I think the small effect sizes are not as important for predicting overall cognition as the larger ones. If damage happens to a specific part of the brain and not others (as in e.g. most non-fatal strokes), that causes a lot more functional impairment than you would expect if you focused on all of the many parts of the brain that weren’t damaged.
If I’m understanding you correctly, you’re saying that you think that a 1-2 statistically significant moderate effect sizes are more worrisome than if this research had discovered statistically significant but small effect sizes in every category. You expect that functional impairment on real-world tasks would be serious and wide-ranging as a consequence of moderate impairment in a single cognitive domain.
First, I think this is a great crux, and highlights the importance of carefully framing a research question.
Second, I think we ought to consider a few possible functional performance scenarios:
Tasks that stretch cognitive capacity to the maximum extent, often by design. Example: a chess championship or math exam.
Tasks that typically use limited cognitive capacity, but in which mistakes can have severe consequences and rare emergencies can occur at any time that will result in much higher cognitive demands than normal. Example: jet pilots, soldiers on patrol.
Tasks that use limited cognitive capacity, allow for corrections and adjustments, and are primarily bottlenecked by time, skill, and material resources. Example: most jobs, tasks of daily living, sailing competitions.
I think that scenarios (1) and (2) are the best candidates for the hypothesis of greater functional impairment effects relative to limited and modest cognitive function deficits, and scenario (3) is best for the hypothesis that modest and limited cognitive function deficits aren’t too concerning.
Some challenges might even be mixed. For example, we could imagine that in school coursework, learning is bottlenecked by hours of study in general. However, on exams, for well-prepared students, hours of sleep might become the limiting factor on their performance. Hence, an idealized study strategy might want to routinely sacrifice some sleep in exchange for more study time, but then increase sleep on the days leading up to an exam.
Under this mixed framework, a strategic sleep regimen would be flexible, based on the specific performance demands the person in question was facing.
Note that one of my meta-analyses found that (the absolute value of) the effect size of sleep restriction on attention is higher when the comparison group undergoes sleep extension rather than when it undergoes no intervention. This is at least weak evidence that the dose-response relationship between sleep duration and cognition doesn’t flatten out once you get to the average amount of time people sleep.
I’ll check into this when I have time—it’s interesting!
The meta-analysis itself tells you the average sleep deficit across the studies compared with age-adjusted sleep recommendations, as I pointed out in my post, and that number is probably more relevant than those averages.
The sleep times I referred to track closely with the average age-adjusted sleep deprivations, but you are right that it would have been more apt to choose the latter instead of the former. I simply find it more intuitive to gauge what X hours of sleep feels like than Y hours of sleep deprivation.
Edit:
As a followup, I’m not sure how to think about the relationship between single-factor cognitive effects and broad functional effects.
At first glance, it seems to me that if we hypothesize that a single-factor cognitive deficit corresponds to broad functional impairment, that we’re essentially saying that you’re held back by whatever cognitive faculty is furthest below baseline. That in turn would indicate that any single-factor cognitive improvements obtained by sleep extension (i.e. alertness) would not correspond to functional improvements, if any cognitive factors fail to be improved by sleep extension.
On the other hand, it could be that, in either direction, the cognitive factors that are harmed/helped by sleep adjustments happen to be the load-bearing factors for most activities. Maybe executive function and sustained attention are the limiting cognitive resource for most people at baseline, and alertness is the typical limiting cognitive resource above that. If that were true, we’d expect to see broad functional improvements or impairments across wider ranges of sleep adjustment.
I’m most inclined toward the idea that tasks have a heterogeneous relationship with particular cognitive capacities, and thus with the impact of sleep. This suggests that people can reap the most benefit by tailoring their sleep schedule to their particular requirements.
Busy parents with regular jobs that aren’t too cognitively demanding or hazardous might benefit most from a couple extra hours of wakefulness every day to get things done, and indeed, they often seem to opt for that strategy. By contrast, airlines apparently have strict policies for how long pilots must sleep before a shift, and I’ve also heard an anecdote from a biologist in a BSL4 facility that they don’t do inoculations unless they’re in absolutely tip-top cognitive shape. If they got bad sleep, or woke up with a crick in their neck, they reschedule.
