Scaffolding the Strategist: Architecture-Dependent Reasoning Interventions in Hotelling Spatial Markets

Researchers at an undisclosed institution tested GPT-4.1-mini and GPT-5-mini on Hotelling's linear city model and found that reasoning interventions interact with model architecture: commitment scaffolding improved the standard model but degraded the reasoning model, while principled separation showed the opposite pattern. Adversarial stress-testing harmed both models, with the reasoning model suffering 2.6 times greater degradation. The study highlights a declarative-procedural gap where models identify correct strategies but fail to execute them.

arXiv:2607.09743v1 Announce Type: new Abstract: We investigate whether structured reasoning interventions improve the strategic economic reasoning of large language models, and whether their effects depend on model architecture. Using Hotelling's linear city model as a diagnostic vehicle, we evaluate GPT-4.1-mini a standard instruction-following model and GPT-5-mini a reasoning-optimized model under five conditions - an unscaffolded baseline and four reasoning interventions - across eight questions spanning deductive and abductive reasoning, three prompt framings, and three repetitions per condition, yielding 720 individually judged responses. We find a statistically significant crossover interaction between scaffolding type and model architecture $t 7 = 4.79$, $p = 0.002$, $d = 1.69$ : commitment scaffolding improves the standard model $+0.21$ while degrading the reasoning model $-0.63$ , and principled separation shows the opposite pattern $-0.40$ vs. $+0.31$ . Both crossovers are individually significant commitment: $p = 0.040$; separation: $p = 0.002$ and hold across all eight questions with 7/8 directional consistency. Adversarial stress-testing harms both models, with $2.6\times$ greater degradation for the reasoning model $-1.47$ vs. $-0.57$; $p = 0.038$ , and the damage correlates negatively with baseline difficulty $R^2 = 0.36$, $p = 0.014$ . We further document a persistent declarative-procedural gap in which both models identify correct strategies at rates far exceeding their ability to execute them; separation fully closes this gap for the reasoning model while no intervention helps the standard model.