Congenital aphantasia reveals frontotemporal and cingulate structural alterations underlying conscious access to imagery

Article activity feed

1. 

   eLife

   Jul 10, 2026

   eLife Assessment

   This valuable study provides novel evidence that congenital aphantasia is associated with structural differences in frontotemporal and cingulate systems, with relative sparing of early visual regions and major visual pathways. The multimodal structural imaging approach is carefully implemented and will be of interest to researchers studying mental imagery and aphantasia. However, the strength of evidence is incomplete because the data cannot adjudicate between alternative cognitive interpretations, and the multiple discovery streams make the findings better viewed as key exploratory evidence, rather than as establishing a definitive structural phenotype of aphantasia.

   Read the original source
2. 

   eLife

   Jul 10, 2026

   Reviewer #1 (Public review):

   Summary

   In this paper, the authors provide a systematic investigation of structural brain differences associated with congenital aphantasia (self-reported lifelong absence of voluntary visual imagery). Specifically, the authors analysed a structural neuroimaging dataset involving 18 individuals with aphantasia and 18 visualizers to test two competing hypotheses: (1) that aphantasia reflects alterations in visual pathways and early visual cortex, and (2) that it instead reflects differences in higher-order frontotemporal and cingulate systems. To test these hypotheses, the authors employed multiple analysis approaches (e.g., cortical morphometry, tractometry, graph-theoretic network analysis).

   They report structural differences between the two groups in frontotemporal and cingulate systems. In contrast, they …

   Reviewer #1 (Public review):

   Summary

   In this paper, the authors provide a systematic investigation of structural brain differences associated with congenital aphantasia (self-reported lifelong absence of voluntary visual imagery). Specifically, the authors analysed a structural neuroimaging dataset involving 18 individuals with aphantasia and 18 visualizers to test two competing hypotheses: (1) that aphantasia reflects alterations in visual pathways and early visual cortex, and (2) that it instead reflects differences in higher-order frontotemporal and cingulate systems. To test these hypotheses, the authors employed multiple analysis approaches (e.g., cortical morphometry, tractometry, graph-theoretic network analysis).

   They report structural differences between the two groups in frontotemporal and cingulate systems. In contrast, they found no reliable group differences in early visual cortex or major visual tracts. On this basis, they propose that aphantasia is primarily associated with differences in higher-order systems supporting integration and conscious access to internally generated representations, rather than with deficits in sensory visual representations themselves.

   Strengths

   (1) The present work addresses an important gap in the mental imagery literature, providing a systematic investigation of structural neuroimaging differences in congenital aphantasia. By showing that structural differences between aphantasics and visualizers are mainly concentrated in frontotemporal and cingulate systems (rather than in visual cortex), it makes an important step toward a better understanding of individual differences in mental imagery and provides a set of candidate regions for future mechanistic work.

   (2) A key strength of the study is the multimodal approach employed to address the main research question, integrating tractometry, functional region-of-interest (fROI)-based tractography, graph-theoretic network analysis, and surface-based cortical morphometry, which provide a converging assessment of structural differences between aphantasics and visualizers.

   (3) The complementary use of Bayesian analyses alongside NHST to assess evidence for null results is a further strength of this work.

   Weaknesses:

   (1) A weakness of this work is related to aspects of the framing and, in particular, what can be confidently inferred from the results. The framing of existing accounts of aphantasia in the Introduction appears limited in that it reduces the views on aphantasia to two options (sensory strength account versus conscious access account) without acknowledging a third distinct position, namely that aphantasia reflects a specific deficit in the voluntary generation of imagery (Milton et al., 2021; Zeman et al., 2015, 2020; Whiteley, 2021; Cavedon-Taylor, 2022). Like the conscious access account, the view that aphantasia involves a deficit in the generation of sensory representation also speaks against the hypothesis of reduced sensory strength of internally generated representations. This third view could be acknowledged/discussed as it also maps quite well onto the presented results.

