cause these effects. When participants watched neutral clips, such as footage of themselves filling out a questionnaire, the strong gap between expectation and reality disappeared. The effect was specific to the moments of performance and social evaluation.

The researchers found that the type of performance mattered, too. In the second study, they added a new category where participants had to rap nursery rhymes. They discovered that dancing generally received more positive ratings and feelings than the other forms of performance, while the rapping often received less favorable reactions. This indicates that not all social violations are created equal; some forms of expression carry more social risk than others.

Ultimately, this work provides a clear window into how we navigate the social world. It shows that our brains are constantly running a simulation of how we are perceived, and that simulation is often wrong. We tend to think we are being judged more harshly than we are, and when we see evidence to the contrary, it takes time for our minds to adjust. The study confirms that embarrassment is not just a vague feeling of shame, but a specific signal that our social predictions have failed. By measuring this signal, scientists can now study how people learn from social mistakes, how they recover from them, and why some people seem to feel the sting of embarrassment more deeply than others. The "Philadelphia's Got Talent" paradigm offers a new tool to explore these questions, moving beyond simple surveys to capture the real-time dynamics of social error and correction.