Characterizing the cognitive computations underlying spatial navigation

点击翻译按钮获取中文摘要

PROJECT SUMMARY Deficits in spatial navigation are associated with a number of mental disorders including anxiety, autism, depression, and schizophrenia. Previous work to investigate these deficits has relied on relatively coarse behavioral measures, such as error rates and completion times, whose relation to the underlying neural computations is unclear. Moreover, by using desktop Virtual Reality (VR) tasks, in which participants navigate a virtual world while sitting at a computer, many spatial navigation experiments ignore one of the defining features of spatial navigation – that it involves movement of the body through space. Because of these limitations in measurement and ecological validity, the computational mechanisms underlying spatial navigation deficits in mental illness are poorly understood. The objective of this proposal is to develop and test new computational models and behavioral paradigms that can pick apart the cognitive computations underlying spatial navigation. Our central hypothesis is that people perform spatial navigation tasks using a mixture of two interacting processes: Path Integration, based on integrating body-based cues about one’s own rotations and translations (e.g., sensory inputs about the movement of one’s limbs), and Landmark Navigation, based on processing environmental cues (e.g. visual landmarks). Work in this grant will develop computational models of both processes, including how estimates from the two processes are combined. To test these models we make use of a new technology known as immersive VR. In immersive VR, a virtual environment is rendered on a head-mounted display while participants walk freely in a real room. Thus, participants experience identical body-based cues to real navigation but with visual cues that are under experimental control. Using immersive VR, we will create two behavioral tasks in which participants either turn (Rotation Task) or walk (Translation Task) to a remembered heading or location in the dark. In the No Feedback condition (Aim 1), participants will complete these movements with no visual feedback, allowing us to quantify Path Integration. In the Feedback condition (Aim 2), participants will receive a brief flash (300 ms) of visual information that is potentially offset relative to the true location, allowing us to test how they combine Path Integration with Landmark Navigation from visual cues. Finally, in both experiments, we will test whether performance is predictive of more general spatial navigation ability using measures of real-world navigation ability (Aim 3). The innovation in our proposal lies in both our experiments, which leverage new immersive VR technology to induce mismatch between Path Integration and Landmark Navigation systems, and in our models, which provide precise, concise, and quantitative descriptions of behavior. Beyond the foundational work in this grant establishing the paradigms and models, our experiments are readily translatable to patients and animals, allowing for insight into the clinical implications and neural mechanisms of deficits in spatial navigation.