   (2) Relatedly, I think the main weakness of the paper concerns the interpretation of results being restricted to a lack of "conscious access". The paper frames its findings as mainly evidence for a conscious access failure, the view that visual representations are generated by aphantasics but cannot be consciously accessed. However, the structural findings are equally consistent with a voluntary generation failure, especially since the same higher-order regions examined can also be implicated in the top-down generation and control of imagery. The authors themselves initially define aphantasia as "lifelong absence of voluntary visual imagery". Given the nature of structural imaging data (as opposed to functional data), it is not possible with the present study to distinguish between a lack of generation versus a lack of conscious access. As such, examining this alternative interpretation appears appropriate, and it would considerably strengthen the paper. Structural MRI alone is not sufficient to dissociate imagery generation from conscious access, as these are fundamentally functional questions.

   (3) Some inconsistency and lack of clarity around the specific choice of regions/networks, which could be better motivated and explained. E.g., the "core imagery network" analysed in the white-matter connections analysis was derived from a previous 7T study (with which the sample partially overlaps) and is not necessarily the network most commonly associated with visual imagery in the literature (e.g., see Dijkstra et al., 2019; Pearson, 2019). It is, for instance, unclear why V1 was examined in the cortical thickness analysis but not in the previous one, given that both analyses are related to the visual pathway hypothesis. Related to this, in the graph-theoretic analysis, the rationale for network selection is inconsistently established in the Introduction. The attention and salience networks do have some grounding in the Introduction through the mention of specific regions such as FEF and anterior insula, though these are discussed as individual regions rather than as networks. However, the default mode network receives no motivation in the Introduction. More explicit elaboration on these choices would be appropriate.

   (3) The interpretation provided in the Discussion tends to oversimplify what is in fact a heterogeneous and rich set of structural findings into a relatively coherent mechanistic account. The observed differences are spatially and directionally variable across tracts, cortical regions, and metrics: e.g., FA is reduced in the UF and posterior interparietal corpus callosum but increased in the dorsal cingulum; cortical thickness is reduced in aPFC but increased in medial temporal regions, and so forth. The Discussion acknowledges this in part (e.g., proposing increased dorsal cingulum FA as potentially compensatory) but does not address the directional heterogeneity systematically. The authors could discuss more explicitly what the opposing directions of effects mean for their overall interpretation. Relatedly, some parts of the Discussion link specific structural findings to specific imagery processes in ways that go beyond what the current data can support. The authors could more clearly distinguish between what the structural data show and what functional interpretations are taken from prior work.

   Read the original source
3. 

   eLife

   Jul 10, 2026

   Reviewer #2 (Public review):

   Summary:

   This paper addresses whether congenital aphantasia reflects an alteration of visual representations themselves, or rather of the systems that allow internally generated representations to reach conscious experience.

   Strengths:

   The study is novel and ambitious. The authors combine several complementary structural MRI approaches in a rare and well-characterised population, and the convergence of the findings toward frontotemporal and cingulate systems, with relative sparing of early visual cortex and major visual pathways, is particularly interesting because it could affect the way visual imagery is modelled and tested experimentally and clinically.

   Weaknesses:

   Overall, I found the manuscript conceptually and methodologically strong. My main concern regards the interpretation of the anatomical …

   Reviewer #2 (Public review):

   Summary:

   This paper addresses whether congenital aphantasia reflects an alteration of visual representations themselves, or rather of the systems that allow internally generated representations to reach conscious experience.

   Strengths:

   The study is novel and ambitious. The authors combine several complementary structural MRI approaches in a rare and well-characterised population, and the convergence of the findings toward frontotemporal and cingulate systems, with relative sparing of early visual cortex and major visual pathways, is particularly interesting because it could affect the way visual imagery is modelled and tested experimentally and clinically.

   Weaknesses:

   Overall, I found the manuscript conceptually and methodologically strong. My main concern regards the interpretation of the anatomical findings, rather than the findings per se. The authors discuss their results within a rich cognitive framework. However, the current dataset does not appear to include independent behavioural or neuropsychological measures that would allow the proposed cognitive interpretation to be tested in the same participants. As a result, the manuscript sometimes moves quite rapidly from 'these structural differences involve systems associated with higher-order control, salience, conscious access' to 'these structural differences may explain the cognitive mechanisms of aphantasia'. I agree that this is the most interesting interpretation, and probably the right one to explore. Although plausible, it remains indirect. The authors already acknowledge this point when discussing memory, affective control, and semantic processing. However, the same logic should be extended to the interpretation of the full set of findings. For example, if the salience/anterior insula findings are interpreted in relation to access to internally generated representations, it would be useful to know whether aphantasic participants also differ behaviourally on tasks tapping interoception or related aspects of internal monitoring. I appreciate that collecting additional behavioural data may not be feasible at this stage, especially given the difficulty of recruiting participants with such a specific manifestation. However, I think it should be acknowledged more explicitly in a dedicated limitation paragraph.

