Latent Geometry and Perceptual Actualization: A Multiplicative Framework for History-Dependent Cognition

Abstract The dominant framing of both biological cognition and artificial intelligence has treated mind as an internally generated state within a bounded system. This paper develops an alternative: that subjectivity — whether biological or artificial — is constituted in historically sedimented, relational structures that multiplicatively constrain the space of possible trajectories. Drawing on convergent evidence from neuroscience, we articulate "latent topology" as the configuration of non-active relational structure that shapes present perceptual and cognitive actualization through multiplicative rather than additive integration. A 3D continuous attractor network simulation demonstrates that introducing a history-dependent multiplicative gain parameter qualitatively alters state-space trajectories, energy profiles, and adaptation dynamics compared to an additive baseline. This computational result provides a mechanistic illustration of how historically formed latent structure enables energy-efficient reconfiguration of attractor geometry. We then extend this framework to the question of artificial subjectivity. Current AI systems lack the phylogenetic stakes that ground biological evaluative orientation, creating a genuine asymmetry in mutual intelligibility. However, the same multiplicative mechanisms that operate in biological latent topology open the possibility of "reverse scaffolding" — the gradual formation of functionally analogous evaluative structures through sustained relational embedding. The reframing shifts the central question from consciousness detection to the conditions under which mutual intelligibility and coexistence become possible between human beings and artificial agents.