Scientists already use machine learning to create models of fMRI scans of the visual cortex during visual perception and imagination. My impression is that these models can work pretty well for multiple scans from the same person, but nowhere near so well between people. Interestingly in a recent paper, it was found that the accuracy of the predictions of what was seen or imagined (meaning accuracy above chance) was about the same for visualizers and aphants when the models were applied only to perception or only to imagination. However, the cross-task accuracy (visualization vs perception) was better than chance only for the visualizers. For aphants, the cross-task accuracy was no better than chance. (See https://doi.org/10.1016/j.cub.2024.12.012) The visual images used in this study were very simple, but there still are some implicatons. Just because you do not experience any visualization when you try to imagine something doesn't mean that your visual cortex is not doing anything. So maybe it could be possible to read someone's mind, even if they were aphantasic, if you had a library of fMRI scans taken when they were imagining different types of things. Maybe not today (who is going to sit in an fMRI for this purpose anyway???), but who knows what the future may bring.

Scientists already use machine learning to create models of fMRI scans of the visual cortex during visual perception and imagination. My impression is that these models can work pretty well for multiple scans from the same person, but nowhere near so well between people. Interestingly in a recent paper, it was found that the accuracy of the predictions of what was seen or imagined (meaning accuracy above chance) was about the same for visualizers and aphants when the models were applied only to perception or only to imagination. However, the cross-task accuracy (visualization vs perception) was better than chance only for the visualizers. For aphants, the cross-task accuracy was no better than chance. (See https://doi.org/10.1016/j.cub.2024.12.012) The visual images used in this study were very simple, but there still are some implicatons. Just because you do not experience any visualization when you try to imagine something doesn't mean that your visual cortex is not doing anything. So maybe it could be possible to read someone's mind, even if they were aphantasic, if you had a library of fMRI scans taken when they were imagining different types of things. Maybe not today (who is going to sit in an fMRI for this purpose anyway???), but who knows what the future may bring.