A new monograph is out from RAND's Ben Connable. You can find a link to the full text here. The abstract is below.
In the hostile, complex, and chaotic counterinsurgency environment, people can support the government and the insurgency to varying degrees at the same time — and be similarly resentful of both. Identifying all but the unequivocally irredeemable as an "enemy" and labeling anyone wearing a government uniform as a "friend" not only creates a false paradigm of human identity, but it also artificially bounds the U.S. military's options for influencing a population during a counterinsurgency operation. Analyzing complex environments, such as Iraq or Afghanistan, through simplified approaches that incorporate color-coding and enforce a strict division of analytic specialties can lead analysts to make unhelpful and logically unsound assumptions about human identity. Color-coded, enemy-centric analyses also reinforce the inaccurate and unhelpful notion that the enemy and society are separate constructs in the counterinsurgency environment, or separate subsystems (or groups) within a larger societal system. On the contrary, what is needed is an all-source, holistic, fused approach to analysis that takes into account sociocultural ambiguities. This paper proposes a paradigm shift in how intelligence is combined for analysis and how the product of that analysis can provide a more complete picture of counterinsurgency operations for commanders and other decisionmakers. The concept of behavioral intelligence analysis discards the old method of color-coding in favor of a spectrum of hostility. In other words, analysts would work from the assumption that all actors might have the capacity to behave in a way that is more or less conducive to the U.S. military's objectives in a conflict.