Figures and data
Tractometry reveals selective white-matter microstructural differences in aphantasia.
(A) Tracts showing reduced fractional anisotropy (FA) in aphantasic individuals relative to visualizers (Vis. > Aph.): left uncinate fasciculus (UF), right UF, and posterior interparietal corpus callosum. Upper panels show tract profiles (mean ± SEM); lower panels show tract reconstructions from a representative participant, color-coded by segment position (nodes 1-100). Red lines indicate aphantasic individuals and blue lines indicate visualizers. Shaded gray regions mark segments showing significant group differences (p < 0.05, FWE-corrected). (B) Tracts showing increased FA in aphantasic individuals relative to visualizers (Aph. > Vis.): left and right dorsal cingulum. Upper panels show tract profiles (mean ± SEM); lower panels show representative tract reconstructions, color-coded by segment position. Red lines indicate aphantasic individuals and blue lines indicate visualizers. Shaded gray regions and black bars on the x axis mark significant segments (p < 0.05, FWE-corrected). Thin lines show individual participant profiles; thick lines show group means.
Structural network analysis reveals reduced local segregation and increased inter-network hubness in congenital aphantasia.
(A) Analysis pipeline. Whole-brain tractography streamlines were weighted using spherical-deconvolution informed filtering of tractograms (SIFT2) and normalized by ROI volume to generate a 400 × 400 structural connectivity matrix based on the Schaefer 17-network parcellation. Node-level graph-theoretic metrics were then computed from the weighted undirected graph. (B) Node strength was reduced in the left frontal eye field (FEF; dorsal attention network B, DAN B) in aphantasic individuals relative to visualizers, indicating weaker overall structural connectivity at this node. (C) Clustering coefficient was reduced in the left anterior insula (salience/ventral attention network A, Sal/VAN A) and left dorsolateral prefrontal cortex (dlPFC; Default B), indicating reduced local network segregation in salience and association-network regions. (D) Participation coefficient was increased in the right dorsomedial prefrontal cortex (dmPFC; Default A) in aphantasia, indicating greater inter-network hubness at this node. Blue indicates Vis. > Aph.; dark red indicates Aph. > Vis. Boxplots show median and interquartile range; *p < 0.05, **p < 0.01, ***p < 0.001, FWE-corrected.
Regional differences in cortical thickness between aphantasic individuals and visualizers.
(A) Whole-brain t map of the group comparison. Blue indicates regions of reduced cortical thickness and red indicates regions of increased cortical thickness in aphantasia, with whole-cortex FDR correction. Boxplots show individual participant values (dots), median, and interquartile range. *p < 0.05, **p < 0.01, ***p < 0.001. Vis., visualizers; Aph., aphantasia. aPFC: anterior prefrontal cortex; PoCG: post central gyrus; PHG: parahippocampal gyrus; EC: entorhinal cortex; ITG: inferior temporal gyrus; RSC: retrosplenial cortex; PCC: posterior cingulate cortex. (B) Functional term associations derived from meta-analytic decoding of the t map using Neurosynth. Blue terms are associated with regions of reduced cortical thickness in aphantasia, and red terms with regions of increased cortical thickness. Term size reflects the strength of the Pearson correlation with the corresponding meta-analytic activation map.
Overview of the analysis pipeline.
Three data streams were processed in parallel. Diffusion MRI data underwent preprocessing and fiber reconstruction to generate whole-brain tractography. Anatomical T1-weighted images were processed for grey matter/white matter (GM/WM) segmentation and cortical surface extraction. Functional ROIs from a previous 7T fMRI study of mental imagery (Liu et al., 2025b) were registered to each participant’s native space. These inputs were then used for four complementary analyses: (1) tractometry, in which white-matter bundles were segmented and microstructural properties, including fractional anisotropy (FA), were quantified along tract profiles; (2) fROI-based tractography, in which functional ROIs were projected to the GM/WM interface to identify streamlines connecting regions of the core imagery network; (3) graph-theoretic network analysis, in which cortical parcellation and SIFT2-weighted streamline counts were used to construct structural connectivity matrices and derive node-level network metrics; and (4) surface morphometry, in which cortical thickness was estimated from the extracted cortical surfaces and compared between groups across the whole brain.
