AIM Framework glossary
Source Inference
Canonical definition
The reconstruction of the source-decomposition of an integrated valuation from evidence other than the integrated Scalar itself, principally the Behavioural Signatures of the contributing sources together with prior knowledge that a given source exists and of its characteristic operating dynamics. Source Inference is the permitted complement to Source Opacity, which establishes only that the decomposition cannot be recovered by any function of the integrated Scalar alone. It does not attempt to invert the many-to-one Common-Currency Integration map, which cannot be done, but instead estimates the composition of a valuation by fitting known source-models to observable evidence: the A-signature of state-dependence and satiation, the I-signature of engagement that persists without an audience, and the M-signature of non-satiation and, in M's special case, audience-dependence and reference-point behaviour. The Two-Question Diagnostic is the framework's canonical Source Inference procedure at the level of the individual valuation, and the Aggregation Gradient is its correlate at population scale. Source Inference yields an approximation of the decomposition whose accuracy improves as the priors supplied to it sharpen, and it is the mechanism by which the Knowledge Patch routes around Source Opacity rather than curing it.
Explanation
Because the Decision Hub blends Appetites, Intrinsic Motivation, and Mimetic Desire into a single wanting, you cannot look inward and read off how much of a desire came from which source. That tag is gone, and no amount of reflection brings it back. But you are not therefore stuck. If you already know the three sources exist and know how each one behaves (Appetites fade once satisfied, Intrinsic Motivation keeps going even when nobody is watching, and Mimetic Desire never settles and depends on an Audience), you can work out a good estimate of what was really driving a choice by looking at how it behaves rather than how it feels from the inside. That is Source Inference, and it is exactly what the Two-Question Diagnostic does. You cannot un-mix the blend, but you can make a reliable estimate of what went into it.
Epistemic status
Framework-Specific Construct. Derived from Source Opacity (Axiom 3) and the Non-Recoverability Is Not Non-Existence corollary (Chapter 9, Section 9.9.5).