Examining Insula–Default Mode Network Functional Connectivity and Its Relationship with Heart Rate Variability

The insular cortex (insula) is central to interoception, emotional processing, cognitive control, and autonomic regulation, with strong connectivity to regions such as the amygdala, ACC, and prefrontal cortex. The DMN (mPFC, PCC/precuneus, lateral parietal cortex) supports self-referential cognition and is predominantly active at rest. Integrating insula-mediated interoceptive and salience processing with DMN self-referential functions is thought to facilitate emotional regulation by aligning bodily states with personal context. Prior work links HRV to autonomic and emotional regulation and to brain network activity, including insula and DMN connectivity, and insula structural measures have been associated with HRV. Research Problem: The combined relationship between insula–DMN functional connectivity and HRV remains underexplored. Hypotheses: (1) Insula functional activity and gray matter volume (GMV) would be related to HRV; (2) Insula activation would be functionally connected with the DMN; (3) Stronger insula–DMN connectivity would correlate with higher HRV.

Evidence indicates the insula integrates interoceptive signals with emotional and cognitive control, influencing heart rate and HRV via the central autonomic network. The DMN provides self-referential context for emotional experiences and may support autonomic balance. Prior neuroimaging shows HRV associations with resting-state functional connectivity (Chang et al., 2013) and that HRV, respiration, and heart rate affect DMN connectivity (Yoshikawa et al., 2020). Structural studies relate insula and central autonomic network GMV to HRV (Matusik et al., 2023; Wei et al., 2018). DMN–insula connectivity has been implicated across disorders (Hsiao et al., 2017; Tsurumi et al., 2020). Despite these findings, the joint contribution of insula–DMN connectivity to HRV had not been specifically tested prior to this study.

Design: Resting-state structural and functional MRI acquisition coupled with resting-state HRV measurement and correlational analyses. Participants: Forty-three healthy, right-handed Caucasian adults (23 males, 20 females; age 23–39 years; M = 31.14, SD = 4.54). Exclusion criteria: dementia, neuropsychiatric or neurodegenerative disorders; alcohol/drug dependence or abuse within past year; MRI contraindications (e.g., metallic implants, pregnancy); age <20 or >50. Ethics: Approved by Comissão Ética do CIPsi (CIPSI/CE/2013/02); informed consent obtained. MRI Acquisition: 3T Siemens Magnetom Tim Trio. Structural T1: 192 sagittal slices; TR = 2000 ms; TE = 2.33 s; flip angle = 7°; slice thickness = 0.8 mm; gap = 0 mm; pixel size = 0.8 × 0.8 mm²; FoV = 256 mm. Resting-state fMRI: 7 min; BOLD EPI; TR = 2000 ms; TE = 29 ms; flip angle = 90°; FoV = 222 × 222 mm; voxel size = 3 × 3 mm; 39 slices; 210 volumes; participants eyes closed, awake, still. Structural Preprocessing: FreeSurfer with manual verification (Talairach coregistration; skull stripping; cortical/subcortical boundaries). ROI GMV extracted for left and right insula; intracranial total volume computed. Functional Preprocessing: FSL v5.09. First five volumes removed; slice-timing correction; motion correction (rigid body); motion scrubbing based on frame displacement and DVARS; normalization to MNI space; regression of motion parameters and mean white matter and CSF signals; band-pass filter 0.01–0.08 Hz; spatial smoothing 8 mm FWHM; visual quality checks. Functional Connectivity Extraction: Time series for insula and DMN extracted using FSL fslmeants with region-specific masks; resulting BOLD time series used to compute insula–DMN connectivity; Fisher transformation applied to correlation values for normalization prior to HRV correlations. Cardiac Measures (HRV): BIOPAC system; abstinence from alcohol/nicotine/caffeine ≥4 h prior. Resting ECG recorded ~6 min at 1000 Hz using 3-electrode Lead-II configuration (right and left mid-clavicle, upper left shoulder). Skin cleaned; Ag–AgCl electrodes (EL-503). BioNomadix RSPEC-R wireless module with MP150; AcqKnowledge v4.4. ECG filtered (1 Hz IIR high-pass; 35 Hz low-pass); ectopic beats/artifacts visually inspected and corrected if needed. HRV computed in frequency domain using FFT per standards; HF-HRV extracted as average per individual. Statistical Analysis: Normality verified via Shapiro–Wilk (p > 0.05). Pearson correlations tested relationships between HRV and insula GMV (right, left, total, adjusted for intracranial volume), insula resting-state activity (mean time series), and Fisher-transformed insula–DMN connectivity. Significance threshold p < 0.05. Software: Jamovi v2.5.

