Selective Perception: Why Fans of Both Teams Saw a Different Game

Perception is not a recording. What reaches awareness has already been sorted by attention, which goes to some things and not others, and shaped by interpretation, which decides what an ambiguous event means. Both steps are guided by what the perceiver already believes and wants. A fan who expects the other team to play dirty looks for contact by the other team, finds it, and reads it as a foul; the same contact by their own team is a hustle play, or is not noticed at all. Nothing dishonest has happened. Each fan reports what they saw, and they saw different games.

Hastorf and Cantril put the point sharply in 1954: there is no such thing as a game out there that people merely observe. The game exists for a person only as the events they pick out and give meaning to, and those events are selected from a huge field of possible ones. Of the hundreds of collisions in a football game, a spectator notices a few, and which few depends on which team they came to see win. Which side was in fact rougher, the authors wrote, is not a question the study answers.

The same selection runs on expectation as well as loyalty. In Vallone, Ross and Lepper's 1985 study, pro-Israeli and pro-Arab students watched the same network news coverage of the 1982 Beirut massacre, and each group rated it as biased against their own side. The authors suggested that partisans on each side had noticed and remembered more of the content that went against them. A report that strikes both camps as hostile is a sign that the camps are perceiving selectively, not evidence about the report.

Selective perception is cousin to confirmation bias but sits earlier in the process: it is about which parts of a scene reach a person and what those parts seem to be, before any evidence is weighed. The demo above cannot show the attention half of that, since every play is described in full. What it can show is the interpretation half: the same words, read as a foul or as a fair play depending on which side the player wears.

On November 23, 1951, Dartmouth played Princeton at Princeton's Palmer Stadium. Princeton was unbeaten and its star, Dick Kazmaier, was playing his last game; he left in the second quarter with a broken nose, and a Dartmouth player later left with a broken leg. The game was rough and the penalties heavy, and in the following week the student newspapers of the two colleges accused each other's teams of dirty play. Albert Hastorf of Dartmouth and Hadley Cantril of Princeton saw a natural experiment and published "They saw a game: A case study" in the Journal of Abnormal and Social Psychology in 1954.

The study had two parts. A week after the game, a questionnaire went to 163 Dartmouth and 161 Princeton undergraduates in psychology courses. Asked which team started the rough play, 86 percent of the Princeton students said Dartmouth; among Dartmouth students, 36 percent said Dartmouth and 53 percent said both sides started it. Asked whether the game had been clean and fair or rough and dirty, no Princeton student chose clean and fair and 93 percent chose rough and dirty; Dartmouth students split, with 39 percent writing in an answer of their own, rough and fair.

In the second part, 48 Dartmouth students and 49 Princeton students watched the same film of the game and checked every infraction they saw as it happened, marking each as mild or flagrant. Princeton students saw the Dartmouth team commit 9.8 infractions on average, more than twice the 4.2 they saw their own team commit, and they rated the Dartmouth infractions flagrant about twice as often as mild. Dartmouth students saw 4.3 infractions by Dartmouth and 4.4 by Princeton, about the same number for each side. The two groups had watched the same reel and counted different games.

The paper did not set out to decide who had actually played rough, and its authors said so. Its point was that the game each group reported was a product of what that group brought to it. Vallone, Ross and Lepper's 1985 study of the hostile media phenomenon, with 68 pro-Israeli, 27 pro-Arab and 49 neutral students, is the best known follow-up.

Selective perception operates wherever people with different stakes look at the same scene. The rows below are illustrations, not findings from the studies cited on this page.

Perceiving selectively is not a choice, so the counter-measures change what gets recorded and who records it.

• Decide the criteria before the event. A list of what would count as a foul, a success or a failure, written down in advance, is much harder to bend to a loyalty than a judgment made on the spot.
• Keep a record that does not depend on memory. A transcript, a video, a tally sheet or a log can be checked by someone with the opposite stake; a recollection cannot.
• Get the opposite stake to look. If two readers with opposing loyalties both find a report fair, that is stronger evidence than one reader's confidence.
• Separate counting from judging. Count every instance of the behavior first, from both sides, and only then decide which ones mattered. Hastorf and Cantril's students did the two together and the counts followed the loyalties.
• Treat your own certainty as a symptom. The feeling of having plainly seen what happened is exactly what both sets of students reported.

Selective perception is the front end of a family of biases that keep beliefs in place, and the family members are easily run together.

• It is not confirmation bias, which is about seeking and weighting evidence to support a belief already held. Lord, Ross and Lepper showed the weighing step in 1979, when supporters and opponents of the death penalty each rated the study that favored their side as the better piece of research. Selective perception comes first: it decides which parts of a scene register and what they appear to be, before any weighing is done.
• It is not naive realism, the conviction that you see the world as it is and that anyone who disagrees must be biased or misinformed. That conviction is what makes selective perception hard to notice from the inside, but it is a belief about perception rather than the selection itself.
• It is not attentional bias, the narrower finding that attention is drawn to stimuli linked to a current concern, such as threat words for an anxious person. Selective perception covers interpretation and memory as well as attention, and it is driven by expectations and loyalties rather than by emotional state alone.