NODDI microstructural profiles in white-matter tracts showing significant FA differences.
(A) Neurite density index (NDI) tract profiles. (B) Orientation dispersion index (ODI) tract profiles. Profiles are shown for the bilateral uncinate fasciculus, posterior interparietal corpus callosum, and bilateral dorsal cingulum, the tracts identified in the main tractometry analysis. Red lines indicate aphantasic individuals and blue lines indicate visualizers. Thin lines show individual participant profiles; thick lines show group means ± SEM. Shaded grey regions and black bars indicate significant between-group differences (p < 0.05, FWE-corrected).
Marginal FA reductions in the right SLF-II and right ventral cingulum are accompanied by increased orientation dispersion.
(A) Fractional anisotropy (FA) tract profiles for the right SLF-II (p = 0.082, FWE-corrected) and right ventral cingulum (p = 0.079, FWE-corrected), with representative tract reconstructions shown below. Shaded grey regions and black bars indicate segments showing marginal between-group differences. Red lines indicate aphantasic individuals and blue lines indicate visualizers; thin lines show individual participant profiles and thick lines show group means ± SEM. (B) Orientation dispersion index (ODI) profiles for the same tracts. In both tracts, marginal FA reductions overlapped with increased ODI in aphantasic individuals, suggesting that these effects may be influenced by local fiber geometry rather than robust microstructural differences. These findings were therefore not interpreted further.
White-matter connections of the core imagery network.
(A) White-matter streamlines associated with each functional ROI of the core imagery network, shown on sagittal MRI sections: left fusiform imagery node (FIN), left orbitofrontal cortex (OFC), left anterior prefrontal cortex (aPFC), and right anterior intraparietal sulcus (aIPS). (B) Streamline counts normalized by ROI volume for each functional ROI. The FIN was connected primarily by the long segment of the arcuate fasciculus (AF), posterior arcuate fasciculus (pAF), and vertical occipital fasciculus (VOF). The left OFC was connected by both the inferior fronto-occipital fasciculus (IFOF) and uncinate fasciculus (UF), whereas the left aPFC was connected mainly by the IFOF. No direct structural connection was identified between the FIN and left aPFC in this analysis. The right aIPS was connected predominantly by the third branch of the superior longitudinal fasciculus (SLF-III). A nominal increase in normalized streamline count was observed in aphantasic individuals for the left AF associated with the FIN (uncorrected p = 0.039, U = 229, Cliff’s d = -0.41), but this effect did not survive correction for multiple comparisons. Bayes factor analysis provided moderate evidence for no group difference in VOF streamline count associated with the FIN (BF = 0.32). (C) Proportion of streamlines assigned to each tract of interest for each ROI. Light bars indicate visualizers and dark bars indicate aphantasic individuals. Error bars show SEM and dots indicate individual participants. IFOF, inferior fronto-occipital fasciculus; ILF, inferior longitudinal fasciculus; AF, arcuate fasciculus long segment; pAF, posterior arcuate fasciculus; VOF, vertical occipital fasciculus; UF, uncinate fasciculus; SLF-I/II/III, superior longitudinal fasciculus branches I, II, and III.
Control analyses confirm the robustness of network findings across parcellation schemes.
(A) Gordon 333-region atlas. Convergent group differences were observed across graph-theoretic metrics, with aphantasic individuals showing reduced clustering coefficient in the left anterior insula (Vis. > Aph., blue) and increased participation coefficient in the right dorsomedial prefrontal cortex (dmPFC; Aph. > Vis., dark red). (B) Schaefer-Yeo 100-region atlas. Aphantasic individuals showed increased participation coefficient in bilateral dmPFC (Aph. > Vis.) and reduced participation coefficient in right lateral prefrontal cortex (Vis. > Aph.). (C) White-matter tracts connected to the left anterior insula node in the Schaefer-Yeo 400-region atlas. The inferior fronto-occipital fasciculus (IFOF) and arcuate fasciculus (AF) converge on the left anterior insula, the node showing reduced clustering coefficient in the main analysis. The left anterior prefrontal cortex (aPFC) and fusiform imagery node (FIN) are linked by distinct white-matter pathways, primarily IFOF and AF, respectively. Dark red indicates Aph. > Vis. and blue indicates Vis. > Aph. All group differences shown in panels A and B were significant at p < 0.05, FDR-corrected. The left frontal eye field (FEF) was not identified consistently across atlases, consistent with uncertainty in its parcellation boundaries.
