finding

active

finding:normal-0-9-and-chronic-0-1-agents-in-objective-only-non-stationary-category-perform-best-with-opposite-learning-rates

Normal (α=0.9) and chronic (α=0.1) agents in Objective-only non-stationary category perform best with opposite learning rates

Suggests fundamental differences in learning dynamics between normal and chronic perception models

Source paper

extracted_from

Exploration Through Introspection: A Self-Aware Reward Model

(2026) · Michael Petrowski · Milica Gašić

Neighborhood — ranked by edge-count

Claims (1)

claim

The chronic pain model outperforms the normal pain model in non-stationary environments despite producing negative well-being
supports
Surprising finding that maladaptive perception can yield superior task performance in changing environments

Questions (1)

question

Why does a maladaptive chronic pain perception outperform normal pain in non-stationary environments?
answered_by
Empirical puzzle raised by the surprising chronic model results

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

Chronic pain agent achieves M=4235.5, SD=180.3 COR in non-stationary All category (n=300), highest across all chronic resultsfinding0.804
Peak performance of chronic pain agents across all reward categories in non-stationary environment
Introspective agents generally outperform standard no-pain baseline agents across environments and reward categoriesclaim0.782
Central empirical claim of the paper supported by statistical tests
All three agent types (active inference, Q-learning, Bayesian RL) perform adequately in stationary FrozenLake; only active inference achieves Bayes-optimal behavior in non-stationary settings.finding0.779
Key empirical result validating online planning capability of active inference.
Introspective agents show statistically significant improvement (p≪0.05) over no-pain baselines across most reward categories and both environmentsfinding0.779
Main empirical result of the paper establishing general superiority of introspective agents
Under reward shaping (G=100, H=-100, F=0), Active Inference scored 99.52, Bayesian RL 99.77, Q-learning 95.56, with nearly identical behavior between belief-based agents.finding0.769
Table 2, row 3, showing equivalence when prior preferences match rewards.
No-pain baseline achieves M=1586.5, SD=631.2 COR in non-stationary Objective-only category (n=300)finding0.763
Baseline for non-stationary Objective-only; dramatically lower than both pain models
Smith & Boyd (1991) criteria are irrelevant across the vast majority of the space of possible agents.claim0.762
Dismissal of earlier criteria as too narrow.
The chronic agent's high performance despite negative well-being aligns with findings on chronic pain and quality of life in humansclaim0.760
Cross-domain interpretive claim linking computational results to human chronic pain literature