   Read the original source
4. 

   eLife

   Jul 10, 2026

   Reviewer #3 (Public review):

   Summary:

   The authors investigate the structural brain basis of congenital aphantasia, a condition characterised by a lifelong absence of voluntary mental imagery. They test two competing accounts: one predicting structural differences in early visual pathways, the other predicting differences in higher-order frontotemporal and cingulate systems. To do this, they combine four complementary structural imaging approaches: white-matter microstructure profiling along anatomically defined tracts, tractography seeded from functional regions of interest, whole-brain structural network analysis, and cortical thickness mapping. The main finding is that white-matter differences are selective for frontotemporal and cingulate pathways and absent in early visual pathways, which the authors interpret as support for the …

   Reviewer #3 (Public review):

   Summary:

   The authors investigate the structural brain basis of congenital aphantasia, a condition characterised by a lifelong absence of voluntary mental imagery. They test two competing accounts: one predicting structural differences in early visual pathways, the other predicting differences in higher-order frontotemporal and cingulate systems. To do this, they combine four complementary structural imaging approaches: white-matter microstructure profiling along anatomically defined tracts, tractography seeded from functional regions of interest, whole-brain structural network analysis, and cortical thickness mapping. The main finding is that white-matter differences are selective for frontotemporal and cingulate pathways and absent in early visual pathways, which the authors interpret as support for the higher-order account.

   Strengths:

   The multi-modal design is a genuine strength: running four independent analyses increases the chance of detecting real effects and of identifying false positives that appear in only one stream. The statistical choices within each analysis are appropriate. Permutation-based correction with a threshold-free method is well-suited to the tract-level comparisons. The use of Bayes factors to quantify evidence for null results, rather than simply reporting non-significant tests, is particularly valuable here, since the absence of visual pathway differences is central to the argument. The robustness checks across multiple brain parcellations for the network analysis strengthen confidence in those findings.

   Weaknesses:

   The main limitation concerns the relationship between two of the analysis streams. The measure used to weight structural connections in the network analysis is calibrated to match fiber density estimates derived from the same diffusion signal that drives the white-matter microstructure differences. If the two groups differ in tissue organisation in certain pathways (which the microstructure analysis suggests they do), that difference will feed into both measures. The authors should acknowledge this dependency when discussing convergence across analyses.

   More broadly, the imaging metrics used throughout (measures of fiber organisation and weighted connection counts) reflect what the diffusion model captures from the tissue and cannot be directly read as measures of axon number or connection strength. This is a known limitation of the field, but it is relevant to the strength of structural claims made in this paper.

   The network analysis is presented without comparison to a null network. Without this, it is hard to know whether the node-level differences reflect specific network topology or simply follow from overall differences in connectivity weight or density between groups.

   The study runs four separate discovery analyses on the same 36 participants, each corrected within itself but with no control across analysis streams. At 18 participants per group, this is exploratory work. Some of the language used in the abstract and discussion, like "first comprehensive characterization" and "selective structural phenotype", reads as more definitive than the data support at this sample size. Framing the results as hypotheses to be replicated would make the paper stronger.

   The paper frames the results as distinguishing between two competing accounts. The positive evidence for the higher-order account is clear. The absence of differences in visual pathways is a different kind of result: it means such differences were not detected in this sample, not that visual pathways are uninvolved. The discussion at times moves toward that stronger conclusion, which the data do not support.

   The cortical thickness analysis finds one cluster in the predicted direction, while the other analyses each return multiple effects. One cluster in a whole-brain search with 18 participants per group is not strong evidence and should not be presented as equivalent to the other results.

   Effect sizes are reported without confidence intervals throughout. With 18 participants per group, the uncertainty around those estimates is large, and confidence intervals would give readers a more accurate sense of what can be concluded.

