Are we aware of neural activity in imagination? The attention model of conscious imagery and aphantasia

Dr. Jianghao Liu is a cognitive neuroscientist at the Paris Brain Institute, where he studies the neural mechanisms underlying mental imagery, aphantasia, attention, and consciousness. He recently co-founded, together with Christian O. Scholz, the Interdisciplinary Reading Club on Aphantasia (IRCA; https://jianghao-liu.github.io/irca/).

A post by Jianghao Liu

Please imagine a red Gala apple. Is it darker or lighter than a cherry? Most of us can easily conjure a mental image, a perceptual-like experience that feels almost as if we were seeing it. For centuries, mental imagery has long been considered a core component of consciousness in both philosophy of mind and neuroscience (Nanay, 2023). Yet, emerging evidence from individuals of aphantasia, who report no voluntary imagery (Zeman, 2024), raises a striking question: can there be unconscious mental imagery (Nanay, 2021; Michel et al., 2024)? In this post, I present the attention model (Liu, 2025) of the neural basis of conscious imagery and aphantasia, thereby aiming to bridge empirical neuroscience and philosophical accounts of imagination and awareness.

Traditional empiricists such as Hume viewed imagination as composed of faint “copies” of sensory impressions or mental pictures derived from experience. Following this view, Kosslyn (1980) famously proposed that quasi-perceptual imagery arises when higher-level representations reactivate a “visual buffer” (a kind of internal screen) in the primary visual cortex. This perspective has profoundly influenced subsequent theories of imagery, which typically conceive imagination as vision in reverse (Pearson, 2019). But is this sensory reactivation enough to produce the subjective feeling of imagery? Aphantasia provides a natural experiment in which imagery experience is entirely absent or severely diminished. In this post, I explore how this condition reshapes our understanding of the neural and phenomenological basis of imagery.

In our studies, we first asked individuals with congenital aphantasia to compare different visual properties of objects including colours, shapes, faces, written letters, even a map of France (Liu & Bartolomeo, 2023). These participants consistently denied any visual experience, yet they could accurately recall visual details from memory—a paradox revealing preserved access to visual knowledge without phenomenology. Using high-resolution 7T fMRI, we observed specific reactivation in higher visual areas when aphantasic participants attempted to imagine—the very same regions active when they actually saw those images, e.g., areas activated when participants saw pictures of faces (Liu et al., 2025). This preserved activity may explain their access to visual knowledge despite lacking visual experience, a phenomenon that cannot be fully explained by nonvisual strategies such as verbal labels or sensorimotor coding (Reeder et al., 2024).

However, a closer look revealed two key neural differences in aphantasia (Liu et al., 2025). First, in aphantasics, neural patterns in the visual cortex during imagery diverged from those during perception. In contrast, stronger pattern similarity between imagery and perceptual representations predicted higher vividness in typical individuals. This absence of representational overlap challenges the notion of unconscious mental imagery, as it suggests a lack of perceptual-like activation that is necessary for genuine imagery (Scholz et al., 2025). Second, aphantasic individuals showed markedly reduced communication (i.e., long-range functional connectivity) between the visual cortex and a left prefrontal region known to regulate the threshold of conscious access and metacognition (Liu & Bartolomeo, 2025). Together, these findings suggest that people with aphantasia may form unconscious visual representations that fail to cross the threshold of awareness.

If mere sensory reactivation in imagery generation is insufficient, what additional processes allow imagery to become conscious? The attention model (Liu, 2025) offers the first attempt to establish a neuromechanistic framework for how conscious imagery arises from unconscious sensory reactivation. This model outlines a hierarchical architecture across three successive stages: (i) generation, where top-down signals initiate weak reactivations in sensory areas; (ii) integration, where the visual cortex binds conceptual knowledge with visual features to assemble coherent preconscious perceptual-like content; and (iii) amplification, where top-down attention and prefrontal recurrence boost this content into global awareness. In short, after initial reactivation, fragmented sensory traces are locally combined in the visual cortex into preconscious images, which are then amplified through higher attention systems, thus crossing the threshold into awareness. This process unfolds through a frontoparietal-visual imagery network (Liu, 2025), shaped by interacting dorsal and ventral attention networks that guide information from unconscious processing to conscious experience (Liu et al., 2023).

Crucially, the attention model predicts subtypes of aphantasia, each reflecting a specific breakdown in one or multiple stages before conscious imagery. Returning to our “red apple” example, we can now distinguish “knowing about an object” from “feeling it in the mind” (Figure 1). On this model, congenital aphantasia reflects intact generation but disrupted integration or amplification, due to weakened top-down modulation associated with attention networks. In contrast, acquired aphantasia (from brain injury) often involves the loss of visual knowledge itself, suggesting damage to brain areas associated with the imagery generation stage (Bartolomeo, 2002; Liu et al., 2022; Thorudottir et al., 2020). More provocatively, the model suggests we might temporarily induce aphantasia in typical individuals by interrupting integration or blocking access to internally generated content, a kind of induced aphantasia (Liu, 2025). Testing such predictions using specific tasks or neuromodulation techniques could open a new empirical path toward the neural mechanisms of imagination and consciousness.

For decades, both philosophers and neuroscientists studying consciousness have focused mainly on perceptual awareness, often overlooking inner phenomena such as imagination. This neglect stems from the assumption that mental imagery is necessarily conscious—an idea now challenged by aphantasia. The attention model provides a bridge between the fields of imagery and consciousness, offering testable accounts of both conscious and unconscious internal representations, and inviting renewed dialogue between empirical science and phenomenology. The phenomenon of aphantasia further illuminates this complex and largely unexplored landscape, with the potential to reveal the invisible scaffolding that underlies consciousness of internal representations. As Sartre (1940) argued in L’Imaginaire, imagery is not a “picture” inside the mind but an intentional act directed toward the absent in consciousness. The attention model offers a neural complement to this insight by explaining how intentional direction might fail when neural amplification breaks down. Studying unconscious imagery may therefore uncover a hidden dimension of consciousness—one that turns inward, revealing how the mind perceives its own thoughts and exposing a phenomenological structure fundamentally distinct from perception.