finding

active

finding:in-single-agent-simulation-of-32-trials-performance-becomes-perfect-after-trial-14-without-bayesian-model-reduction-with-confidence-increasing-progressively

In single-agent simulation of 32 trials, performance becomes perfect after trial 14 without Bayesian model reduction, with confidence increasing progressively.

Baseline learning curve for pure epistemic learning without structure learning

Source paper

extracted_from

Active Inference, Curiosity and Insight

(2017) · Karl Friston · Marco Lin · Chris Frith · Giovanni Pezzulo +2

Neighborhood — ranked by edge-count

Hypotheses (1)

hypothesis

Human participants in the rule-learning paradigm should acquire insight after approximately 7–8 trials, fewer than the ~14 required by Bayes-optimal inference without model reduction, suggesting they perform Bayesian model selection.
supports
Based on informal audience experiments; implies people use prior knowledge about rule structure

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.

Nearly all of 64 simulated agents attain 100% performance at around the 10th trial when allowed to perform abductive Bayesian model reduction after each trial.finding0.887
Group-level simulation result showing generalizability of BMR benefit across agents
Simulated agent achieves perfect performance after trial 12 following Bayesian model reduction (sleep), versus ~14 trials without BMR.finding0.864
Key result showing BMR dramatically accelerates rule learning in simulation
Bayesian model reduction after 12 trials correctly removes off-diagonal (redundant) parameters from the likelihood array, recovering the true contingency structure.finding0.819
Validation that BMR correctly identifies and prunes wrong connections in the likelihood mapping
A naive agent equipped with reduced priors from an experienced agent performs perfectly with maximum confidence from the first trial.finding0.810
Demonstration that model-level priors (not parameter-level knowledge) suffice for immediate transfer
In simulations, positive evidence threshold for Bayesian model reduction corresponds to ΔF ≤ −3, equivalent to odds ratio of exp(−3) ≈ 0.05 (reduced model ~20 times more likely than full model).finding0.793
Quantitative threshold used for accepting reduced models; linked to Bayes factor of ~20
The strict version of the simulation objective is optimized by the actual time evolution rule that created the training samples.claim0.767
Equivalence of optimal predictor to the physics of the data.
3 of 64 simulated agents exhibited superstitious (incorrect) abduction, leading to persistently poor performance, demonstrating a trade-off between ampliative benefit and susceptibility to false insight.finding0.767
Demonstration of failure mode of abductive model reduction
A model whose objective is prediction can simulate agents who optimize toward any objectives, with any degree of optimality (bounded above but not below by the model's power).claim0.766
Prediction orthogonality thesis.