   Read the original source
5. 

   eLife

   Jul 10, 2026

   Author response:

   Reviewer #1 (Public review):

   Summary:

   In this paper, the authors provide a systematic investigation of structural brain differences associated with congenital aphantasia (self-reported lifelong absence of voluntary visual imagery). Specifically, the authors analysed a structural neuroimaging dataset involving 18 individuals with aphantasia and 18 visualizers to test two competing hypotheses: (1) that aphantasia reflects alterations in visual pathways and early visual cortex, and (2) that it instead reflects differences in higher-order frontotemporal and cingulate systems. To test these hypotheses, the authors employed multiple analysis approaches (e.g., cortical morphometry, tractometry, graph-theoretic network analysis).

   They report structural differences between the two groups in frontotemporal and cingulate systems. …

   Author response:

   Reviewer #1 (Public review):

   Summary:

   In this paper, the authors provide a systematic investigation of structural brain differences associated with congenital aphantasia (self-reported lifelong absence of voluntary visual imagery). Specifically, the authors analysed a structural neuroimaging dataset involving 18 individuals with aphantasia and 18 visualizers to test two competing hypotheses: (1) that aphantasia reflects alterations in visual pathways and early visual cortex, and (2) that it instead reflects differences in higher-order frontotemporal and cingulate systems. To test these hypotheses, the authors employed multiple analysis approaches (e.g., cortical morphometry, tractometry, graph-theoretic network analysis).

   They report structural differences between the two groups in frontotemporal and cingulate systems. In contrast, they found no reliable group differences in early visual cortex or major visual tracts. On this basis, they propose that aphantasia is primarily associated with differences in higher-order systems supporting integration and conscious access to internally generated representations, rather than with deficits in sensory visual representations themselves.

   Strengths:

   (1) The present work addresses an important gap in the mental imagery literature, providing a systematic investigation of structural neuroimaging differences in congenital aphantasia. By showing that structural differences between aphantasics and visualizers are mainly concentrated in frontotemporal and cingulate systems (rather than in visual cortex), it makes an important step toward a better understanding of individual differences in mental imagery and provides a set of candidate regions for future mechanistic work.

   (2) A key strength of the study is the multimodal approach employed to address the main research question, integrating tractometry, functional region-of-interest (fROI)-based tractography, graph-theoretic network analysis, and surface-based cortical morphometry, which provide a converging assessment of structural differences between aphantasics and visualizers.

   (3) The complementary use of Bayesian analyses alongside NHST to assess evidence for null results is a further strength of this work.

   Weaknesses:

   (1) A weakness of this work is related to aspects of the framing and, in particular, what can be confidently inferred from the results. The framing of existing accounts of aphantasia in the Introduction appears limited in that it reduces the views on aphantasia to two options (sensory strength account versus conscious access account) without acknowledging a third distinct position, namely that aphantasia reflects a specific deficit in the voluntary generation of imagery (Milton et al., 2021; Zeman et al., 2015, 2020; Whiteley, 2021; Cavedon-Taylor, 2022). Like the conscious access account, the view that aphantasia involves a deficit in the generation of sensory representation also speaks against the hypothesis of reduced sensory strength of internally generated representations. This third view could be acknowledged/discussed as it also maps quite well onto the presented results.

   (2) Relatedly, I think the main weakness of the paper concerns the interpretation of results being restricted to a lack of "conscious access". The paper frames its findings as mainly evidence for a conscious access failure, the view that visual representations are generated by aphantasics but cannot be consciously accessed. However, the structural findings are equally consistent with a voluntary generation failure, especially since the same higher-order regions examined can also be implicated in the top-down generation and control of imagery. The authors themselves initially define aphantasia as "lifelong absence of voluntary visual imagery". Given the nature of structural imaging data (as opposed to functional data), it is not possible with the present study to distinguish between a lack of generation versus a lack of conscious access. As such, examining this alternative interpretation appears appropriate, and it would considerably strengthen the paper. Structural MRI alone is not sufficient to dissociate imagery generation from conscious access, as these are fundamentally functional questions.