Cortical NODDI microstructural differences between aphantasic individuals and visualizers.
Cortical NODDI metrics were compared between groups using random field theory as implemented in BrainStat, with cluster-level FWE correction (p < 0.05). Aphantasic individuals showed increased neurite density index (NDI) in bilateral medial temporal regions, including the parahippocampal gyrus and entorhinal cortex, as well as in the left precuneus, and reduced NDI in the left dorsomedial prefrontal cortex. Orientation dispersion index (ODI), isotropic volume fraction (ISOVF), and mean diffusivity (MD) also showed significant group differences. Regions of reduced MD overlapped spatially with areas of increased NDI in medial temporal cortex, a pattern consistent with denser neurite packing and lower overall diffusivity in these regions.
Tract-level diffusion and NODDI differences between aphantasic individuals and visualizers.
Significant group differences in tract-profile diffusion tensor imaging (DTI) and neurite orientation dispersion and density imaging (NODDI) metrics are shown for tracts identified in the tractometry analysis. Values indicate the start and end nodes of each significant cluster, followed by the FWE-corrected p value, test statistic, and Cohen’s d in parentheses. Direction of effect is indicated separately for each metric column (Vis. > Aph. or Aph. > Vis.). Only clusters meeting the reporting threshold of at least 3 consecutive significant nodes are listed. For the main analyses, significance was defined as p < 0.05, FWE-corrected; marginal effects (p < 0.1, FWE-corrected) are reported in the text but are not included in this table.
Nodal graph-theoretic network differences between aphantasic individuals and visualizers.
Significant group differences in nodal graph-theoretic metrics are reported for node strength, clustering coefficient, and participation coefficient. Group comparisons were performed using the Brunner-Munzel test with false discovery rate (FDR) correction within each metric. MNI coordinates (x, y, z) correspond to parcel centroids. Network affiliations are defined according to the Schaefer 400-parcel, 17-network atlas. Vis. > Aph. indicates lower metric values in aphantasic individuals, whereas Aph. > Vis. indicates higher metric values in aphantasic individuals. B-M statistic denotes the Brunner-Munzel statistic, and Cliff’s delta the effect size. Significance threshold: p < 0.05, FDR-corrected.
Gray matter volume in occipital cortical regions.
Regions were defined using the Neuromorphometrics atlas in CAT12. Gray matter volumes were normalized by total intracranial volume (TIV). No significant group differences were observed in any occipital cortical region. Bayes factors provided moderate evidence for an equal volume between the two groups in the left and right calcarine cortices. Values are reported as mean ± SD.
Cortical thickness in early visual areas in aphantasic individuals and visualizers.
Mean cortical thickness values (mm ± SD) for atlas-defined early visual regions are shown for the aphantasia and visualizer groups. Regions of interest were defined using the HCP-MMP atlas. No significant group differences were observed in any region. Bayes factors indicated moderate evidence for the absence of a group difference in 4 of the 6 regions (BF+₀ < 0.33). l, left hemisphere; r, right hemisphere.
Cortical thickness differences between aphantasic individuals and visualizers.
Significant group differences in cortical thickness are reported for clusters comprising more than 30 vertices. Group comparisons were performed using a linear model with FDR correction. aPFC: anterior prefrontal cortex; PoCG: Post central gyrus; PHG: parahippocampal gyrus; EC: Entorhinal cortex; ACC: Anterior cingulate cortex; RSC: retrosplenial cortex; PCC: Posterior cingulate cortex.
