To create a coherent view of any scene, the visual system must determine how inputs from the scene relate to one another and link them together to ultimately produce percepts of coherent objects separate from other objects and the background. This dissertation aimed to determine how the visual system links spatially separate and neurally ambiguous regions of the visual field. To this end, experiments here used multiple stimuli that create neural ambiguity presented in spatially separated locations within the visual field. Observers reported when the multiple stimuli resulted in “grouped” percepts. Grouped percepts occurred when all of the multiple ambiguous stimuli in view appeared the same or similar in certain features above chance levels. The experiments here made use of multiple methods to create neural ambiguity and used color perception as a model system to measure grouping under different conditions. This allowed testing of different neural theories about how the visual system segments the world. The overarching hypothesis here was that a grouping process acts on low-level representations of visual input and the competition between these low-level representations determines if two or more regions are grouped. Experiment 1 examined whether grouping is driven by competition between percept-level representations (i.e., representations of the possible percepts, regardless of what neural processes gave rise to these representations) or if grouping is limited by different neural levels of competition in the multiple locations. Experiment 2 measured grouping of a novel percept, the superposition of two orthogonal gratings creating a plaid, in order to determine if grouping can act on neural representations of binocularly-integrated components. Experiments 3a and 3b eliminated the possibility that an inhibition of interocular suppression drives the perception of plaids. Experiment 4 used specific physical chromaticities within chromatic surrounds to determine if grouping is driven by a shared color percept or a shared physical chromaticity. Experiment 5 followed up the results from Exp. 4 and tested if grouping persists when there is no shared chromatic information and if grouping can be driven by one chromatic signal (L/(L+M), for instance) in the presence of a conflicting chromatic signal (S/(L+M), for instance). Experiment 6 tested if both of the two competing neural representations must match exactly, or if grouping can be driven by a single shared competitive neural representation paired with unshared neural representations.