Latent Task Architecture: Modeling Cognitive Readiness Under Unresolved Intentions

The Latent Task Architecture (LTA) proposes a control-theoretic framework for modeling how deferred, unresolved cognitive structures modulate ignition dynamics and effort regulation. Extending the Cognitive Drive Architecture (CDA) and Cognitive Thermostat Theory (CTT), LTA introduces the construct of Latent Load (LL), a bounded, continuous signal that estimates the influence of non-executed task representations on readiness parameters: Grain, Anchory, and CAP. Through mathematical formalization, simulation scenarios, and predictive modeling, LTA demonstrates that latent structures may act as dynamic modulators of cognitive engagement, although causality remains to be empirically isolated. Critical considerations highlight that LL formalization, while computationally tractable, may not fully capture the stochastic, fragmented nature of real latent cognition; further, boundary conditions between active and latent states require sharper definition. Ethical risks of latent-state sensing, especially regarding cognitive privacy and autonomy, are recognized as first-order design constraints, not secondary concerns. Moreover, the necessity of multi-timescale modeling is emphasized, given cognition's sensitivity to both rapid cue dynamics and slow latent accumulation. Rather than presenting a finalized model of mind, LTA opens a structured dialogue, framing cognitive stability as an emergent negotiation between present action and latent potential, an architecture still to be fully formalized, empirically validated, and ethically governed.