Data for 'A stochastic world model on gravity for stability inference'

The dataset contains all code and related data that can replicate the results of the publication.<br><br>PaperCode: <br>Fig1_GenerateStacks: Code to generate stable and unstable configurations. The configurations used in the manuscript were randomly selected from stable_configurations_info.pkl and unstable_configurations_info.pkl.<br>Fig1_GravityCharacters: Experimental code to measure human gravity sensitivity.<br>Fig2_MeasureStability: Experimental code to measure humans' stability inference and build relation with simulated results.<br>Fig3_GravityOrigin_RLmodel: Code to train a reinforcement learning model. Note that you can re-run this code to generate training trajectory and converged gravity with different configurations, thus we no longer provide data in this part.<br>Fig4_AccSpeedTradeoff: Code to build decision-making model and evaluate the judgment efficiency for stability inference under stochastic gravity.<br>utils: Some basic functions to build a basic environment, measure the stability of configuration in a physical environment, and perform manipulation by adjusting gravity (e.g., collapse detection, trajectory recording, etc.).<br><br>PaperData: <br>Figure1_configurations: Generated configurations from code in Fig1_GenerateStacks.<br>Figure1_subjbeh: Human behaviors collected from code in Fig1_GravityCharacters.<br>Figure2_stableJudgment_stimulus: stimulus used for stability judgment. <br>Figure2_subjbeh: Human behaviors collected from code in Fig2_MeasureStability.<br>Figure2_heightbias: Data for measuring height bias under different gravities.<br>Figure3: please generate data directly from code in Fig3_GravityOrigin_RLmodel.<br>Figure4: the relationship between simulated stability and human stability inference under different simulation times, and gravity variances. <br><br>Please also refer to the Methods for the computational details.