20.1  Introduction

In order to assess the effectiveness of prosthetic vision in context, we must ask the question “Effectiveness for what”? Answering this question requires an understanding of the realistic goals for this new technology. The prospect of “restoring vision to the blind” (e.g., [40]) has, of course, been received with great enthusiasm. However, the richness of our visual experience belies the complexity of the neural system delivering it, thus making it unrealistic for a prosthetic device to provide vision in its full and richly complex form. For instance, the number of electrodes in current devices (16–1,500) is many, many orders of magnitude fewer than required to carry the wealth of the information from the 120 million photoreceptors in each eye along over a million fibers of each optic nerve to the 140 million highly organized neurons in each hemisphere of primary visual cortex, which in turn send them on to the many other regions of the cortex devoted to specific aspects of visual processing.

The multitude of neurons form a complex network with feed-forward, lateral and feed-back signaling giving the visual system its complex imaging power. Subsequent to loss of photoreceptors in outer retinal diseases such as retinitis pigmentosa, there are significant losses in both the inner nuclear layer (e.g., bipolar cells) and ganglion cells [50, 81, 91]. Furthermore, there is significant remodeling of neural retina and thus local neural networks following photoreceptor loss [64, 65]. Thus, whether implanted in the retina or the cortex, the implants will not have the full analytic power of the intact visual system. Currently, prostheses developers are working to increase the number of electrodes and their density to improve resolution (e.g., 1,024 by Troyk’s group [98]). Performance has been shown to improve with increasing number of electrodes [30, 42].1 However, no matter how many electrodes in use, multiple phosphenes (discrete sensations of light), of various sizes shapes and hues, depending on the state of the post-receptoral retina and visual system, will be generated. How will these phosphenes be used to represent the environment? Past visual experience will certainly influence interpretation, but the recipient will have to undergo prolonged training to learn to use these signals.

1 The success of cochlear implants is often cited as a hopeful indication of what can be accomplished by a sensory prosthesis. Only six electrodes stimulating the cells of the auditory nerve enable the wearer to understand speech at near-normal levels. If one assumes the same ratio of electrodes to nerves, hundreds of electrodes are projected to be required [108].

Thus, prosthesis recipients may have to make sense of a “blooming, buzzing confusion” that William James (1842–1910) said faced a newborn.

Bearing these considerations in mind, it becomes clear that “restoring vision to the blind”, is an amorphous goal that is unreachable in the foreseeable future. In fact, the goal of prostheses is not to recreate normal vision, but to provide visual perception that however limited in scope is useful to the individual [69, 70, 78, 108].

How can we determine whether prosthesis aided or provided vision is useful to the individual? This requires well-defined, measurable goals for evaluation of progress and demonstration of successful outcomes. Widely accepted outcome measures and means of measuring them, and criteria for “success” have not been developed for prosthesis implantation. Standards have not been set. The fact that clinical trials are already in progress (e.g., SecondSight, Intelligent Medical Implants are conducting Phase 2 Clinical Trial) and others are planned makes more pressing the establishment of relevant outcome measures as criteria for success. Once such outcome measures are selected and developed, the means to assess performance with respect to these outcomes are specified, and testing has been carried out, progress can be assessed. The lack of well-specified relevant outcomes and means to assess them is a major hurdle in the future of prostheses.2

For most of us, activities of daily life rely on vision. Visual performance (actions incorporating and guided by visual input) is a very complex phenomenon involving other senses, motor skills, memory, prior knowledge, feedback, experience, practice, etc. The basic building blocks for vision performance are sensory visual functions such as motion, color, luminance, contrast, and orientation. These contribute to object identification and localization. These in turn feed into higher order visual areas that integrate the visual information with other senses, motor systems, cognitive systems, and memory the ensemble of which guides the actions that form the tasks of daily living. In assessing prosthetic vision, we should ideally find ways of measuring these functions and task performance.

For simplicity, we restrict the discussion that follows to consideration of individuals with particular characteristics. Most prosthesis recipients to date have lost vision as a result of retinitis pigmentosa (e.g., [21, 22, 30, 51, 105, 118]). Therefore, we assume a target population of individuals with long-standing retinitis pigmentosa (RP), a progressive disease of the photoreceptors. Nearly all RP patients have some residual vision, typically one or more small peripheral islands in an otherwise non-functional retina [22].3 Since these individuals have had some vision well into

2 Lengthy discussions of these issues took place at a special interest group meeting at ARVO 2007 (organized by author MES and contributed to by author GD) [82] and a symposium hosted by the Smith-Kettlewell Eye Research Institute in San Francisco (October 2007) [83]. These meetings highlighted the complexity of the problems and demonstrated that more work needs to be done before specific recommendations of protocols and tests will be established. Some of the content of this chapter has been gleaned from those meetings.

3 The degree of residual vision function is highly light-level dependent in RP patients; individuals with retinal disease often require unusually high light levels to attain their best vision function [85, 92].