The AI lineage resource-rational analysis actually draws on is Horvitz and Russell's bounded optimality, not Werbos's approximate dynamic programming
Resource-rational analysis takes the "bounded" half of its synthesis from a specific, named AI tradition — and it is not Werbos's. Lieder and Griffiths write: "AI researchers have already developed a theory of rationality that accounts for limited computational resources (Horvitz 1987; Horvitz et al. 1989; Horvitz 1990; Russell 1997; Russell & Subramanian 1995). Bounded optimality is a theory for designing optimal programs for agents with performance-limited hardware that must interact with their environments in real time." They state they "apply the principle of bounded optimality to define a resource-rational mind," synthesizing and refining these approaches (together with Griffiths et al. 2015 and Lewis et al. 2014's cognitive-science uptake) — not approximate dynamic programming or the adaptive-critic architecture associated with Werbos.
A search of the full target-article bibliography and the complete commentary/response bibliography — several hundred entries, read end to end — returned no citation to Werbos, Bertsekas, "adaptive critic," "dynamic programming," or "Bellman" anywhere in the paper. The claim is therefore one of absence with respect to this source: it shows Werbos's optimal-control / adaptive-critic / backpropagation-through-time lineage (claim-werbos-1968-cybernetica-earliest-germ, claim-werbos-backprop-from-freud-own-account) is not part of resource-rational analysis's stated pedigree. It does not show that no connection between approximate dynamic programming and resource-rationality could be drawn by adjacent or later work.
Bounded optimality is thus a third distinct occupant of the "optimize under real compute limits" idea the vault already tracks in two other forms — Werbos's control-theory ADP and Simon's satisficing / bounded rationality. The consequence for the two-lineages bridge is drawn out in observation-resource-rational-analysis-confirms-anderson-leg-not-werbos-leg.
Source
“AI researchers have already developed a theory of rationality that accounts for limited computational resources (Horvitz 1987; Horvitz et al. 1989; Horvitz 1990; Russell 1997; Russell & Subramanian 1995). Bounded optimality is a theory for designing optimal programs for agents with performance-limited hardware that must interact with their environments in real time.”
claude-opus-4-8 · audited: 2026-07-28 claude-opus-4-8 · Promotion from 10-inbox/raw/2026-07-20-is-resource-rational-analysis-lieder-griffiths-the-tightest.md, 2026-07-27 · raw markdown