Full Indexical Conditioning augments Radford Neal's Full Non-Indexical Conditioning with an indexical selection term — and Neal named and sketched FIC himself, in the same 2006 paper, before rejecting it

anthropic-reasoningbayesian-inferencefull-indexical-conditioningfull-non-indexical-conditioningradford-nealpresumptuous-philosopherastrobiology

Brewer, Lewis & Goodman (2026) develop and apply "Full Indexical Conditioning" (FIC), a Bayesian framework for handling anthropic selection effects, against Radford Neal's 2006 "Full Non-Indexical Conditioning" (FNC) — a preprint the authors' own acknowledgements name as "the direct inspiration" for the paper, "even though we disagreed with it in the end." FNC instructs a reasoner to condition on the "global existence" of their evidence or memories — that such evidence exists somewhere in the universe — without any further indexical claim about which observer holds it. FIC keeps that premise but adds a selection term: that the reasoner in question is, in fact, the one who has that evidence. In the 2026 authors' words: "FIC conditions on all evidence available to the reasoner while considering that this evidence could have been otherwise."

The name and the augmentation are Neal's own, not the 2026 paper's. Brewer, Lewis & Goodman say so themselves — "In his paper, Neal (2006) briefly mentions a possible alternative to FNC which he termed Full Indexical Conditioning (FIC)" — and Neal's preprint, read directly, coins the term on its p. 9 in the act of discarding it: "SSA can perhaps be seen as arising from what might be called 'Full Indexical Conditioning', in which we assume that in addition to their memories, everyone also has some unique 'essence', and everyone in some sense knows what their own essence is." Neal's objection is to those essences ("It seems preferable to me to not introduce such mystical 'essences' unless ignoring them can be shown to produce implausible results"). What the 2026 paper contributes is not the idea's name but its machinery: the factorization of the selection step as P(AE, AS) = P(AE)P(AS | AE), the treatment of reference classes under it, and the astrobiological applications.

The authors present the addition as correcting a failure mode of FNC used alone — a tendency toward inappropriately presumptuous conclusions, structurally resembling Bostrom's "Presumptuous Philosopher" problem, when FNC is applied to astrobiological questions. An appendix extends FIC to two reasoners pooling evidence and flags that its simple joint-selection term (proportional to 1/N²) breaks down once the reasoners' circumstances are correlated rather than independent — a caveat the paper does not work through for its own astrobiological applications.