How much does sleep restriction actually hurt cognition?
Measurably, and unevenly across mental abilities. This meta-analysis pooled 61 studies drawn from 71 different populations using random-effects models, and found that experimentally restricting sleep produced a small to moderate decline in cognitive performance overall.
The damage was not spread evenly. Sustained attention took the clearest hit. Several other domains had too little evidence to reach a conclusion at all.
Effect sizes by domain (Hedges g, where negative means worse performance):
- Sustained attention: g = -0.409, the largest confirmed effect
- Overall cognitive processing: g = -0.383
- Executive functioning: g = -0.324
- Long-term memory: g = -0.192, the weakest of the confirmed effects
- Attention, multitasking, impulsive decision-making, and intelligence: insufficient evidence to detect an effect
Reading effect sizes without overstating them
A Hedges g of roughly -0.4 is conventionally a small to moderate effect. It means the average sleep-restricted person performs about four tenths of a standard deviation below the average rested person. That is real and it is worth avoiding. It is not the same as saying performance drops by half.
I am flagging this because the popular framing of sleep research tends to convert effect sizes into dramatic percentages that the underlying papers never reported. This meta-analysis reports standardized effect sizes, not percentage decrements, and the honest presentation keeps them in that form.
Dr. Kumar’s Take
The finding I would actually change behavior over is that sustained attention is the most affected domain with the clearest evidence.
Sustained attention is what keeps you tracking something monotonous over time. It is the ability that fails on a long drive, in the back half of a night shift, and in the ninth hour of a case. It is also the ability people are worst at self-monitoring, because the failures are brief lapses rather than a continuous sense of being impaired. You do not feel yourself stop paying attention. That combination, largest measured deficit and poorest self-awareness, is what makes it the practically dangerous one.
Equally important is what this analysis could not establish. Attention as a broader category, multitasking, impulsive decision-making, and intelligence all had insufficient evidence. That is not a finding that sleep loss spares them. It means the studies to answer the question have not been done in sufficient number or quality. Absence of evidence is a genuinely different thing from evidence of absence, and a lot of confident writing about sleep and decision-making runs past that distinction.
The long-term memory effect at g = -0.192 is small enough that I would not build advice on it from this analysis alone.
Study Limitations
Meta-analyses inherit the weaknesses of what they pool, and the included studies varied in how sleep was restricted, for how many nights, and which cognitive tests were used. Four of the domains examined lacked sufficient evidence for any conclusion. The analysis covers experimentally manipulated sleep restriction in study conditions, which is not the same as chronic real-world sleep debt accumulated over months or years.
Related Studies and Research
- Sleep Deprivation vs Alcohol: 17-19 Hours Awake Equals 0.05 BAC
- One Night of Partial Sleep Loss Triggers Insulin Resistance
- Acute Sleep Loss Impairs Physical Performance
- Overnight Therapy: How Sleep Processes Emotions
- The Two-Process Model of Sleep Regulation
FAQs
How many studies were included?
61 studies drawn from 71 different populations, analyzed with random-effects models.
Which mental ability suffers most?
Sustained attention, at g = -0.409, the largest effect with sufficient evidence behind it. Executive functioning followed at g = -0.324.
Does sleep restriction lower intelligence?
This analysis could not tell. Intelligence was among the domains with insufficient evidence to detect an effect, which means the question is unresolved rather than answered in either direction.
What does a Hedges g of -0.4 mean in practice?
It is a small to moderate effect: the average restricted person scores roughly four tenths of a standard deviation below the average rested person. Meaningful, but not a collapse in function.
Bottom Line
Across 61 studies, experimental sleep restriction produced a small to moderate decline in cognition overall (g = -0.383), with sustained attention most affected (g = -0.409) and long-term memory least (g = -0.192). Attention as a broad category, multitasking, impulsive decision-making, and intelligence had too little evidence to judge.
Read the study
Lowe CJ, Safati A, Hall PA. The neurocognitive consequences of sleep restriction: A meta-analytic review. Neuroscience and Biobehavioral Reviews. 2017;80:586-604.
