Modelling eye-head coordination without pre-planning--a reflex-based approach

The gaze orientation system is a prime example of the CNS using multiple platforms to achieve its goal. To move the gaze in space, the eyes, head, and body cooperate to place the image of the target on the fovea. Understanding the underlying neural circuitry innervating this collaboration could also be a cue to understanding other movement related CNS tasks involving multiple platforms, i.e., posture and locomotion. Basically two major network topologies for modeling the gaze orientation system have been proposed: the independent controller model and the shared gaze feedback controller model. In the independent controller model, each platform (i.e., eyes, head or trunk) receives its own share of the retinal error (distance of the target from the current gaze position) independent from other platform(s) and its goal is to null its individual error, whereas, in the shared gaze feedback controller all platforms collaborate to null the shared global error, which is calculated on the fly using feedback from all platforms or reflexes. Each of the mentioned general topologies has its own supporters and the question is which does the CNS actually use. In this article, based on evidence from neurophysiology and behavior, complemented by simulation data, it will be shown why a shared feedback controller is the better candidate for this task. More specifically, simulations of an updated Prsa-Galiana model (the Shared Sensory-Motor Integration (SMI) model) will be discussed in more detail and, where applicable, compared with other popular models, including independent and shared controller models. It provides plausible explanations for observations on gaze shifts with various interventions.