Project Summary To optimally estimate properties of the environment, one should use all available sensory and prior infor- mation. Sensory input arises from multiple sensory modalities, and is uncertain due to physical and neural noise. An ideal observer must combine all cues, taking into account cue reliability and considering alternative causes for discrepancies between cues such as inferring that they derive from different objects (i.e., separate causes) or that one or both sense modalities are miscalibrated. Human behavior is often consistent with “opti- mal” cue integration defined as maximizing combined-cue reliability. Can observers access internal estimates of reliability for confidence judgments to inform subsequent behavior? What computations does the brain use to solve these problems and how do they unfold over time? We tackle these questions by combining computational theory and behavioral experiments to elucidate the dynamics and time course of multisensory cue integration. The standard framework for understanding sensory integration incorporates five key components: (1) reliability-based cue weighting, (2) integration of prior infor- mation, (3) causal inference, (4) cross-modal recalibration and (5) confidence in the sensory estimates. Past work on multisensory cue integration has primarily dealt with situations that are static, but the world is dynamic and ever-changing, many activities (locomotion, driving, sports, etc.) depend critically on combining information over time and on temporal estimates, and sensory cue integration unfolds over time. In Aim 1, we will develop and test models of multisensory temporal perception, both for relative timing and perception of duration as well as understanding how spatial and temporal correlation jointly contribute to causal inference. In Aim 2 we look at sensory-motor integration from a similar perspective: the planning and execution of movement also involves multisensory integration prior to, during, and after a movement, both for movement itself and for judgments about movement success. Finally, in Aim 3 we look at the use of multiple visual and other sensory cues for perception of heading and object motion from optic flow and other cues. We will study whether the five compo- nents of multisensory integration help us to better understand the analysis of optic flow over time, and integrat- ing optic flow with natural self-motion in a virtual environment, for the estimation of heading and detection of moving objects. These studies will shed light on the dynamic processes used to combine sensory information to form a co- herent percept, the unfolding of sensory integration as perceptual decisions are made, and how our senses guide us in an ever-changing world. A better understanding of the dynamics of these processes will inform fu- ture studies on the neural implementation of these computations. These experiments on healthy humans will provide a starting point for understanding multisens...