The perception and production of speed during self-motion: evidence for non-optimal compensation mechanisms

The thesis describes a series of studies on the perception of speed of self-motion in realistically simulated environments. We carried out several experiments in order to explore from both the behavioral and the perceptual point of view how visual changes in the environment lead to a misperception of self-speed. More specifically, we investigated the issues of image contrast reduction, gaze direction and field-of-view and their influence on the perceived and produced speed of self-motion. Virtual environments technology was employed to display visual motion for both driving and walking speeds. Given the first results obtained from the implementation of a realistic contrast reduction in a driving simulation we formulate the hypothesis that changes in the perceived and produced speed are due to a non-optimal combination of the retinal angular velocities during self-motion. Indeed, the retinal projection of the environment during forward self-motion consists of expanding optic-flow patterns in which the angular velocities of the objects on the scene vary gradually according to their position and speed relatively to the moving observer. Objects very far away appear as moving slowly in the centre of the field-of-view, while the proximal regions appear as moving faster at the periphery of the visual field. The second part of the thesis is then dedicated to a systematic investigation on how the angular velocities from central and peripheral regions of the field-of-view contribute to the generation of the percept of a unique speed of forward translation. In Chapter 1 we present a brief overview of the principal aspects of visual motion processing. We report some psychophysical and physiological findings that have been produced in the last decades of research, together with some models that try to describe the functional aspects of motion and speed processing. Moreover, we introduce and explain the theme of the visual speed of self-motion. Finally, we describe the technical and methodological tools that have been adopted to implement and execute the experiments of the following chapters. In Chapter 2 experiments are described in which we address the question how image contrast reduction affects the driving behavior. We put particular emphasis to a naturalistic implementation of contrast reduction, namely fog. We show that a realistically simulated fog causes drivers to reduce the driving speed. We provide experimental evidence that this effect has a perceptual origin, due to the increased perceived speed while driving in fog. The behavioral effect of simulated fog is consistent through different experimental setups. Moreover, we show that the effect of an increased perceived speed is enhanced accordingly to the fog density. We hypothesize an explanation for these results based on the fact that the motion signals within the visual field are selectively masked. In fact, the three-dimensional spatial distribution of the exponential fog model used in our experiments reduces mainly the visibility of the lower velocity signals, situated in the distal region of the environment, centrally in the visual field. Furthermore, we compare the effect of simulated fog to the effect of a type of contrast reduction that attenuates the visibility uniformly all over the visual field, independently of the three-dimensional structure of the environment. We demonstrate that the perception of both the driving and walking speed is independent of a spatially uniform contrast reduction and of the region within the visual field in which this reduction is applied (central or peripheral). The experiments of Chapter 3 show that during self-motion the estimation of the self-speed relies on retinal angular velocities that are likely to be combined in a non-optimal solution. We provide evidence that the speed estimate is biased towards the available motion signals from the environment, both in the driving and in the walking speeds domain. When the visibility of the central region of the field-of-view is precluded the perceived speed is higher, and conversely, when the peripheral region is not visible, the perceived speed is lower. This suggests that the speed estimation process takes into account the velocity signals from both central and peripheral areas of the visual field. However, we provide also evidence that the central area is necessary and nearly sufficient to build a correct speed estimate, even in large field-of-view virtual environments. Finally, we report also that the speed estimate depends on the gaze direction and can be impaired when the vertical field-of-view is limited. Summary of the results and general discussion are presented in Chapter 4.