Unconscious imagery in aphantasia? Spoiler: we still don’t know

Dr Andrea Blomkvist
andrea.blomkvist@glasgow.ac.uk
Centre for the Study of Perceptual Experience
University of Glasgow

One of the most imagination-capturing concepts discussed in Bence Nanay’s recent book Mental Imagery (Nanay, 2024) is unconscious mental imagery. For many of us, mental imagery seems inherently phenomenally conscious (see for example Kung, 2010). But in Nanay’s book, mental imagery is defined as perceptual representation that is not directly triggered by sensory input (Nanay, 2024: 4). From this definition, it follows that we do not necessarily have a phenomenally conscious experience of mental imagery. Mental imagery can remain unconscious, and our perceptual and cognitive systems can still access the representation for computations; imagery can be accessible but unconscious (see Block, 2011 for a discussion on access vs. phenomenal consciousness).

An intriguing suggestion from Nanay is that some people with aphantasia could be characterised as having such unconscious, but accessible, imagery. Aphantasia is commonly defined as a condition where the vividness of mental imagery is reduced or completely absent (Zeman et al., 2015). But note that this definition is put in terms of vividness, a concept relating to the phenomenal experience of imagery. Despite not reporting any vivid mental imagery, Nanay suggest that people with aphantasia could have unconscious imagery.

Now, at the time of the publication of the book, little research had been devoted to studying the possibility of unconscious imagery in aphantasia. Nanay presents us with one case of a person with aphantasia who is nonetheless able to solve working memory tasks thought to rely on mental imagery (Nanay, 2021; for a criticism, see Blomkvist, 2022), and suggests that unconscious imagery could be doing the explanatory work here. Acknowledging that aphantasia is a heterogeneous condition, Nanay suggests that at least some people with aphantasia might have intact unconscious imagery.

A few years down the line, neural evidence now stands to potentially strengthen this hypothesis. We can investigate this using decoding technology reliant on machine learning, whereby representational content from early visual cortex can be decoded. We should expect that if people with aphantasia retain unconscious imagery during a task, this content is decodable. In particular, we should expect to be able to use a decoder trained on data from perceptual trials to decode content from imagery trials (Koenig-Robert and Pearson, 2019). Why? Because the representational content in perception and (unconscious) imagery should be the same, hence, content should be cross-decodable.

Let’s look at a study which has done just that, and which claims to have found ‘imageless imagery in aphantasia’ (Chang et al., 2025). This sounds suspiciously like unconscious imagery – a mental image with no accompanying phenomenology. In the study, 16 participants with aphantasia (as operationalised by VVIQ scores (Marks, 1973) and binocular priming (Keogh and Pearson, 2018)) were compared to controls whilst perceiving or imagining stimuli in an fMRI scanner. The stimuli were Gabor patterns – circular coloured patches with gratings. As expected, content could be cross-decoded in the way described above in the control participants with phenomenally conscious imagery.

But the results showed an unexpected difference between people with aphantasia and controls. Surprisingly, content could not be cross-decodable in people with aphantasia, although it could be independently decoded in imagery trials and perception trials. But if these participants generated unconscious imagery during the imagery trials, it should also have been possible to cross-decode the content. After all, the unconscious imagery should have had the same content (because of the stimuli) and format (because it is imagery) as the perceptual representations. Let’s take these results at face value and see where it leads us.

Have we found unconscious imagery in aphantasia? Despite the authors claiming to have found ‘imageless imagery’, I think there are some obstacles to this interpretation. One obstacle has to do with the content of the representation, and the other with the format of the representation.

Regarding content, we should expect the content of unconscious imagery to correspond to content in perception. Most of Nanay’s work has included this requirement of correspondence – e.g. “mental imagery is perceptual processing (that is, processing in the early sensory cortices, V1, V2,V4/V8, MT) that is not triggered by corresponding sensory stimulation in the relevant sense modality” (Nanay, 2021). However, in Mental Imagery, Nanay explicitly removes the correspondence requirement, but still holds these two definitions to be co-extensive in the vast majority of cases (Nanay, 2024: 8). Perhaps the case raised by the current study is one where correspondence does not hold? It is possible, but unlikely, given that unconscious imagery is supposed to be able to do the computational work in solving visual working memory tasks. The way it could do so is by retaining a content correspondence between perceptual stimuli and imagery. It is in virtue of this correspondence that unconscious imagery could play an explanatory role solving visual working memory tasks. So, we should expect the content of unconscious imagery to correspond to perceptual content, and therefore to be cross-decodable.

