been working through the scholz et al. (2026) paper on what they’re calling the integration model of aphantasia. feels like a useful consolidation of a field that’s been spinning its wheels on two unsatisfying camps for a while.
three theories, one problem
generation failure: ventral-dorsal misconnection preventing imagery generation. clean mechanistically but doesn’t explain preserved performance on tasks that should require imagery.
unconscious imagery: the “black box” hypothesis. scholz et al. push back effectively — fmri data doesn’t show shared representational geometry between aphantasics and controls, and there’s no binocular rivalry priming effect. strong strike against the unconscious account.
integration/attention/interoception: the paper’s main contribution. unifies the attention stream (dorsal, AIC-coordinated: generation → integration → amplification) and the interoception stream (ventral, PIC-coordinated: embodied simulation drawing on interoceptive input). this is the part i wasn’t expecting — it implicates the body, not just visual cortex, in the substrate of mental imagery. the interoceptive route isn’t decorative; it’s load-bearing in the model. nagai’s 2025 paper “interoception predicts mental imagery vividness” is the obvious prior work here and scholz et al. lean on it — the finding that interoceptive sensitivity correlates with imagery vividness is what motivates routing the ventral stream through the PIC in the first place. it reframes the whole thing: if your interoceptive signal is weak or poorly integrated, your imagery may be dim not because of a visual processing deficit but because the body-state input that helps construct and stabilize the image is missing. that’s a genuinely different kind of explanation than anything in the generation failure or unconscious imagery camps.
sensory precursors
rudimentary preconscious codes that exist prior to full imagery generation. explains how aphantasics answer “are strawberries red?” without visualizing a strawberry. reframes aphantasia less as an absence and more as a truncation — the process starts but doesn’t amplify into awareness. neuromechanistic framing anchored in global neuronal workspace theory (GNWT): imagery failure as a broadcast failure rather than a generation failure.
meeting with dr. monzel — evolutionary framing
the core premise: mental imagery was almost certainly adaptive for early humans. aphantasia would have been a meaningful fitness disadvantage — largely self-erasing through natural selection. the wrinkle is the cultural threshold: once humans started outsourcing imagination onto physical media (cave walls, writing, everything else), selective pressure on internal imagery weakens. modern aphantasia is downstream of that shift.
the “superpower” framing: aphantasics may approach creative work without interference of preconceived mental images — seeing things “for the first time every time.” testable-ish claim worth including in the framing section. next step is a lit review on the exographic memory side — extended mind literature could scaffold the evolutionary transition argument.