Bipartite Functional Fractionation within the Default Network Supports Disparate Forms of Internally Oriented Cognition

Abstract Our understanding about the functionality of the brain’s default network (DN) has significantly evolved over the past decade. Whereas traditional views define this network based on its suspension/disengagement during task-oriented behavior, contemporary accounts have characterized various situations wherein the DN actively contributes to task performance. However, it is unclear how different task-contexts drive componential regions of the DN to coalesce into a unitary network and fractionate into different subnetworks. Here we report a compendium of evidence that provides answers to these questions. Across multiple analyses, we found a striking dyadic structure within the DN in terms of the profiles of task-triggered fMRI response and effective connectivity, significantly extending beyond previous inferences based on meta-analysis and resting-state activities. In this dichotomy, one subset of DN regions prefers mental activities “interfacing with” perceptible events, while the other subset prefers activities “detached from” perceptible events. While both show a common “aversion” to sensory-motoric activities, their differential preferences manifest a subdivision that sheds light upon the taxonomy of the brain’s memory systems. This dichotomy is consistent with proposals of a macroscale gradational structure spanning across the cerebrum. This gradient increases its representational complexity, from primitive sensory-motoric processing, through lexical-semantic representations, to elaborated self-generated thoughts.

3 using affine transformation. In the process of creating the group's template brain using individual T1, for each individual DARTEL estimated 'flow fields' that contained the parameters for contorting native T1-weighted images to the group template. SPM8 deformation utility was then applied to combine group-to-MNI affine parameters with each participant's 'flow fields' to enable tailored warping into the MNI space with better accuracy.
GLM analysis. For each participant, contrasts of interest were estimated using general linear model (GLM) convolving the experimental design matrices with a canonical haemodynamic response function, with resting periods modelled implicitly. Motion parameters were entered into the model as covariates of non-interest. For Experiment 2, we separately modelled the events of interest (i.e., the 15-sec interval of AM, ToM, and VS) and the button-response interval at the end so that the results are not contaminated by response preparation or execution. Moreover, we included each participant's reaction times (RTs) of all active-task performance as parametric modulators, allowing us to rule out any brain activation driven by task difficulty or cognitive effort when assessing the effects of experimental manipulation.
Low-frequency drifts were removed using a high-pass filter of 128 sec. Contrast images from the individual-level (1 st -level) analyses were then submitted to random-effect models in the group-level (2 nd -level) analyses. present study, their coordinates in the MNI stereotaxic space, and location pinpointed on the MNI template by the yellow crosshair. Also rendered on the template is the meta-analysis outcomes, based on the NeuroSynth database, of brain regions related to default-mode network and semantic memory.