   (3) Some inconsistency and lack of clarity around the specific choice of regions/networks, which could be better motivated and explained. E.g., the "core imagery network" analysed in the white-matter connections analysis was derived from a previous 7T study (with which the sample partially overlaps) and is not necessarily the network most commonly associated with visual imagery in the literature (e.g., see Dijkstra et al., 2019; Pearson, 2019). It is, for instance, unclear why V1 was examined in the cortical thickness analysis but not in the previous one, given that both analyses are related to the visual pathway hypothesis. Related to this, in the graph-theoretic analysis, the rationale for network selection is inconsistently established in the Introduction. The attention and salience networks do have some grounding in the Introduction through the mention of specific regions such as FEF and anterior insula, though these are discussed as individual regions rather than as networks. However, the default mode network receives no motivation in the Introduction. More explicit elaboration on these choices would be appropriate.

   (4) The interpretation provided in the Discussion tends to oversimplify what is in fact a heterogeneous and rich set of structural findings into a relatively coherent mechanistic account. The observed differences are spatially and directionally variable across tracts, cortical regions, and metrics: e.g., FA is reduced in the UF and posterior interparietal corpus callosum but increased in the dorsal cingulum; cortical thickness is reduced in aPFC but increased in medial temporal regions, and so forth. The Discussion acknowledges this in part (e.g., proposing increased dorsal cingulum FA as potentially compensatory) but does not address the directional heterogeneity systematically. The authors could discuss more explicitly what the opposing directions of effects mean for their overall interpretation. Relatedly, some parts of the Discussion link specific structural findings to specific imagery processes in ways that go beyond what the current data can support. The authors could more clearly distinguish between what the structural data show and what functional interpretations are taken from prior work.

   We will add two recent in-press Cortex papers to the Discussion. One provides lesion-based double-dissociation evidence against V1 as a necessary causal substrate of visual imagery. The other shows that aphantasic individuals can display visualizer-like oculomotor patterns during mental map exploration despite reporting little or no imagery vividness. Together, these studies help clarify our interpretation of our null V1 findings and structural effects in higher-order brain regions, which are consistent with aphantasia involving altered integration or access rather than a primary V1-dependent imagery deficit.

   Reviewer #2 (Public review):

   Summary:

   This paper addresses whether congenital aphantasia reflects an alteration of visual representations themselves, or rather of the systems that allow internally generated representations to reach conscious experience.

   Strengths:

   The study is novel and ambitious. The authors combine several complementary structural MRI approaches in a rare and well-characterised population, and the convergence of the findings toward frontotemporal and cingulate systems, with relative sparing of early visual cortex and major visual pathways, is particularly interesting because it could affect the way visual imagery is modelled and tested experimentally and clinically.

   Weaknesses:

   Overall, I found the manuscript conceptually and methodologically strong. My main concern regards the interpretation of the anatomical findings, rather than the findings per se. The authors discuss their results within a rich cognitive framework. However, the current dataset does not appear to include independent behavioural or neuropsychological measures that would allow the proposed cognitive interpretation to be tested in the same participants. As a result, the manuscript sometimes moves quite rapidly from 'these structural differences involve systems associated with higher-order control, salience, conscious access' to 'these structural differences may explain the cognitive mechanisms of aphantasia'. I agree that this is the most interesting interpretation, and probably the right one to explore. Although plausible, it remains indirect. The authors already acknowledge this point when discussing memory, affective control, and semantic processing. However, the same logic should be extended to the interpretation of the full set of findings. For example, if the salience/anterior insula findings are interpreted in relation to access to internally generated representations, it would be useful to know whether aphantasic participants also differ behaviourally on tasks tapping interoception or related aspects of internal monitoring. I appreciate that collecting additional behavioural data may not be feasible at this stage, especially given the difficulty of recruiting participants with such a specific manifestation. However, I think it should be acknowledged more explicitly in a dedicated limitation paragraph.

   We thank the reviewer for this thoughtful and constructive comment. Lack of introspective report of voluntary imagery is arguably the defining signature of aphantasia. This motivated us to primarily interpret our anatomical findings in a broader cognitive context of higher-order control, internal monitoring, and conscious access in aphantasia. We expect that a reliable behavioural test measuring imagery sensitivity and accessibility would allow us to direct link these findings to individual imagery ability. Nevertheless, to our best knowledge, this kind of test on imagery is still missing. Instead, our findings point to some plausible structural signature or brain regions that may be related to conscious imagery, which motivate future studies to examine their direct or causal roles. We agree with the reviewer, future studies should test the relationship between these anatomical structures and the accessibility to internal representation, together with related aspects of internal monitoring. We will therefore add a dedicated paragraph to discuss the plausible cognitive mechanisms during the revision.

