“You’re not even listening” is a line that arrives late in a day, and it almost always sounds like a verdict on the other person.
A pair of studies published together in 2014 suggests it can be a report on the night before as well, and, more oddly, on the night the person saying it had.
The diaries by the bed
The first study was small and patient. Seventy-eight people in romantic relationships, taking part alone rather than as couples, kept a sleep diary by the bed for two weeks and filled in an online diary each night. Three filed only one diary and were excluded; the rest filed 908 between them, about twelve nights each.
Sleep was scored the way sleep researchers usually score it: latency, duration, awakenings, quality and daytime tiredness, each recoded so that higher meant worse and summed into one number running from 0 to 15. Conflict was a single question at the end of the day, on a scale from one to five: did you and your partner experience conflict in your relationship today.
The average conflict score across those two weeks was 1.49. These were not people in warring relationships; they were having mostly uneventful days. The average poor-sleep score was 3.87, well down the scale it sits on.
Amie Gordon and Serena Chen, both at Berkeley at the time, found that people reported more conflict on days that followed a relatively poor night, and that this held with the previous day’s conflict statistically controlled, which rules out the tidier explanation that a bad fight had ruined the night’s sleep and then carried on into the morning.
The effect had two components, unequally strong. People who slept worse than others across the fortnight reported measurably more conflict than people who slept better. Within a single person, sleeping worse than their own usual went with more conflict the next day only marginally, at p under .06 — and that second version is the one a reader applies to themselves. Last night was bad, so today will be worse. It is the weaker of the two.
The link survived controls for how stressed, how anxious and how low the person had felt the day before, each entered separately, and for their satisfaction with the relationship on that same previous day. The authors also checked whether the partner had been the source of the bad night and report that the link held, though the numbers for that step are unusable: the paper prints an identical coefficient and test statistic for it as for the test immediately before it, on a different measure, so one of the two is almost certainly a transcription error.
The five-minute conversation
The second study brought couples in. Seventy-one heterosexual couples from the community took part, and two were excluded for not following instructions. They were young: average age twenty-two, though the range ran to fifty-six. On average they had been together about twenty-one months.
Each partner rated the previous night’s sleep on arrival. Both listed the three biggest sources of conflict in the relationship, one partner was picked at random to choose which they would discuss, and the couple was given five minutes on camera to work toward a resolution.
Afterwards, each person rated ten emotions they had felt during the conversation, and how much of each they thought their partner had felt. The gap between the guess and the partner’s own answer is one of the two measures this research cares about most: not how you felt, but how badly you read the person opposite. It also has the lowest internal consistency the study reports, .64, and it carries the finding people will remember. Three coders rated how positive and how negative each partner had been on the tapes, and agreed closely with each other.
People who had slept badly reported a lower ratio of positive to negative feeling through that conversation. The coders saw the same thing, though at that point the result was only marginal.
Poor sleep travels across the table
Then the part that gives this research its shape. The effects did not stay with the person who had slept badly. When one partner had slept poorly, the other also showed a lower ratio of positive to negative feeling, marginally so by self-report and significantly so in the coders’ ratings.
On the reading measure both directions held: the poorly slept partner read the other less accurately, and the other read the poorly slept partner less accurately too. The authors’ guesses are that a tired face gives less away, or that tired people say less about what they feel, so tiredness flattens the signal as well as the receiver.
Resolution was reported most often when both partners were well rested, and the authors’ summary is that having just one partner who slept poorly was enough to hinder it. When your partner had slept well, your own sleep still tracked whether the two of you got anywhere. When your partner had slept badly, that link no longer reached significance.
Women reported a worse emotional balance when their partner had slept badly; men’s self-reported feeling was not moved by their partner’s sleep. The interaction behind that sat at p under .06, the value this same paper calls merely marginal a few pages earlier, so it is thin.
Controlling for both partners’ depressive symptoms, and separately for relationship satisfaction, dropped several of the feeling results below even marginal significance. What held throughout every analysis were the two findings about reading each other and about reaching a resolution.
A Geneva team took the sleep away
Sleep in both studies is only what people said about it, rated night and morning in the diaries and at the door of the laboratory in the second. Nobody wore a monitor; nobody slept in a lab; nothing was manipulated, so no firm causal conclusion is available. The authors say all of this themselves, and recommend that someone run the experiment.
Someone did, in the years that followed, publishing online at the end of 2021. A team in Geneva recruited thirty couples, screened them for medical, psychiatric and sleep disorders, and assigned them at random either to a night of total sleep deprivation under supervision or to a normal night at home with a monitor on the wrist. The next day each couple discussed a recurring conflict for fifteen minutes, with saliva cortisol and self-reported feeling taken before and after.
That study found much less positive feeling in the sleep-deprived couples, both before the argument and after it. Cortisol was higher in them after the argument, though it was already higher just before it, while the couples were being given their instructions, at much the same size, and the overall time-by-condition test was not significant. Its authors point at the holes: the control couples spent the night at home while the deprived couples spent it in a laboratory, and the whole sample was unusually contented; the groups also differed at baseline on relationship satisfaction, which the analysis adjusted for.
It also reports, in its own words, that it “did not corroborate” the earlier correlational findings on reduced conflict resolution, reduced emotion recognition and increased aggression, naming Gordon and Chen among the work it did not bear out. They offer a small sample, a short discussion, the timing of the conflict measures and self-report for negative feelings among the reasons. On agreement the numbers even lean the other way, ten of the eighteen couples who reached one having been kept awake all night, though the test found nothing in that difference, and neither satisfaction with the outcome nor the rated severity of the conflict differed between the groups.
One claim has an experiment behind it
The split here is finer than it usually gets reported. That poor sleep drags on the positive feeling around a difficult conversation has support from a correlational and an experimental study both, but only if the correlational half is taken from the coders rather than the participants: sleep predicted the coders’ ratings of positive feeling, significantly for the partner and only marginally for the sleeper, and not people’s own reports of it, where the effect ran through negative feeling instead. Geneva found the reverse split, positive feeling down sharply and no detectable difference in negative feeling.
That poor sleep gets in the way of two people reaching a resolution rests on one correlational study Geneva did not corroborate. That it makes two people misread each other stands on the same study and was not tested in Geneva at all, which measured recognition of emotions in video clips of actors rather than accuracy about your own partner, and whose authors name impaired empathic accuracy among the psychological mechanisms they say still need work. Most of what both papers measured is what people said about themselves on a form; the Berkeley coders’ ratings, and Geneva’s saliva samples, wrist monitors and video-clip test, are the exceptions. What is on offer here is a reading of two papers by writers, with no clinical training behind it and no diagnosis in it, and the misreading claim, the one in the title, is held up by that single study alone.
There is no instruction hiding in any of this. An early night will not settle a specific argument, and the diary study’s own data point backwards as well as forwards: on days people felt more stressed, anxious or low, they slept worse that night, and on the nights they slept badly they were also more likely to say their partner had disturbed them, a figure that is one of the pair the paper prints identically and is best read loosely. Sleep and the day’s mood lean on each other. If the arguments in a relationship have taken a shape that a night’s sleep does not explain, a couples or family therapist is the person for that; if the sleep itself is bad most nights rather than occasionally, that belongs with a doctor.
Two people, two nights, and how much either of them actually slept: it is a duller sort of evidence than an argument usually invites, and it is the kind nobody consults at the time. The tired one is not the only one who stops seeing straight.
