A Developmental Model of Congenital Nystagmus

Purpose: Congenital nystagmus (CN) is a spontaneous oscillation of the eyes with an onset in the first few months of life. In 90% of affected children there is an associated underlying sensory defect (foveal hypoplasia, cone dysfunction, cataracts, etc.). In 10% no underlying visual defect can be found, and the nystagmus is labelled as ‘idiopathic’. CN appears to be a developmental anomaly of sensorimotor integration, as it is not have an onset later in infancy or beyond, but why such a wide variety of early onset visual defects should lead to life-long oscillation of the eyes is a mystery. Previous models have focussed on a systems level approach to explain how CN might be generated by known oculomotor circuits. We ask, instead, why CN might occur. Model: Our basic tenet is that infant visuomotor development is highly plastic during some early ‘critical’ period. A defect of foveal vision occurring during (and only during) this period leads to an anomalous connectivity in the oculomotor circuitry, which becomes permanent thereafter. We propose that circuitry normally used for precise foveal registration of a visual object (gaze holding, fixation, and smooth pursuit) develops to maintain some degree of image motion, as this would maximise contrast for a low spatial frequency system. However, this motion is in conflict with maintaining the image on the fovea (or its remnant). We explore the best oculomotor strategy to cope with this conflict. Results: The optimal strategy (in the least squares sense) is to oscillate the eyes in one meridian with alternating slow and quick (saccade) phases. Remarkably, the optimal waveform profile has an increasing-velocity profile. Many of the unique waveforms seen empirically in CN are also optimal strategies given realistic uncertainty in the initial position of a slow phase. Using non-linear dynamical systems analysis, we show that these ‘optimal’ oscillations have similar fractional correlation dimensions to observed data. We also show that a ‘null region’, as commonly observed in CN, would be an inevitable consequence of a velocity driven oculomotor system. Conclusions: We have developed a new approach to understanding oculomotor development, in which we examine the best strategy to maximise visual contrast. In a normal foveate visual system with fine oculomotor control, the best strategy is to develop good foveal registration, which we call ‘fixation’, and ‘smooth pursuit’. If, however, the fovea is absent or not being stimulated (eg. cataracts), the best strategy would be to develop oscillations of the type seen in CN. It implies that the chaotic oscillations are the result of a physiological developmental adaptive process. This is in contrast to the prevailing view that CN is a disease that can be ‘cured’. It is not surprising that CN has proven remarkably refractory to therapeutic intervention with only minimal (if any) long-term successes using drugs, surgery, or even biofeedback. We argue that CN is as adaptive and permanent as normal eye movements are in a normally sighted individual