Functionally, a vertically split fovea should confer an advantage to the processor. Visual stumuli arriving to each eye would be vertically split and the two parts sent to different hemispheres, obeying the crossed nature of the visual pathways. I test the prediction of a functional advantage for the separate lateralisation of text processing from the two eyes. I explore this hypothesis by means of psycholinguistic experimentation and cognitive modelling. I employed a haploscope to show foveated text to the two eyes separately, controlling for location and presentation duration, and guaranteeing that each eye could not see the other eye's stimuli. I carried out a series of experiments, based on this novel paradigm, to explore the effects of a vertically split fovea on correctness of word perception. The experiments showed: (i) words presented exclusively to the contralateral hemifoveas are more correctly reported than words presented exclusively to the ipsilateral hemifoveas; (ii) the same full word shown to both eyes and available for fusion led to better perception; (iii) word endings with fewer type-count neighbours were more accurately reported, as were beginnings with larger type-count neighbours; (iv) uncrossed-eye stumuli were better perceived than crossed-eye stimuli; (v) principled roles in a model of isolated word recognition for lexical and sublexical neighbourhood statistics, syllabicity, hemispheric fine- and coarse-coding differences, sex of the reader, handedness, left and right eye, and visual pathways. Finally, I propose a connectionist model of visual word recognition that incorporates these findings and is a basis for further exploration.

This thesis project was undertaken with the intent to discover the source of a known but hitherto unexplained error in the calibration of the wings for a haploscope used in depth perception studies. The angles of the haploscope wings are used to control the vergence angle of the virtual images projected into each eye. This accounts for a strong depth cue used in AR and depth perception studies. Two experiments were devised to both display and attempt to characterize the error between the theoretical wing angles needed to cause a user's vision to verge at some focal depth and the actual wing angles that caused vergence. The investigation revealed a near-constant offset between the theoretical and actual angles needed. This suggests that the error may not stem from the haploscope alignment itself, but from how the center of the user's eye is currently modelled.