Prompted by how she kept mixing up similar-looking characters among the cast of hundreds, Christel, a lecturer in the University’s School of Psychology, joined with fellow researchers Annabelle Wride and Associate Professor Gina Grimshaw to develop a test based on the show to provide fresh insights into human face recognition.
“We had noticed significant limitations in existing tests,” says Christel. “Some, for instance, use strictly controlled faces of strangers—stripped of hair, glasses, and other adornments—that people study and learn in the lab. To us, this seemed very unlike the way we learn faces in the real world, where we encounter people in many different situations and without any real intent to study their facial features.”
According to laboratory-based tests, Christel is supposedly good with faces, but she wasn’t when it came to Game of Thrones and its more challenging conditions.
So she and her colleagues tested participants who had watched the first six seasons of the show, just once and when they first screened, thus ensuring the same viewing experience. They showed them 90 headshots of actors with different levels of exposure (main heroes, lead characters, support characters, and bit players), mixed with 90 strangers.
Participants judged whether each face was familiar and rated their confidence in their answer, before trying to identify and name the character or actor. Half were shown pictures where actors looked similar to their character and the rest pictures where appearance differed.
Among the cases of mistaken identity that ensued were Jack Gleeson, who plays despised Joffrey Baratheon, taken for Maisie Williams, who plays beloved Arya Stark; Sibel Kekilli, who plays Tyrion Lannister’s mistress Shae, taken for Isaac Hempstead Wright, who is supernatural seer Brandon Stark; and a grizzled bearded stranger taken for Kristian Naim, the hulking Hodor.“These and other misidentifications were often based on ostensibly superficial features like hairstyle or colour and facial hair,” says Christel.
Only the most prominent actors were correctly identified and named more often than just recognised, and identification errors occurred at all exposure levels. Bit players were never correctly identified and only rarely recognised, even if their brief appearance in the show had a dramatic impact.
“The participant who recognised the most actors did so for about 80 percent of actors, but also falsely recognised about 50 percent of strangers. This shows that being good with faces is also about knowing when you encounter a face for the first time. Some participants were extremely confident and others really hesitant, but confidence didn’t predict performance.
“Our research—showing that even people with superior recognition abilities also forget faces and make identification errors—has important implications for eyewitness testimony in criminal court cases and law enforcement agencies’ use of super-recognisers.”
It can also make the rest of us feel better about our own difficulties with Game of Thrones characters.