- Insula–DMN functional connectivity: Significant positive correlation between insula and DMN time series (r = 0.493, p = 0.003, n = 35), indicating strong synchronous fluctuations during rest. - Insula morphometry and HRV: Significant positive correlation between left insula GMV and HF-HRV (r = 0.365, p = 0.016, n = 43; volumes adjusted for intracranial volume). Right insula GMV not significant (r = 0.017, p = 0.916, n = 43). Total insula GMV not significant (r = 0.202, p = 0.192, n = 43). - Insula resting-state activity and HRV: No significant correlation (r = 0.244, p = 0.158, n = 35). - Insula–DMN connectivity and HRV: No significant correlation between Fisher-transformed insula–DMN connectivity and HRV (r = 0.004; R² ≈ 0.002).

Findings partially support the hypotheses: the insula was strongly functionally connected with the DMN at rest, and left insula volume related positively to HF-HRV, suggesting potential lateralization in autonomic–emotional integration consistent with prior evidence of left anterior insula involvement in heart rate reduction and HF-HRV during emotion. However, stronger insula–DMN connectivity did not correlate with HRV, indicating that intrinsic coupling between these regions may not directly reflect parasympathetic cardiac regulation as indexed by HF-HRV. This dissociation implies that HRV is likely influenced by additional neural pathways (e.g., central autonomic network, hypothalamus, amygdala, brainstem) and immediate physiological factors (respiration, stress, hormonal status, sleep quality) that may not be captured by insula–DMN resting connectivity alone. The results emphasize that insula–DMN interactions may predominantly support higher-order self-referential and salience processes rather than immediate autonomic output. Methodological context suggests that resting-state measures might miss task-evoked dynamics more tightly coupled to autonomic changes, and the non-concurrent acquisition of HRV and fMRI may have obscured associations.

The study demonstrates robust resting-state functional coupling between the insula and DMN and identifies a positive association between left insula volume and HF-HRV, but finds no direct link between insula–DMN connectivity and HRV. These outcomes challenge the assumption that stronger insula–DMN coupling predicts higher parasympathetic regulation and underscore the complexity of brain–autonomic relationships. Future research should prioritize concurrent HRV–fMRI acquisition, task-based paradigms that elicit autonomic responses, longitudinal designs to capture temporal variability, and multimodal approaches (including salience and central autonomic networks). Potential applications include targeted biofeedback and real-time fMRI neurofeedback interventions aimed at improving emotional regulation by engaging insula–DMN and autonomic circuits.

- HRV and fMRI acquisitions were conducted on different days, risking temporal mismatch given rapid HRV fluctuations due to stress, sleep, physical activity, and hormonal factors. - Resting-state-only paradigm may lack specificity to detect task-evoked connectivity changes more tightly coupled with autonomic modulation. - HRV assessed as a single resting HF-HRV metric; immediate modulators (e.g., respiration) can influence HRV independently of observed resting connectivity. - Potential confounds such as respiration patterns, menstrual cycle, and sleep quality were not concurrently measured and controlled during fMRI/HRV recording. - ROI-based approach focused on insula and DMN may have missed contributions from other central autonomic and salience network regions.