Increasing spatial competition enhances visual prediction learning

Our previous work provides support for the idea that the development of visual perception in early infancy depends on progressive improvements in oculomotor skill. In particular, we have proposed and tested an eye-movement model that successfully reproduces infants' gaze patterns on two measures of visual attention. However, this result is due to explicit hand-tuning of a key parameter in the model. In the current simulation study, we investigate whether manipulating this parameter (i.e., the duration of spatial competition) enhances visual prediction learning. As expected, we find that prediction learning becomes more accurate as spatial competition in the eye-movement model is increased. This finding suggests that visual prediction learning can provide a meaningful error-feedback signal, which can be used to modulate spatial competition in the eye-movement model.