You would think that being frozen out by people you cannot stand is the one kind of rejection that slides off.
Psychologists thought so too, and they had reasons better than a hunch. Social bonds are built through group membership. The need to belong is served by the people you count as your own, not by strangers you would cross a road to avoid. And we tend to move away from outgroups anyway, without being asked twice. Put all of that together and being shut out by a group you want nothing to do with should cost you approximately nothing.
Two experiments have now gone looking for that discount. Neither has found it.
The reasons it should have been true
Some time in the mid-2000s, Karen Gonsalkorale and Kipling Williams sat volunteers down at computers to play a simple ball-tossing game with two strangers. What the volunteers had been told about the strangers was the variable. Some believed they were playing with people who shared their politics. Some believed they were playing with supporters of the main rival party. And some believed they were playing with people who leaned toward the views of the Australian Ku Klux Klan.
Then the ball stopped coming.
Williams later summarised the result in a review of the field: “Despite strong reasons to discount ostracism by an outgroup or, especially, a despised outgroup, the distress of ostracized participants was unmoderated by the psychological closeness of the ostracizing group.” Being frozen out by the Klan registered about the same as being frozen out by your own.
The explanation Williams offers is a two-stage account he had laid out years earlier and restated in the Annual Review of Psychology. First comes what he calls a reflexive painful response to any form of ostracism, which he describes as unmitigated by situational or individual difference factors. Only afterwards does a reflective stage arrive, in which people appraise the situation, work out who did it and why, and start shoring up whichever need took the hardest hit. On that account, the identity of the people ignoring you is information that arrives late. The pain does not wait for it.
All of which rested, on the despised-outgroup question, on one study reporting one absence of an effect.
Someone ran it again with a Front National flag
In 2014, four researchers did exactly that. Writing in Social Psychology, Marie-Pierre Fayant, Dominique Muller, Chris Hartgerink and Anthony Lantian recruited sixty French undergraduates, reproduced the Australian procedure closely, and swapped in French politics. Days before the session, students picked a party they would affiliate with and rated all of them. The Parti Socialiste served as the ingroup. The Front National served as the group nobody wanted anything to do with: asked whether “this group disgusts me”, students gave the Front National 4.29 and their own party 1.77.
The game is called Cyberball, and it is deliberately trivial. Players are told they are throwing a virtual ball with two other people elsewhere in the building; the two other people are the computer. Players being excluded got a couple of tosses at the start and then nothing at all, while players being included got roughly a third of the throws. Afterwards everyone answered the same twelve questions, three each on four things exclusion is thought to threaten: belonging, self-esteem, a sense of control, and the feeling that your existence registers on anyone. “I felt rejected.” “I felt powerful.” “I felt invisible.”
Nine people were dropped before analysis, for housekeeping reasons: one worked out the hypothesis, three misreported which party their co-players belonged to, four turned out to lean right themselves, and one was a statistical outlier. Fifty-one were left.
Exclusion landed hard, well over a point down on the combined needs measure. Whose flag was on the screen made no significant difference to that, and what difference there was ran backwards from the intergroup prediction.
Then the team did something better than repeat the original. They doubted the ruler. You can agree that you felt invisible, they pointed out, simply because you were in fact being ignored, without any of it having hurt. So they built a second measure that asked directly. After the standard questionnaire, each student was shown their own answers back — “to the question ‘During the game, I felt rejected’ you answered 5” — and asked how that had felt, on a scale running from really hurtful at one end to really pleasant at the other.
It came out the same way. Excluded players landed on the hurtful side of the midpoint and included players on the pleasant side, and the despised flag did not soften it. The new measure has a crack in it, which the authors name: because students were shown their earlier answers before rating them, they may have been reaching for consistency, and the two sets of responses did correlate. Within this design there is no way to separate those.
Pooling their study with the original in a small meta-analysis, the authors found no reliable interaction on the combined measure or on any of the four needs taken separately. One result cut against the grain and they flagged it. On the need to belong specifically, the flag did matter, at conventional significance, but in the opposite direction, with exclusion by the Front National biting harder than exclusion by their own party. They decline to build anything on it, since it was not predicted and the meta-analysis does not confirm it. It also disappears if the nine dropped participants are put back in.
The trouble with an absence
A failure to find something is not a finding that there is nothing there. The pooled confidence interval on the flag effect runs from well below zero to well above it, a range comfortable with no discount at all and with a fairly substantial one. Nobody has established that being iced out by people you loathe is exactly as bad. What has been established, twice, is that nobody has managed to show it is better.
The samples do not help. Fifty-one French undergraduates, playing in groups of up to eight for the chance of winning an mp3 player, is thin. The authors say as much in a footnote, conceding the sample “might seem too low to accept a possible null effect” and pointing to the meta-analysis as the corrective. The meta-analysis contains two studies.
And there is a serious argument that the whole question has been posed badly. In 2016, Selma Rudert and Rainer Greifeneder ran four studies on whether the first sting depends on the rules of the situation. In one, they rewrote Cyberball as Dislikeball, telling players to throw only to the person they liked least. Being left out of that game hurt measurably less than being left out of the ordinary version, even though everyone understood perfectly well that they were being excluded. Where the norm was on the excluders’ side, the flinch was smaller. “Based on our current findings,” they write, “we are confident to say that norm-consistent exclusion causes less reflexive pain reactions than its norm-violating counterpart.”
Their reading of the Klan result is the interesting part, and it is not that it was wrong. Cyberball’s unspoken rule is that everyone gets a fair share of throws, so refusing to pass to someone breaks the rule no matter who is doing the refusing. On that view, as they put it, “the finding that reflexive reactions are unalterable may not so much reflect a true state of the world, but a specific choice of experimental situations in which exclusion violates an implicit inclusion norm.” They are careful to add that this does not diminish the earlier results, since exclusion and rule-breaking travel together often enough. Their own effects attenuate the hurt rather than removing it, one of their four studies was underpowered, and another asked people to imagine a debate rather than sit through one.
No clinician has been anywhere near this article, and none was anywhere near the studies either. Cyberball measures a flinch in the first minute, in a room, with a stranger’s flag on a screen. That is the whole of its reach. The real version of this is being cut out at work or at home, month after month. Ask a GP or a counsellor about that one.
Fayant and colleagues name the experiment they want next, and it is a good one. A reflexive flinch should not need spare mental capacity. A correction should. So run the whole thing again with people’s attention loaded up elsewhere, and watch whether the discount for being snubbed by a party you despise fails to arrive. If contempt only protects you when you have the spare capacity to remember you feel it, is it protecting you at all?