My guess is that many people are modestly strategic about their sleep budget. But if this hypothesis is correct, there may be substantial gains to be reaped by individualized, flexible tailoring of the sleep schedule, rather than issuing a blanket recommendation of X hours of sleep across the board.
If I’m understanding you correctly, you’re saying that you think that a 1-2 statistically significant moderate effect sizes are more worrisome than if this research had discovered statistically significant but small effect sizes in every category. You expect that functional impairment on real-world tasks would be serious and wide-ranging as a consequence of moderate impairment in a single cognitive domain.
First, I think this is a great crux, and highlights the importance of carefully framing a research question.
Second, I think we ought to consider a few possible functional performance scenarios:
Tasks that stretch cognitive capacity to the maximum extent, often by design. Example: a chess championship or math exam.
Tasks that typically use limited cognitive capacity, but in which mistakes can have severe consequences and rare emergencies can occur at any time that will result in much higher cognitive demands than normal. Example: jet pilots, soldiers on patrol.
Tasks that use limited cognitive capacity, allow for corrections and adjustments, and are primarily bottlenecked by time, skill, and material resources. Example: most jobs, tasks of daily living, sailing competitions.
I think that scenarios (1) and (2) are the best candidates for the hypothesis of greater functional impairment effects relative to limited and modest cognitive function deficits, and scenario (3) is best for the hypothesis that modest and limited cognitive function deficits aren’t too concerning.
Some challenges might even be mixed. For example, we could imagine that in school coursework, learning is bottlenecked by hours of study in general. However, on exams, for well-prepared students, hours of sleep might become the limiting factor on their performance. Hence, an idealized study strategy might want to routinely sacrifice some sleep in exchange for more study time, but then increase sleep on the days leading up to an exam.
Under this mixed framework, a strategic sleep regimen would be flexible, based on the specific performance demands the person in question was facing.
I’ll check into this when I have time—it’s interesting!
The sleep times I referred to track closely with the average age-adjusted sleep deprivations, but you are right that it would have been more apt to choose the latter instead of the former. I simply find it more intuitive to gauge what X hours of sleep feels like than Y hours of sleep deprivation.
Edit:
As a followup, I’m not sure how to think about the relationship between single-factor cognitive effects and broad functional effects.
At first glance, it seems to me that if we hypothesize that a single-factor cognitive deficit corresponds to broad functional impairment, that we’re essentially saying that you’re held back by whatever cognitive faculty is furthest below baseline. That in turn would indicate that any single-factor cognitive improvements obtained by sleep extension (i.e. alertness) would not correspond to functional improvements, if any cognitive factors fail to be improved by sleep extension.
On the other hand, it could be that, in either direction, the cognitive factors that are harmed/helped by sleep adjustments happen to be the load-bearing factors for most activities. Maybe executive function and sustained attention are the limiting cognitive resource for most people at baseline, and alertness is the typical limiting cognitive resource above that. If that were true, we’d expect to see broad functional improvements or impairments across wider ranges of sleep adjustment.
I’m most inclined toward the idea that tasks have a heterogeneous relationship with particular cognitive capacities, and thus with the impact of sleep. This suggests that people can reap the most benefit by tailoring their sleep schedule to their particular requirements.
Busy parents with regular jobs that aren’t too cognitively demanding or hazardous might benefit most from a couple extra hours of wakefulness every day to get things done, and indeed, they often seem to opt for that strategy. By contrast, airlines apparently have strict policies for how long pilots must sleep before a shift, and I’ve also heard an anecdote from a biologist in a BSL4 facility that they don’t do inoculations unless they’re in absolutely tip-top cognitive shape. If they got bad sleep, or woke up with a crick in their neck, they reschedule.
My guess is that many people are modestly strategic about their sleep budget. But if this hypothesis is correct, there may be substantial gains to be reaped by individualized, flexible tailoring of the sleep schedule, rather than issuing a blanket recommendation of X hours of sleep across the board.