UC San Diego

People vary greatly in how they experience their thoughts. Some say they think in vivid mental images, whereas others say they think in words. These internal representations play a central role in cognition, supporting memory, language, planning, and imagination. This raises the possibility that diversity in how people internally represent the world produces diversity in how cognition unfolds across minds. In this dissertation, I examine this possibility across three cases: when internal representations are maintained from visual input, constructed from language, or generated with little external constraint on their content.First, I examine visual working memory, when people hold visual information in mind across a short delay. Using a task with naturalistic images, I show that people with different imagery profiles may rely on different representational routes even when performing the same task with the same stimuli. These routes differ in what kind of information they preserve, how well they support simultaneous maintenance of multiple items, and whether similar visual input during maintenance interferes with or reinforces the memory trace. Second, I turn to language comprehension, when people internally represent meaning conveyed through a non-visual medium. Embodied theories of meaning suggest that understanding language can involve recruiting sensory and motor systems. In a task previously used to index perceptual simulation during comprehension, I find that imagery vividness does not reliably modulate overt behavior. Neural responses, however, suggest that similar behavior may reflect different processing routes. Participants across the imagery spectrum show sensitivity to the task manipulation in a later window linked to meaning processing, whereas stronger visual imagers also show sensitivity in an earlier window linked to visual-perceptual processing. These findings suggest that the extent to which language comprehension recruits perceptual simulation may depend on imagery vividness.Third, I explore internally generated visual experience through Ganzflicker-induced hallucinations, where visual content can emerge freely rather than being specified by a particular image or sentence. Analyzing free-text descriptions from over four thousand participants, I identify systematic differences in the kinds of content people report experiencing across the imagery spectrum, from simpler visual phenomena such as colors, flashes, and geometric patterns to complex objects, faces, scenes, and richer sensorimotor content.Finally, I synthesize these findings into a broader framework for imagery research. I argue that mixed findings in the literature may reflect bottlenecks in how imagery is defined, measured, and linked to cognitive tasks, and propose some concrete ways forward. Together, these studies suggest that imagery diversity is not merely a curiosity about private experience. It is a tool for revealing how different minds can reach similar cognitive outcomes through different internal routes.