The Four-Dimensional Structure of Semantic Transitions: A Model of Contextual Jumps Derived from the Free Energy Principle

Background: Conventional three-dimensional semantic network models fail to account for the low-cost nature of instantaneous semantic transitions across contextual frames. Under the free energy principle (FEP), such transitions should incur cumulative complexity costs, yet the brain performs them efficiently.Hypothesis: We propose the Semantic Tesseract Hypothesis, redefining semantic space as a four-dimensional hypercube in which contextual switching occurs via a dedicated W-axis, bypassing associative distance entirely.Methods: Whereas v1 sampled KL divergence from scalar values, v2 represents each semantic node as a multimodal feature vector (visual, auditory, emotional, spatial, motor; 5 dimensions) inspired by Barsalou's Perceptual Symbol Systems. Monte Carlo simulations (N = 10,000 trials) compared cumulative costs between the two models.Results: The W-axis advantage emerged gradually, increasing rapidly from N ≥ 4 and reaching complete dominance at N ≥ 9 (P = 1.00). For short-range transitions (N = 1), local associative paths were more cost-efficient (P = 0.23).Conclusion: The model describes cognitive flexibility as adaptive selection between W-axis transitions and three-dimensional associative paths, presenting falsifiable predictions including an N400 inversion effect.