6.4. Discussion 6 107 with visual prostheses. For more realistic evaluation of functional value of prosthetic vision, it is therefore crucial to include eye-tracking in the experimental setup, similar to, for instance, (Dagnelie et al., 2006; McIntosh et al., 2013; Paraskevoudi & Pezaris, 2021; Sommerhalder et al., 2004; Titchener et al., 2018; Vurro et al., 2014). 6.4.4. Learning effects Importantly, the use of prosthetic vision, and especially gaze-locked vision requires intensive training (Sabbah et al., 2014). In both experiments, most of the participants reported that they had sufficient training with the simulation conditions during the practice trials. Indeed, we did not find a significant training effect in the primary experimental outcomes, with one exception: the reduced trial duration in Experiment 1 indicated improved mobility performance over the course of the experiment. We found no interaction between the different study conditions. Although previous suggested that gaze-contingent viewing can be adopted with relatively minimal effort Paraskevoudi and Pezaris (2021),wedid not find any differences in the learning abilities with the different conditions. 6.4.5. Strategies The participants did not receive any specific instructions for coping with the different gaze conditions. The reduced gaze-velocity in the gaze-locked compared to the gazecontingent study condition indicates that subjects intuitively learned to suppress eyemovements with gaze-locked vision. Indeed, the majority of subjects indicated in the exit interview of Experiment 2 that they tried to keep their eyes still. Note that most of the subjects that tried to actively suppress eye-movements experienced difficulties in succeeding so. These responses are similar to the experiences described by users of head-steered prostheses (Caspi et al., 2018; Sabbah et al., 2014), suggesting fundamental challenges with compensation training based on eye-movement suppression. Some complicating factors are reflexive eye-movements (e.g., vestibulo-ocular reflex) which remain difficult to suppress. When asked about other strategies, more that half of the participants in both experiments indicated that they used frequent or exaggerated head movements. Interestingly, the average head velocity in Experiment 2 was significantly higher with gaze-contingent vision compared to the gaze-locked condition. This finding was unexpected, as the reestablished visual scanning with eye-movements would alleviate the need for head movements. We speculate that the reduced rotational head velocity with gaze-locked vision partly reflects uncertain scanning behaviour caused by perceptual instability. One participant indicated in the free written responses that they ‘decreased the exaggerated head movements due to nausea’. In both experiments, the side effects such as disorientation and nausea were reportedly lower with the gaze-contingent versus the gaze-locked vision. Noteworthy, there was a large variety in self-reported strategies for performing the experimental tasks. Incontestably, besides the functional performance, it is important to consider these subjective measures of convenience and (dis)comfort for the development of visual prostheses, accounting for personal differences in preference. 6.4.6. Limitations and future directions There are several limitations to the current study. One practical limitation is the indirect user control in the mobility experiment that was used for moving through the virtual
RkJQdWJsaXNoZXIy MTk4NDMw