Behavioural results for the semantic and spatial context tasks inside the scanner. SCB = Same Category Buildings: all the items in the building were taken from the same semantic category. MCB = Mixed Category Buildings: the items in the buildings were drawn from different semantic categories.

Probe responses.

Warm colours = semantic > spatial context probes. Cool colours = spatial context > semantic probes. Left panel: Univariate results from Study 1, contrasting semantic and spatial context probes. Right panel: Intrinsic connectivity results from Study 3 using semantic and spatial context probe activation within visual networks as seeds. Panel a: Brain maps depicting the supra-threshold univariate activation results for the probe phase of the semantic and spatial context tasks. Panels b and c: Axial slices showing the overlap of these univariate results with Scene and Object localiser maps from Study 2 (the localiser maps are in green, and the univariate results maps are in warm and cool colours). Panel d: ROI analysis examining the activation in the three default mode subnetworks of the Yeo 17 parcellation during the probe phase of the semantic and spatial context tasks. The error bars in the bar plots depict the standard error of the mean (Note: ***p<.001); the ROIs are shown to the right of the bar plots. Panel e: Brain maps depicting the seeds and intrinsic connectivity results for the semantic and spatial context probe regions. Panel f: Word clouds depicting the cognitive decoding of unthresholded connectivity maps for semantic and spatial context probe seeds using Neurosynth (bigger words reflect stronger correlation of the functional maps with the terms); the colour-code follows that of the brain maps. Panel g: Brain maps showing the overlap of these intrinsic connectivity maps for semantic and spatial context probes with the default mode network from the 7-network parcellation from Yeo et al. (2011).

Decision responses.

Warm colours = semantic > spatial context decisions. Cool colours = spatial context > semantic decisions. Left panel: Univariate results from Study 1 contrasting semantic and spatial context decisions. Right panel: Intrinsic connectivity results from Study 3 using semantic and spatial context decision activation within DMN networks as seeds. Panel a: Brain maps depicting the supra-threshold univariate activation results for the decision phase of the semantic and spatial context tasks. Panels b and c: Axial slices showing the overlap of these univariate results with the FT and MT default mode subnetworks of the Yeo 17 network parcellation (the default mode maps are in green, and the univariate results maps are in warm and cool colours). Panel d: ROI analysis examining the activation in the Scene and Object localiser maps from Study 2 during the decision phase of the semantic and spatial context tasks. The error bars in the bar plots depict the standard error of the mean (Note: ***p<.001, * p < .05); the ROIs are shown in supplementary Figure S3. Panel e: Brain maps depicting the seeds and intrinsic connectivity results for the semantic and spatial context decision regions. Panel f: Word clouds depicting the cognitive decoding of unthresholded connectivity maps for semantic and spatial context decision seeds using Neurosynth (bigger words reflect stronger correlation of the functional maps with the terms); the colour-code follows that of the brain maps. Panel g: Brain maps showing the overlap of these intrinsic connectivity maps with the visual network from the 7-network parcellation from Yeo et al. (2011).

Panel a:

Warm colours = common regions showing stronger intrinsic connectivity to semantic decision regions in DMN and semantic probe regions in visual cortex; Cool colours = common regions showing stronger intrinsic connectivity to spatial context decision regions in DMN and spatial context probe regions in visual cortex. Panel b: The cognitive decoding of these spatial maps using Neurosynth following the same colour code as Panel a. Panel c: Network composition showing the percentage of each pathway map overlapping with the three DMN and two Visual subnetworks defined by the Yeo et al. (2011) 17-network parcellation. Panels d, e and f: These panels depict the seeds, ROIs and their connectivity. The bar plots in panels e and f show the connectivity between DMN decision regions and probe visual regions. Panel g: Results of spatial correlation analysis comparing the semantic and spatial context pathways with non-pathway maps (derived from the conjunction of the connectivity of probe and decision seeds across different tasks, e.g., probe spatial context ∩ decision semantic connectivity). We assessed the spatial similarity of these pathway and non-pathway maps to the univariate activation during the probe and decision phases for each task and each participant. Panel h: Results of the structural connectivity analysis.

Panel a:

Analysis situating the position of the pathways in a whole-brain connectivity gradient space (Margulies et al., 2016). The scatterplots depict the position of each participant’s peak response to the semantic and context task in this gradient space (the big circles represent the mean of each task for that phase). The bar plots compare the mean of each gradient across tasks. The inset on the bottom left of the panel displays Margulies’ et al. (2016) original gradient space. Panel b: Psychophysiological interaction analysis of the connectivity from the spatial context and semantic probe regions to FT and MT-DMN subnetworks. This analysis collapses the SCB and MCB conditions, which showed no significant differences. Note. * = p<.05, ** = p<.01.

Univariate results of the probe phase for the analysis contrasting same category vs mixed category trials separately for the semantic and spatial context tasks. Panel a: univariate results contrasting spatial context same- > mixed-category building trials during the probe phase. Panel b: univariate results of a task by condition (same/mixed-category building) interaction. Panel c: univariate results contrasting semantic same- > mixed-category building trials during the probe phase. Panel d: spatial relations of the same > mixed spatial context cluster with the semantic and spatial context pathways outlined in Figure 4. Panel e: intrinsic connectivity seed-to-ROI results using the three univariate results clusters shown in the top panel as seeds and the pathways as ROIs. The error bars depict the standard error of the mean. *** p<.001.

Results of the representational similarity analysis. Top left panel: within-task RSA results correlating BOLD activity from the probe phase of semantic mixed-category building trials with the semantic similarity matrix described in the Methods section, and BOLD activity from the probe phase of context mixed-category building trials with the context similarity matrix. Top right panel: cross-task similarity analysis correlating BOLD activity from semantic trials with the context similarity matrix. Bottom panel: The left part depicts the spatial relations of the cross-task similarity analysis cluster with the semantic and context pathways outlined in Figure 4; the right part shows Intrinsic connectivity seed-to-ROI results using the within- and cross-task RSA clusters shown in the top panel as seeds and the pathways as ROIs. The error bars depict the standard error of the mean. *** p < .001, ** p < .01.

Top panel: the layout of two buildings is shown, one of them contains semantically related items (SCB), the other contains unrelated items (MCB). These items and locations are shown in the example trials below. Bottom panel: trial procedure for Semantic and Spatial Context decisions. The phases of a trial are shown (Probe, Dots, Decision, Arrow task, Fixation), and the red square indicates the correct response (not shown to participants). Participants were required to press ‘left’ or ‘right’ buttons in the Decision phase. No-decision trials omitted the dots and decision phases.