Someone tells you they thought about it for a long time before they landed where they landed. Whether that sentence helps them depends largely on whether you agree with where they landed.

That is the finding in a paper by Zachary Grey, Paul Bloom, Jason Plaks and Yoel Inbar, published open access in Personality and Social Psychology Bulletin on 3 September 2026. Across three studies with 896 American adults recruited through Prolific, the researchers described a stranger taking a position on a contested political issue, and varied whether participants were told the stranger had deliberated before answering. Then they asked how good a person the stranger was.

Participants who disagreed with the stranger rated them higher when the stranger had deliberated. Among participants who agreed, the difference never reached significance in any of the three studies.

A pause carries two readings at once

The second study measured why. Alongside the character rating, 400 participants rated how carefully the stranger had considered the question, and how much hesitation, conflict and reservation the stranger seemed to have about the decision. The second set of items was reverse-scored, and the paper calls the resulting measure certainty.

Both readings moved, and they moved in opposite directions. A stranger described as deliberating was seen as considerably more thoughtful, and thoughtfulness tracked strongly with higher character ratings. The same stranger was also seen as considerably less certain.

The second inference is the one that does the work here, because its value is not fixed. Whether looking unsure makes you seem like a better or a worse person depends on what the observer thinks of your conclusion.

The reading that helps an opponent

For a participant who disagreed with the stranger, looking unsure about the position was not a failing; it was a concession. Both inferences now pointed the same way, and the effect on character judgments was large enough to see. In the second study, strangers who disagreed with participants scored 3.90 out of 7 when described as answering quickly and immediately, against 4.53 when described as deliberating. In the first study the pair came in at 3.71 and 4.06.

The certainty route reached conventional significance only in the third study, where the indirect effect through lower certainty came in at p = .005. In the second study the same path sat at p = .051, and the paper reports it as directionally positive rather than reliable.

The third study also added a condition the first two lacked. Every participant read about a stranger described as answering “after deliberating”, one described as answering “without deliberating”, and a third who simply answered, with nothing said about how. Among participants who disagreed, that silent third stranger was rated no differently from the visible non-deliberator: 3.69 against 3.75, a difference at p = .382. The one who deliberated came in at 4.00 and beat both. On this side of the design, an unexplained answer read like an unconsidered one.

Every participant saw all three versions across the twelve scenarios, so nobody was judging a deliberator in isolation.

That third study is also where the central claim gets thinner. Grey and colleagues had preregistered the interaction between agreement and deliberation as the key test, and it came back at p = .09. What held were the pairwise contrasts inside the disagreement condition, which were also preregistered and corrected for multiple comparisons, but a pattern of simple effects is weaker evidence for an asymmetry than the interaction test itself. The same model did find an overall effect of deliberation on character ratings, at p = .001, averaged across both sides of the design. With the interaction non-significant, the third study cannot establish that the effect belongs to one side rather than the other.

In every study the average rating for a disagreeing stranger stayed well below the average for an agreeing one, and in none of the three did deliberation lift the disagreement condition into the agreement range. Across the three studies the benefit among people who disagreed ran between a fifth and just over half a standard deviation.

The same pause goes flat among allies

Run the identical setup past someone who already holds the position and the two inferences stop cooperating.

Thoughtfulness still counts in the stranger’s favor. But a person who lands where you landed and looks hesitant about it now reads as insufficiently committed to something you consider obviously correct, so the doubt costs them. In the second study that penalty was real and small, at an indirect effect of −0.15, and it ran against a thoughtfulness bonus more than three times its size, at 0.50. The paper’s reading is that the two worked against each other: the total effect of deliberating on this side came out at p = .201, and in the third study at p = .153.

The averages are correspondingly flat. Agreeing strangers scored 5.14 without deliberation and 5.24 with it in the second study, 5.29 and 5.31 in the first. In the third study the stranger who said nothing about their process actually scored highest of the three, at 5.44 against 5.39 and 5.31, and none of those gaps survived correction for multiple comparisons.

Allies do not punish a deliberator. Among them the two readings simply pull against each other, and nothing survives into the character rating.

Limits the authors put on their own claim

Participants read a paragraph and rated the person in it. A single phrase carried the whole manipulation: “after long and careful deliberation” in the first two studies, “after deliberating” in the third. The silence itself was never staged. Grey and colleagues say directly that they were not trying to estimate what deliberating is worth in a real conversation. Real exchanges carry tone, nonverbal behavior and a social setting, and the paper allows that those cues could tip the reading toward uncertainty rather than thoughtfulness. Separately, it says real-world effects could be either smaller or larger than the ones it measured.

All three main studies showed each participant several versions of the same setup in quick succession, which is not how a disagreement arrives; a between-participants version reported in the supplementary material largely replicates the character result, but there the certainty pathway on the disagreement side flattens to near zero. The data and analysis code are posted, so the reanalysis is open to anyone who wants it. Every participant and every issue was American, and the researchers say the findings are untested in political cultures with different partisan structures. And certainty was never separated from perceived extremity, which was not measured directly, so a stranger read as less certain may simply have been read as holding a milder view.

In the first study, adding participants’ ratings of the stranger’s intelligence as a statistical control removed the interaction’s significance. The authors argue that intelligence is a poor thing to control for here, since a stranger who looks thoughtful will also look intelligent and intelligence is itself relevant to character. They call for designs that could establish whether intelligence sits upstream of the thoughtfulness pathway, downstream of it, or alongside it.

Moral character was the only outcome the studies were built to explain. Whether the deliberating stranger seemed warmer, more conscientious, or more worth collaborating with was never asked.

The 2013 wallet experiment

The counter-evidence is older than the paper. Work by Critcher and colleagues in 2013, one of whose authors is also an author here, found that deliberation makes a person look worse when the right answer is obvious: take a long time to decide against keeping a stranger’s wallet and observers wonder what you were weighing. In that work the effect reversed for actions observers considered wrong, where hesitating before doing the wrong thing read as a point in someone’s favor, which is the closer parallel to what happens in these studies.

Grey and colleagues explain the difference by arguing that many political positions do not directly affect the person holding them, so a pause reads as effort rather than temptation, and that people already believe political decisions ought to be made deliberatively. They add that where self-interest is salient, a tax that would land on the person answering, they would expect the thoughtfulness reading to weaken. That is their interpretation of why the sign flips; the studies did not test it.

The mechanism runs through other people’s inferences, which makes it close to unusable on purpose. Set out to produce the impression of having deliberated and you are doing something other than deliberating, and the paper never tested a performed pause.

Deliberating recovers a little of what disagreeing costs, and only that. Among those who already agree with a position, the reasoning behind it barely registers. The audience least interested in how someone got there is the one that already ended up in the same place.