I get suspicious of studies that come out apparently reporting that a key feature of CFS is only a perception. Research is supposed to control for biases from both the subjects and researchers, but there are some features of CFS which I would be reluctant to accept at face value do not exist, and I would first question the paper before questioning the existence of the phenomenon. This is CFS afterall, where inadequate research practices seem pervasively routine.
I think there are even a few studies which put patients though an exercise test and they basically reported no symptom exacerbations or activity reductions as a result. If you are a patient who has crashed countless times from physical activity, you just know from the deepest recesses of your heart that it can exacerbate symptoms and reduce activity, and no amount of poorly conducted papers on unrepresentative cohorts will sway you into believing otherwise.
We always knew, from personal experience, that deconditioning was an inadequate explanation for primary disability in CFS. Research is showing us to be correct about that. I think we will be generally correct about other things too.
Similarly, if you are a patient who has brain fog which makes you struggle to do anything, difficulties reading and writing, make clumsy mistakes and take hours to finish a simple post, or crash from reading articles, or at times can barely even remain conscious let alone conduct a cognitive test, you cannot accept that you have "normal" cognitive function.
There is previous research studies showing cognitive impairments anyway, although IIRC the association between subjective and objective measures is not particularly consistent across studies. There are multiple possible explanations.
As wdb mentioned, subjective measures may not be reliable. As Bob mentioned, patients may try extra hard to do the same work (and Simon linked to brain scan evidence of increase neural recruitment which suggests this could be the case and it certainly feels like it subjectively too with a cost). As MeSci mentioned, the cognitive impairments may vary at different times. As Snow Leopard said, they need to control for pre-morbid intelligence (which is difficult). As alex3619 and Simon suggested, the tests may be inadequate to detect the problems we do experience in daily life. In a previous study, the deficits only became apparent towards the end of the testing, because of fatiguability. Stress testing also sounds good.
IIRC, the correlation between fatigue severity and physical activity levels is not particularly strong either. I suspect that if true this is because fatigue is not the most important symptom and patients tend to adjust their activity levels to how much fatigue they can tolerate, so the bed bound patients and high functioning patients may have similar levels of fatigue. Formal cognitive testing does not reveal whether someone can function normal over the entire day.