Ablations show the semi-Markov controller is the primary driver of behavioral fidelity (removal pushes DKL to 6.81) and Strategist/Executor separation is essential for epistemic fidelity and realism¶

Claim · All claims
Evidence · 1 study (2 entries) · design r2 · q2
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? Removing the semi-Markov controller causes catastrophic divergence from real student distributions (DKL rises to 6.81, versus ≤0.32 for all other ablations). → Hanchen David Wang 2026 q2 i? Merging Strategist/Executor into a Combined Agent produces the largest realism drop (2.44→1.88) and reduces error recurrence by 21% (86.2%→65.3%). → Hanchen David Wang 2026 (2)

Evidence¶

Hanchen David Wang 2026¶

Hanchen David Wang, Clayton Cohn, Zifan Xu, Siyuan Guo, Gautam Biswas, Meiyi Ma. (2026). BEAGLE: Behavior-Enforced Agent for Grounded Learner Emulation. Preprint. https://arxiv.org/abs/2602.13280 doi:10.48550/arxiv.2602.13280

q2 · i? · design · r2

Ablation study (N=50, T=30, Gemini 2.0 Flash, Table 2) removing individual BEAGLE modules. The semi-Markov ablation's DKL of 6.81 dwarfs all other variants (0.26–0.32), identifying it as the primary driver of behavioral fidelity.

"Its removal causes catastrophic divergence from real student distributions ( DKL rises to 6.81, compared to ≤0.32 for all other ablations), confirming that standard LLMs inherently default to linear construction."

Hanchen David Wang 2026 (2)¶

Hanchen David Wang, Clayton Cohn, Zifan Xu, Siyuan Guo, Gautam Biswas, Meiyi Ma. (2026). BEAGLE: Behavior-Enforced Agent for Grounded Learner Emulation. Preprint. https://arxiv.org/abs/2602.13280

q2 · i? · design · r2

Same ablation study (Table 2): the Combined Agent variant, which merges the two LLM stages, shows that without task framing the LLM reverts to skilled error resolution, silently correcting mistakes instead of exhibiting diagnostic incompetence.

"Merging Strategist/Executor into a Combined Agent produces the largest realism drop (2.44→1.88 ) and reduces error recurrence by 21% (86.2%→65.3% )."

Discussion¶

Related Claims¶

• BEAGLE's error recurrence rate (86.2%) falls within the realistic novice envelope, far above vanilla LLMs (7.8%) and below rule-based SimStudent's mechanical saturation (92.0%) — related
• BEAGLE achieves higher behavioral fidelity to real student cognitive-behavior distributions than LLM prompting and agent baselines on a Particle Simulator task (DKL = 0.31 vs. baselines ≥ 0.53) — related
• BEAGLE generalizes across tasks and LLM backbones: on the out-of-distribution Gradient Descent task it still shrinks DKL by ≥3×, raises error recurrence to ≥79%, and improves all three perceptual scores — related
• Ablations show removing the strategy selector substantially increases leakage and replacing the NLI verifier with same-model self-check reduces performance — related
• In a simulated tutoring agent, removing the behavioral-safety constraint (C3) appears to cause collapse to near-single-action policies, which the authors read as C3 being the most influential safeguard, though its ablation does not differ significantly from removing C1 in RHSI — related
• Federated aggregation is the primary driver of performance gains, contributing an average 11.99% improvement — related
• In the EduAgent framework, removing past cognitive states from memory significantly reduced simulation performance — related