Conscious active inference I: A quantum model naturally implements the path integral needed for real-time planning and control.
Computational and Structural Biotechnology Journal January 1, 2025 Michael C Wiest, Arjan Singh Puniani
Active inference, a framework for modeling how sentient agents optimize behavior and learning, requires probabilistic computations including a path integral over future trajectories. Classical biophysical neural models like Hodgkin-Huxley neurons are unlikely to perform these computations quickly enough in realistic contexts. The path integral underlying active inference is mathematically equivalent to quantum dynamics, suggesting a quantum model provides a natural mechanistic implementation. This first of two companion papers argues that classical neurons are insufficient for temporally deep active inference, while a quantum model offers a biologically plausible alternative. The second paper reviews evidence supporting the Orch OR quantum theory of consciousness in microtubules, explaining discrete perceptual inference cycles.