Another possibility is that there really is corresponding content in the unconscious imagery representation, but we fail to detect it because of a difference in format. This seems to be the interpretation the authors favour, as they suggest that the imagery representation might be ‘warped’ or ‘shifted’, and hence not cross-decodable. But then we are forced to say that unconscious imagery uses a different representational format compared to both conscious imagery and perception, and this would also be problematic for unconscious imagery’s ability to do explanatory work, as format is also appealed to in explanting results from visual working memory and mental rotation tasks. It is worth noting that Nanay’s view on the format of mental imagery is liberal compared to many other philosophers[1], where he claims that it represents magnitudes by magnitudes (Peacocke, 2019). Despite this liberal view, it seems like we would have deny that unconscious mental imagery shares format with conscious mental imagery and perception. If this is correct, one is left wondering whether this should be called ‘imagery’ at all as it seems to lack the very format which gave rise to the term.

Let’s bring this discussion full circle. We set out to see assess the claim that some aphantasics have unconscious imagery. I discussed a recent study which showed great potential in being able to adjudicate on this claim as it investigated exactly the kind of neural activity that we think is related to unconscious mental imagery. I suggested to take the results at face value and see where that gets us. We end up in the following position: either the content of unconscious mental imagery does not correspond to the content of perception/conscious imagery, or unconscious mental imagery does not share the same format as conscious mental imagery and perception. Or both. Theorising further without additional empirical data won’t get us past than this stalemate.

To conclude, let me reiterate that one study does not sink a theory. As always, it remains possible that the hypothesis was not supported because of limitations of the study (I did not discuss these here, but see the authors’ discussion (Chang et al., 2025)). The hypothesis that some people with aphantasia have unconscious mental imagery remains an intriguing possibility, and it warrants further testing.

References

Block, N. 2011: Perceptual consciousness overflows cognitive access, Trends in Cognitive Sciences,. Elsevier Ltd 15(12): 567–75. doi: 10.1016/j.tics.2011.11.001.

Blomkvist, A. 2022: Aphantasia: In search of a theory, Mind and Language, (May): 1–23. doi: 10.1111/mila.12432.

Chang, S., Zhang, X., Cao, Y., Pearson, J. and Meng, M. 2025: Imageless imagery in aphantasia revealed by early visual cortex decoding, Current Biology,. doi: 10.1016/j.cub.2024.12.012.

Clarke, S. 2022: Mapping the Visual Icon, Philosophical Quarterly,. Oxford University Press 72(3): 552–77. doi: 10.1093/pq/pqab054.

Fodor, J. A. 2007: Revenge of The Given, Contemporary Debates in Philosophy of Mind,. Edited by B. P. McLaughlin and J. D. Cohen. Blackwell.

Keogh, R. and Pearson, J. 2018: The blind mind: No sensory visual imagery in aphantasia, Cortex,. Elsevier Ltd 105(2015): 53–60. doi: 10.1016/j.cortex.2017.10.012.

Koenig-Robert, R. and Pearson, J. 2019: Decoding the contents and strength of imagery before volitional engagement, Scientific Reports,. Nature Publishing Group 9(1). doi: 10.1038/s41598-019-39813-y.

Kung, P. 2010: Imagining as a guide to possibility, Philosophy and Phenomenological Research, 81(3): 620–63. doi: 10.1111/j.1933-1592.2010.00377.x.

Marks, D. F. 1973: Visual Imagery Differences in the Recall of Pictures, British Journal of Psychology, 64(1): 17–24. doi: 10.1111/j.2044-8295.1973.tb01322.x.

Nanay, B. 2021: Unconscious Mental Imagery, Philosophical Transactions of the Royal Society B,.

Nanay, B. 2024: Mental Imagery. Oxford: Oxford University Press.

Peacocke, C. 2019: The Primacy of Metaphysics. Oxford: Oxford University Press.

Quilty-Dunn, J. 2020: Perceptual Pluralism, Nous,. Blackwell Publishing Ltd 54(4): 807–38. doi: 10.1111/nous.12285.

Zeman, A., Dewar, M. and Della Sala, S. 2015: Lives without imagery – Congenital aphantasia, Cortex,. Elsevier Srl. 73: 378–80. doi: 10.1016/j.cortex.2015.05.019.

---

[1] Others who claim that mental imagery is pictorial/imagistic/iconic/depictive often hold that it meets the Picture Principle and Holism Principle (Fodor, 2007; Quilty-Dunn, 2020). For a helpful discussion, see Clarke (2022).