On the thermodynamics of prediction under dissipative adaptation (2009.04006v1)

Abstract: On the one hand, the dissipated heat of a thermodynamic work extraction process upper bounds the non-predictive information, which the associated system encodes about its environment. Thus, emergent information processing capabilities can be understood from the perspective of a pressure towards high thermodynamic efficiency. On the other hand, the second law of thermodynamics plays a crucial role in the emergence of complex, self-organising dissipative structures. Such structures are thermodynamically favoured, because they can dissipate free energy reservoirs, which would not be accessible otherwise. Thereby, they allow a closed system to move from one meta-stable state to another meta-stable state of higher entropy. This paper will argue, that these two views are not contradictory, but that their combination allows to understand the transition from simple self-organising dissipative structures to complex information processing systems. If the efficiency required by a dissipative structure to harvest enough work from the channeled flow of free energy to maintain its own structure is high, there is a drive for this system to be predictive. Still, the existence of this dissipative system is thermodynamically favoured, compared to a situation without any dissipative structure. Due to the emergence of a hierarchy of dissipative systems, which by themselves are non-equilibrium structures that can be dissipated, such a drive develops naturally, as one ascends in this hierarchy further and further away from the initial driving disequilibrium.