Density effect of 1440 dpi vs. 2880 dpi

I am using Ron Reeder’s technique for printing digital negatives on the Epson SP4000 using Quad Tone RIP and standard Epson Ultrachrome inks. I am using the Matte Black as Ron indicated. At the start I did not print using the Unidirectional or 2880 dpi settings as Ron suggests in his article. In my experience the printing quality between the 1440 and 2880 is negligible to the naked eye at a standard viewing distance and the savings in ink is to my liking. Part way through a series of test prints, I decided to try one with the 2880 setting instead of 1440. To my surprise it made a difference but not in the way I would have expected. There was some perceptual difference in the quality of the negative. In particular there was a reduction in visible banding. Once I made a print from this negative, I also discovered two other differences. In the first, it appears that the densities are more linear with fewer spikes or divets in the curve. I assume this is due to a smoother printing of gradients from the increased dot resolution. The second difference was also in the density of the negative. I had expected that there would be an increase in density over the entire negative as there are more dots and therefore more ink being laid down. Lo and behold, the opposite is true. The density of the negative overall was reduced by about two thirds of a stop in exposure. I am surprised by this and do not understand the reason. If anyone out there has any insight to this, I am interested. I am also uncertain as to whether this reduction in density is an arithmetic or geometric one.