Most Swapped Choices Go Unnoticed, and People Explain Them Anyway

Imagine you're handed two photographs of faces. Which one do you find more attractive? You point to one. The researcher slides the cards across the table, face down, and then hands you back the one you picked.
Except it isn't the one you picked. A little card trick has swapped it for the face you turned down.
"Why did you choose this one?" the researcher asks.
What would you say? You'd probably be sure you'd say, "Hang on, that's not the one I chose." In a 2005 experiment published in Science, most of the time, people didn't: only 13 per cent of the swaps were caught on the spot.
The trick
The study had 120 participants. Each was shown 15 pairs of faces and asked to choose the more attractive one in each pair. On three of those pairs, the researchers used what they called a double-card ploy. Two cards were held together, so that after the choice, the photograph handed back was the face the person hadn't picked.
Then came the question. Why this face?
The researchers counted how often people noticed the switch while it was happening. That's what they called concurrent detection. Out of 354 swapped trials, people caught 46 at the time. That's 13 per cent.
They tried to make it easier. Some people got unlimited time to look at the faces. Some got pairs that looked less alike. Even in the most generous version, with unlimited time and dissimilar faces, no more than 27 per cent of swaps were caught on the spot. Rather surprisingly, how similar the two faces were made no significant difference to the detection rate.
And if you count every way a swap could come to light, including people who mentioned afterwards that something had seemed off, no more than 26 per cent of all the swapped trials, across every condition, were exposed. The researchers point out that even this is generous. Once someone spots one swap, they get suspicious and look harder at the next. Strip out the trials that came after a first detection and the overall rate is significantly lower.
The explanations
Here's where it gets really strange.
On the swapped trials, people were asked to explain their choice, and many of them did. They gave reasons, in the first person, as if they were simply reporting what had gone on in their own heads. Some even pointed to features that belonged only to the face they now held, the one they had turned down. The researchers gave those a category of their own: specific confabulation.
When independent raters scored the explanations for emotion, detail and confidence, the reasons given for swapped faces came out no different from the reasons for faces people had genuinely chosen. On those measures, the made-up reasons looked like the real ones.
That's the "hmmm" at the heart of this. We tend to think of our reasons as a recording of how we decided. These results suggest that, at least sometimes, the reasons get built afterwards, around whatever turns out to be in our hand.
Jam, tea and a supermarket
You might wonder whether faces are a special case. So the researchers took the method somewhere much more ordinary: a supermarket in Lund, in Sweden.
At a tasting stand, shoppers sampled two jams, or smelled two teas, and said which they preferred. The jars were double-ended, so they could be secretly flipped, and the taster was handed the jam or tea they hadn't chosen.
There were 180 shoppers. Only about 14 per cent of the swapped jam trials, and about 14 per cent of the swapped tea trials, were noticed at the time. Adding in people who mentioned it afterwards, and those who reported that the taste or smell seemed to have changed, brought the totals to about a third. So roughly two-thirds of the swaps went straight past.
These weren't subtle pairs, either. Even with flavours as different as cinnamon-apple and grapefruit, no more than a fifth of swaps were caught at the time.
Morals on a page
Then came the bigger question. Does this happen with things people actually care about?
In a 2012 study, people filled in a paper survey about moral questions. Some rated general moral principles. Others rated current moral issues. The survey was rigged so that two of their answers could be secretly reversed. Most of the reversals went undetected, and 69 per cent of the participants failed to spot at least one of the two changes. Many then argued for the opposite of the view they had just ticked.
A 2013 study tried something similar during a Swedish general election. No more than 22 per cent of the altered answers were caught. Ninety-two per cent of participants accepted the altered score they were shown. And around 48 per cent, give or take about 9 percentage points, said they'd be willing to consider moving between the left and right blocs. That's one sample, in one election, after an unusual intervention. It's a finding about that study, not a rule about voters everywhere.
What it doesn't mean
It would be easy to take all this the wrong way, so it's worth being careful.
It doesn't show that people have no preferences, or that their beliefs are empty. Plenty of people did catch the swaps. The figures also depend on what counts as "noticing". Spotting it at the time, admitting to it afterwards and sensing that a taste had changed are different things, and the studies count them separately for a reason.
And none of this is a recipe for tricking people. The studies debriefed participants afterwards, and the later ones report ethics-board approval. What they reveal is narrower, and more interesting. Our explanations for our own choices can come apart from the choices themselves, and we often don't notice when they do.
The story and the choice
It feels as if we have a pretty good line into our own minds. We pick something, and we know why we picked it. That certainty feels like direct access.
These experiments suggest the certainty can survive even when the thing in our hand is the one we turned down. We point, someone swaps, and our story steps in to fill the gap.
Sources
Petter Johansson, Lars Hall, Sverker Sikström and Andreas Olsson, "Failure to Detect Mismatches Between Intention and Outcome in a Simple Decision Task", Science 310 (2005): 116–119. https://doi.org/10.1126/science.1111709 Authors' copy: https://www.lucs.lu.se/fileadmin/user_upload/lucs/2011/01/Johansson-et-al.-2005-Failure-to-Detect-Mismatches-Between-Intention-and-Outcome-in-a-Simple-Decision-Task.pdf (accessed 14 Nov 2026)
Lars Hall, Petter Johansson, Betty Tärning, Sverker Sikström and Thérèse Deutgen, "Magic at the marketplace: Choice blindness for the taste of jam and the smell of tea", Cognition 117 (2010): 54–61. https://doi.org/10.1016/j.cognition.2010.06.010 Authors' copy: https://www.lucs.lu.se/fileadmin/user_upload/lucs/2011/01/Hall-et-al.-2010-Magic-at-the-Marketplace-Choice-Blindness-for-the-Taste-of-Jam-and-the-Smell-of-Tea.pdf (accessed 14 Nov 2026)
Lars Hall, Petter Johansson and Thomas Strandberg, "Lifting the Veil of Morality: Choice Blindness and Attitude Reversals on a Self-Transforming Survey", PLoS ONE 7 (2012): e45457. https://doi.org/10.1371/journal.pone.0045457 (accessed 14 Nov 2026)
Lars Hall, Thomas Strandberg, Philip Pärnamets, Andreas Lind, Betty Tärning and Petter Johansson, "How the Polls Can Be Both Spot On and Dead Wrong: Using Choice Blindness to Shift Political Attitudes and Voter Intentions", PLoS ONE 8 (2013): e60554. https://doi.org/10.1371/journal.pone.0060554 (accessed 14 Nov 2026)









