In natural vision, retinal stimulation is ever-changing due to saccades and observer motion, but our percept of the world remains remarkably stable. How this is achieved is a major unsolved mystery in systems neuroscience. Prior neurophysiological and psychophysical studies have identified three hypothetical encoding strategies that may contribute to visual perceptual stability in the primate brain. First, neuronal representations may carry a prospective code of what is expected to occupy the future receptive fields of neurons even before our eyes land on the saccade target. Second, salient visual objects may be represented in terms of a spatiotopic code, i.e. in world-coordinates (in addition to retinal coordinates), which could provide a perceptual anchor even as the retinal representations shift. Third, neuronal representations may be hysteretic, with responses depending not just on the current retinal image but also on the recent past. These three strategies may operate in parallel across a network of cortical and subcortical brain regions to achieve stability. In this grant cycle, we will investigate the contributions of area V4, a midlevel processing stage in the ventral visual stream that is critical for form encoding and scene perception and is heavily interconnected with parietal, frontal and subcortical brain regions. Using a combination of fixation and naturalistic behavioral paradigms, immersive visual displays of objects, textures and scenes, and recordings with high-density Neuropixels probes in the awake macaque monkey, we will investigate whether: (Aim 1) V4 neurons carry prospective signals that are comprehensive in terms of feature selectivity, spatial coverage and precision, to provide complete and precise trans-saccadic integration, (Aim 2) visual object representations in V4 are modulated by gaze position akin to parietal gain fields, thereby providing an implicit representation in a spatiotopic reference frame at the population level, and, (Aim 3) V4 responses to continuously changing stimuli (mimicking self- motion) are hysteretic, i.e. responses change more slowly than the stimulus itself, thereby providing a neuronal basis for stable perception during observer motion. Our results will: (i) support the development of new models of midlevel cortical representations as a function of extraretinal inputs and current, future and past retinal stimulation, (ii) delineate a framework of representational ebb and flow as animals saccade from one location to the next in the visual world, (iii) provide insights into the neurophysiological basis for myriad psychophysical results that have revealed how visual stimuli are integrated, what frames of reference (retinotopic vs spatiotopic) are relevant and whether and how stimulus history contributes to object invariance and continuity across saccades. Parallel experiments in area V2 will provide insights into how V4 representations may be built. More broadly, the encoding models developed ...