We have absolutely no idea how humans respond to partial gravity. Obviously, we have a bunch of G=1 data, and we have a bit of G=0 (weightless) data from various space-station and lunar missions - and that's it. We also know that 1/6 G is not detrimental to humans over a 2 or 3 day period.
The human response (HR) to partial G is shown on the graph as a scale from 0 to 1. If we arbitrarily define HR=1 to match g=1 (that is, humans function normally under one Earth gravity), we might assign a value of .3 to g=0 (that is, weightless humans don't function normally at all, we lose muscle mass, we lose bone density, we experience digestive troubles, etc.). The million-dollar question is, what does this graph look like between g=0 and g=1, our two known datapoints?
We very strongly suspect the function is continuous, and likely increases monotonically from g=0 to g=1. Four possibilities are shown on the graph:
1) Quadratic. The argument here is that our bodies are made for G=1, and they perform dramatically worse the further away we get from that.
2) Linear. In absence of other data, a linear hypothesis is often the first guess when only two data points are known
3) Pathological: Everyone would be very surprised if this turned out to be our response! We function normally at 1 g and at .25 g, but .62 g kills us. Silly - but we don't have the data to disprove it yet.
4) Expected. This is what I anticipate we'll find as we experiment with fractional g: Our bodies only show the zero g symptoms when we are very close to zero g, and we quickly move toward a HR of 1.0 as we enter the realm of fractional g. It could even be that we perform better at .8 g than we do at 1 g; I suspect a population of senior citizens would think so!
Now, as we think about extended trips to the Moon (about 1/6 g), Mars (just over 1/3 g), etc., we have 3 key questions to answer:
1) What is the human response to partial g? In other words, what does that graph truly look like?
2) Various attributes of low-g debilitation probably each have their own response curve - is there any way to quantify these?
3) We can generate artificial gravity by circular motion. We can separate two masses with a rope in zero G, and spin them about their mutual center of mass, and we'll generate an acceleration (a=v^2 / R). How sensitive are we to the spin rate? Are we OK at 1 RPM, or can we go higher?
This would be an outstanding experiment for the ISS - forget those silly micro-g experiments and start doing something that would actually matter. We'd need two habs (or a hab and a counterweight) that can be winched out on cables and spun. This would allow us to fill out the above graph and experiment with rotation rates. This would be a good prerequisite for any extended stay on the Moon or Mars.
This is what makes the Mars Gravity Biosatellite (link #2) experiment with mice and martian-G so fascinating. A satellite will be launched holding a population of mice. The satellite will spin about its own axis fast enough that the periphery of the satellite experiences 1 Martian gravity, and it will stay up long enough for a generation of mice to be born and mature to adulthood. Then the satellite will return, and the mice will be studied to see if the martian-G effects are visible in an autopsy.
There are some problems with extending this data to humans: We're not mice, and this satellite will spin at about 30 RPM, so it may be difficult to decouple partial-G effects from high-spin-rate effects. But, we currently have exactly zero data about mammalian response to martian-G, and this will at least get us in the game.