The development of retinal implants has become an important field of study in recent years, with increasing numbers of people falling victim to legal or physical blindness as a result of retinal damage. Important weaknesses in current retinal implants include a lack of the resolution necessary to give a patient a viable level of visual acuity, question marks over the amount of power and energy required to deliver adequate stimulation, and the removal of eye movements from the analysis of the visual scene. This thesis documents investigations by the author into a new CMOS stimulation and imaging chip with the potential to overcome these difficulties. An overview is given of the testing and characterisation of the componments incorporated in the device to mimic the normal functioning of the human retina. Its application to in situ experimental studies of frog retina is also described, as well as how the data gathered from these experiments enables the optimisation of the geometry of the electrode array through which the device will interface with the retina. Such optimisation is important as the deposit of excess electrical charge and energy can lead to detrimental medical side effects. Avoidance of such side effects is crucial to the realisation of the next generation of retinal implants.