Supplemental Figures and Discussions
Behavioural analysis. In Experiment 1, participants showed comparable reaction times for the Self-(average±SD: 1071±35 ms) and Other-Referential (1056±39 ms) tasks, with no statistically reliable difference between them (paired t-test, p = 0.42). In Experiment 2, participants also showed comparable reaction times for the AM (807±36 ms) and ToM (832±44 ms) conditions, with them having no reliable difference (p = 0.39; note that these reaction times represent the latency to rate vividness after, rather than during, an AM-or ToM-interval). More importantly, the results of vividness ratings indicated that participants were able to engage in highly vivid recall and imagery (AM: 1.25±0.04; ToM: 1.28±0.05, both approached the ceiling-level on the given scale), while the ratings did not differ between contexts (p = 0.64). Together, these data showed that, in both experiments, the tasks specifically designed to probe the functionality of the default network are matched on the general cognitive effort required (as indicated by reaction times) and the clarity of introspective experiences (as indicated by the vividness ratings). Also note that in all fMRI analysis, reaction times (including those of the visuospatial tasks) were included as additional regressors to factor out their influences on the neural data. Figure S1. (A) Experiment 1 has three task-conditions: Self-Concept (assessing if the adjective fits one's own personality traits), Other-Concept (assessing if it fits the participant's impression for the Queen), Visuospatial (answering if the scrambled patterns are mirror-reversed images); Experiment 2 also includes three task-conditions: Autobiographical Memory (recalling life events related to the topic), Theory of Mind (imagining the feelings and thoughts of the person in the photograph), and Visual Search (looking for a tiny triangle); both Experiment 1 and 2 includes resting-state periods that are randomly interleaved with the task-conditions. (B) The locations of the ten ROIs in our analysis are rendered on a glass brain that highlights their relative position.
Colour coding: red indicates an ROI that tends to be affiliated with the core default network; blue indicates an ROI that tends to be affiliated with the semantic network. (pMTG); there is discrepancy between Experiment 1 and Experiment 2 in the 'diagnostic' regions that dissociate 'Network-A' and 'Network-B' (Braga & Buckner, 2017) at the dorsal PCC, the parahippocampal cortex (PHC), the anterior sector of the inferior parietal lobule (conventionally called the temporoparietal junction/TPJ). We discuss the potential driving factor that causes this discrepancy in the main article.
(B) Task-driven activities (Control > Rest) of Experiment 1 and Experiment 2 are represented using magenta and white, respectively; the areas of conjunction is pink. We observed similar patterns of neural activity driven by the mental rotation task (Experiment 1) and the visual search task (Experiment 2). These two visuospatial tasks activated a set of widely distributed 'task-positive' regions or the 'multiple-demand network' known to underpin spatial attention and executive control. As illustrated in Fig. S2, both tasks elicited robust activity of the frontal and parietal cortices, also the anterior cingulate cortex, that are known to support attention, executive control, and visuospatial working memory (for review, see Fedorenko, As discussed in the main article, core nodes of the default networkthe ventral section of the medial prefrontal cortex (vmPFC), the PCC, and the AGare less involved in this more externally-oriented task of personality assessment, as indicated by the absence of active cluster at these sites. This is in striking contrast with their robust activity in the internallyoriented task-settings during Experiment 2 (lengthy intervals during which participants were immersed in their internal thoughts). (iii) Finally, as a sanity checkthere is robust activity in these task-conditions in the bilateral visual cortices (as the stimuli were presented visually) and in the left sensorimotor cortex (as participants were responding using their right hand), indicating that the analysis are performed accurately that gives the correct neural signatures that ought to appear as a result of our experimental design. Figure S4. In this analysis we specifically focus on the medial temporal lobe (MTL), a neural structure known to be involved in a wide range of internally-directed contexts, such as autobiographical memory, resting-state cognition, episodic-retrieval processes, etc. We select three ROIs based on the peak coordinates of an influential study by Buckner and colleagues (Andrews-Hanna et al., 2010): the retrosplenial cortex/RSC: [-14, -52, 8]; the posterior parahippocampal cortex/PHC: [-28, -40, -12]; the hippocampal formation/HF: [-22, -20, -26].
We examine the neural responses (β-weights contrasted against the resting-state, like the approach used for the analysis in the main article) and compare between the task-conditions of autobiographical memory (AM) and theory of mind (ToM). Results reveal a statistically significant interaction of the three ROIs (RSC, PHC, and HF) × the two tasks (AM vs. ToM), F(1,23) = 27.21, p < 0.001, η p 2 = 0.54. To identify the source of the interaction, we perform post-hoc pair-wise comparisons. As illustrated in Figure S4, we found that neural responses to the AM and ToM conditions do not differ in the PHC and HF (both ps > 0.14). However, there is a significant difference in the RSC (p < 0.001), with RSC activation for the AM task being significantly greater than the ToM task. It is also evident in the figure that responses to both tasks are greater than the resting-baseline in every ROI. These results are consistent with the previous literature regarding the role of MTL in various introspective processes. It also underlines the role of the RSC in autobiographical memorybecause the AM task that we use encourages topic-guided recollection of life experiences (e.g., Recall how you felt when you just learnt the result of the Brexit referendum. Were you exhilarated or disappointed? What did you say to people about it?). This is a highly semantically loaded process (in the example here, understanding the meaning of referendum, exhilaration, and disappointment is essential). This may be particularly reliant on high-order mnemonic regions like the RSC that facilitates the creation of a meaningful mental scenario with combinations of relevant times, places, people, and sentiments.