A neural model of conscious mental imagery and aphantasia

Mental imagery refers to one's quasi-perceptual experience in the absence of direct external input. Yet around 4% of individuals with "aphantasia" report an inability to voluntarily generate such experiences. This phenomenon provides a natural experiment that challenges current theories of imagery, which typically assume that reactivation of the visual cortex in imagery generation is sufficient for experience. In fact, aphantasics can reactivate the visual cortex during imagery tasks, suggesting that additional processes are required for conscious imagery. In this theory paper, I propose a neural model of conscious imagery, specifying multiple stages of processing, in which generation provides initial sensory reactivations, integration binds visual features into coherent perceptual-like objects or scenes, and amplification enhances them for conscious access. Attention appears to be necessary, though not sufficient, for conscious imagery. This architecture of processing is supported by a fronto-parietal-fusiform network shaped by two interacting attention systems. Aphantasia primarily reflects deficits in top-down modulation, with preserved generation but impaired integration and amplification of internal representations, linked to altered interactions between frontoparietal networks and the fusiform gyrus. Overall, this model reframes imagery as an active, constructive process shaped by the dynamic interplay of higher-order networks, including attentional, intentional, and memory-control systems, rather than a passive reactivation of sensory representation. It offers a testable neuromechanistic account bridging research on imagery and consciousness, motivating future research on unconscious and conscious processing of internal representations.