2 20 2. Real-world indoor mobility with simulated prosthetic vision studies. Another factor that may have limited the performance of our participants compared with previous studies, is found in the simulation of the phosphenes. In the current study, phosphenes could take binary states (‘on’ or ‘off’), where Dagnelie et al. (2007) andSrivastava et al. (2009) used 8 or 4 levels of grayscale intensities, respectively. Although some relationship between stimulation parameters and phosphene size has been established (Brindley & Lewin, 1968), at this time cortical visual prostheses do not allow for systematic control over phosphene brightness (Najarpour Foroushani et al., 2018; Troyk et al., 2003). The current simulation with binary phosphenes therefore provides a valuable addition to previous literature that do not implement this constraint. Note that the field is developing rapidly and results from further clinical work can guide SPV research for the development of realistic phosphene simulations, which, vice-versa, can accelerate clinical developments by answering fundamental questions about prosthetic design(Najarpour Foroushani et al., 2018). The curves inFigure2.5suggest that – maybe unsurprisingly – even at higher phosphene resolutions there remains a gap between SPV and normal vision. This implies that besides increasing the number of electrodes there are other challenges to be taken before prosthetic vision approaches the quality of normal sight. Even with the current technological prospects, there are many design choices that influence the utility. For example, in experiments with SPV, Cha et al. demonstrated that, in line with other low-vision research (Marron & Bailey, 1982), the distribution of simulated phosphenes across the visual field can have an impact on mobility (Cha et al., 1992b). Future studies with SPV could further explore the impact of using different electrode locations in the visual cortex on the mobility performance. 2.4.2. The effect of visual complexity Our results demonstrate that scene simplification via removal of background textures and within-object gradients may improve mobility performance at lower phosphene resolutions. In the higher phosphene resolutions this effect was absent or even opposite, indicating an interaction between phosphene resolution and visual complexity. On the one hand, these findings confirm previous suggestions that low-resolution prosthetic vision quickly gets overcrowded (Vergnieux et al., 2017). Post-hoc inspection of the simulated prosthetic percept, as well as the responses on the exit interview revealed that overabundant phosphene activity renders it almost impossible to distinguish the floor, walls and objects. In other words, at low phosphene resolutions visual complexity comes at the cost of interpretability. At the same time, excessive removal of visual information at higher phosphene resolutions may negatively influence the visual processing abilities that are required for mobility. Optic flow processing, for instance, which depends on dynamical tracking of local visual patterns, is an important requirement for the estimation of ego-motion and heading (Lappe et al., 1999; Warren et al., 2001). Especially in low-vision conditions, and even phosphene vision, the removal of optical flow cues may have a negative impact on the recognition of scene structure and foreground objects (Pan & Bingham, 2013; Qiu et al., 2018). Besides serving as dynamical cues for optic flow perception, surface information and background textures may have also directly contributed to the detection of foreground objects, by facilitating figure-ground segregation (Caputo, 1996; Machilsen & Wagemans, 2011). Note that these specific interpretations regarding the underlying visual processing remain somewhat speculative, since downstream visual
RkJQdWJsaXNoZXIy MTk4NDMw