the study, followed by a period of sleep deprivation (four hours in bed) the next six nights of the study, and finally a period of sleep recovery (12 hours in bed) the last seven nights of the study. In reporting the effects on the body (the discussion of glucose metabolism, in the fifth paragraph of the article) the author’s compare the sleep deprivation stage only to the sleep recovery stage, not to normal sleep. This seems to me like doing an experiment on drunkenness and comparing the drunk stage to the hangover stage, without ever reporting what happens when the person is sober. Since normal sleep would seem to be the appropriate control condition here, the absence of results from that condition makes me wonder if something unusual was found there and not reported in the article.Another potential problem comes from the sequential nature of the different sleep conditions. All participants had normal sleep, then sleep deprivation, then sleep recovery (in that order). Therefore, the three conditions differ not only in the amount of sleep, but also in the level of familiarity with the experimental procedures. Why should that make a difference? Well, consider the results reported for the stress hormone cortisol. In comparing people who are sleep deprived (days 4-9 of the study) with those who are in sleep recovery (days 10-16 of the study), they are comparing people early in the experiment to those later in the experiment. If the experimental procedures are themselves stressful (e.g. drawing a blood sample) then a person who has been through it more often might find it less stressful, and therefore not respond so strongly. This fact, rather than the amount of sleep itself, might explain the pattern of results in the physiological data.I’m not sure whether the methodological issues I’ve raised really do account for all the study’s results, but they should be corrected in future research on sleep deprivat...
