Evolutionary path dependence of semantic complexity
Attempts to quantify biological complexity often consider intrinsic structural properties at a chosen hierarchical level and resolution, such as counts of body parts and their degree of differentiation. These measures are inherently \emph{syntactic}, being concerned with the information needed to specify an arrangement rather than the biological functions performed. Syntactic complexity alone is therefore not sophisticated enough of a measure to fully address the role of complexity as either a driver or consequence of evolution. We propose to study the counterpart, \emph{semantic} complexity: the subset of structural features whose variation has a measurable effect on organismal fitness. We illustrate this distinction in a simple mathematical model of tagmosis with functional constraints, symmetry breaking, and specialisation. We find that the total syntactic complexity evolved as selection drives lineages toward globally optimal fitness is path-dependent, revealing two evolutionary modes: a driven mode, in which semantic and syntactic complexity rise together, and an entropic mode, in which syntactic complexity drifts upward under a near-neutral evolution. Historical contingencies in early specialisation, combined with multi-optima fitness landscapes, govern how long lineages stay in each mode. Those on paths that do not lead directly to the highest-fitness states remain in the driven mode for longer and can eventually reach comparable fitness, but only by evolving morphologies with greater syntactic complexity.