We could call one theory "the lost energy theory," though that's not a name used by others. In this theory, dust-laden wind from the west crosses a flattish plain of braided streams with drying mud flats -- a system like the Mississippi River was during the later stages of the Ice Age. When the wind encounters the plain's edge where running water has eroded into the uplands, causing a steep cliff to be formed, the wind's unimpeded progress across the plain is disturbed.
In energetic terms, since the cliff provides friction to the wind above it, the wind must expend some of its energy to reorder itself to flow over the cliff. When the wind spends this energy, it loses some of its ability to hold dust particles in the air, and so it must drop part of its load. As the wind moves across the upland landscape, more and more irregularities below it continue robbing energy, and more and more dust falls out, until eventually there's very little dust left.
Another reasonable-sounding theory could be called "the moistened particle theory," though, again, this isn't a name you find in books.
Here, as dust-laden wind flows across the braided stream system, dust particles bump into water molecules suspended above the braided streams. Each time a molecule of water adheres to the dust particle, the dust particle becomes a little heavier with moisture.
Many particles become so heavy that they drop into the braided stream network and are carried downstream, but some make it to the cliff on the plain's downwind side. Lighter particles make it a little farther. Some make it ten or even twenty miles inland before they fall, and some just keep going.
Earl Manning at Tulane has yet another explanation. In an email he proposes the following scenario which begins with "rock flour," which is silt suspended in water, the silt resulting from heavy glaciers grinding up rock over which they scrape, and the water resulting from the melting glacier:
"Rock flour is washed out of the base of glaciers in southern Illinois by streams. Braided streams distribute the rock flour along the broad middle Mississippi River floodplain. Strong glacial winds (partly created by the cold of the glacial ice) blow down the valley, drying the floodplain, and picking up the silt... the winds were slowed by tall trees in forests along the lower Mississippi River Valley. The trees would baffle the wind enough to slow it. The forest explanation has the advantage of also explaining the basic wedge-shaped cross-section of the loess deposit along the river, thinning as it moves away from the river... The farther away from the water of the river you got in glacial time, the fewer trees. Therefore the less baffling, and the less loess falling out of suspension. In dry glacial time, the area beyond the river was more prairie than forest, like along the Platte today.
This is a good place to remind ourselves that the concepts of "bluff height" and "loess depth" should not be confused. In the lower Mississippi River Valley, much of the height of bluffs and hills in the loess area is contributed not by loess, but rather by sands, gravels, and softer rocks such as siltstone, shale, marl, and nodular limestone lying below the loess. The accumulating loess just added height to them.