Intuition emerges in Maximum Caliber models at criticality
Whether large predictive models merely parrot their training data or produce genuine insight lacks a physical explanation. This work reports a primitive form of intuition that emerges as a metastable phase of learning that critically balances next-token prediction against future path-entropy. The intuition mechanism is discovered via mind-tuning, the minimal principle that imposes Maximum Caliber in predictive models with a control temperature-like parameter $λ$. Training on random walks in deterministic mazes reveals a rich phase diagram: imitation (low $λ$), rule-breaking hallucination (high $λ$), and a fragile in-between window exhibiting strong protocol-dependence (hysteresis) and multistability, where models spontaneously discover novel goal-directed strategies. These results are captured by an effective low-dimensional theory and frame intuition as an emergent property at the critical balance between memorizing what is and wondering what could be.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Statistical models of neural activity, criticality, and Zipf's law
In this overview, we discuss the connections between the observations of critical dynamics in neuronal networks and the maximum entropy models that are often used as statistical models of neural activity, focusing in par…
Modeling and Inferring Metacommunity Dynamics with Maximum Caliber
A major challenge for community ecology is to use distribution patterns to infer basic parameters of dynamical models without conducting laborious experimental manipulations. We present a novel framework drawn from stati…
A maximum-caliber approach to predicting perturbed folding kinetics due to mutations
We present a maximum-caliber method for inferring transition rates of a Markov State Model (MSM) with perturbed equilibrium populations, given estimates of state populations and rates for an unperturbed MSM. It is simila…
Protein FoldingMaximum Caliber Infers Effective Coupling and Response from Spiking Networks
The characterization of network and biophysical properties from neural spiking activity is an important goal in neuroscience. A framework that provides unbiased inference on causal synaptic interaction and single neural …
Path sampling of recurrent neural networks by incorporating known physics
Recurrent neural networks have seen widespread use in modeling dynamical systems in varied domains such as weather prediction, text prediction and several others. Often one wishes to supplement the experimentally observe…
Time SeriesTime Series Analysis