   Reviewer #3 (Public review):

   Summary:

   The authors investigate the structural brain basis of congenital aphantasia, a condition characterised by a lifelong absence of voluntary mental imagery. They test two competing accounts: one predicting structural differences in early visual pathways, the other predicting differences in higher-order frontotemporal and cingulate systems. To do this, they combine four complementary structural imaging approaches: white-matter microstructure profiling along anatomically defined tracts, tractography seeded from functional regions of interest, whole-brain structural network analysis, and cortical thickness mapping. The main finding is that white-matter differences are selective for frontotemporal and cingulate pathways and absent in early visual pathways, which the authors interpret as support for the higher-order account.

   Strengths:

   The multi-modal design is a genuine strength: running four independent analyses increases the chance of detecting real effects and of identifying false positives that appear in only one stream. The statistical choices within each analysis are appropriate. Permutation-based correction with a threshold-free method is well-suited to the tract-level comparisons. The use of Bayes factors to quantify evidence for null results, rather than simply reporting non-significant tests, is particularly valuable here, since the absence of visual pathway differences is central to the argument. The robustness checks across multiple brain parcellations for the network analysis strengthen confidence in those findings.

   Weaknesses:

   The main limitation concerns the relationship between two of the analysis streams. The measure used to weight structural connections in the network analysis is calibrated to match fiber density estimates derived from the same diffusion signal that drives the white-matter microstructure differences. If the two groups differ in tissue organisation in certain pathways (which the microstructure analysis suggests they do), that difference will feed into both measures. The authors should acknowledge this dependency when discussing convergence across analyses.

   More broadly, the imaging metrics used throughout (measures of fiber organisation and weighted connection counts) reflect what the diffusion model captures from the tissue and cannot be directly read as measures of axon number or connection strength. This is a known limitation of the field, but it is relevant to the strength of structural claims made in this paper.

   The network analysis is presented without comparison to a null network. Without this, it is hard to know whether the node-level differences reflect specific network topology or simply follow from overall differences in connectivity weight or density between groups.

   The study runs four separate discovery analyses on the same 36 participants, each corrected within itself but with no control across analysis streams. At 18 participants per group, this is exploratory work. Some of the language used in the abstract and discussion, like "first comprehensive characterization" and "selective structural phenotype", reads as more definitive than the data support at this sample size. Framing the results as hypotheses to be replicated would make the paper stronger.

   The paper frames the results as distinguishing between two competing accounts. The positive evidence for the higher-order account is clear. The absence of differences in visual pathways is a different kind of result: it means such differences were not detected in this sample, not that visual pathways are uninvolved. The discussion at times moves toward that stronger conclusion, which the data do not support.

   The cortical thickness analysis finds one cluster in the predicted direction, while the other analyses each return multiple effects. One cluster in a whole-brain search with 18 participants per group is not strong evidence and should not be presented as equivalent to the other results.

   Effect sizes are reported without confidence intervals throughout. With 18 participants per group, the uncertainty around those estimates is large, and confidence intervals would give readers a more accurate sense of what can be concluded.

   We are grateful to the Reviewer for the constructive and thoughtful assessment of our manuscript. In response to the reviewer’s comments, we will revise the manuscript to clarify the dependency between diffusion-derived analysis streams, to state more explicitly the biological limits of diffusion MRI metrics, to add a null-network sensitivity analysis for the clustering coefficient findings, to include confidence intervals for reported effect sizes, and to temper the interpretation of the cortical thickness result. We will also revise the Abstract and Discussion to better reflect the exploratory nature of the study and to frame the findings as hypotheses requiring replication in larger independent samples. We believe that these revisions will make the manuscript more balanced, transparent, and appropriately cautious, while preserving the central conclusion that congenital aphantasia is associated with structural differences centered on higher-order frontotemporal and cingulate systems.

   Read the original source
6. 

   Version published to 10.64898/2026.03.22.713248 on bioRxiv

   Mar 22, 2026