The Intrinsic Hierarchy of Self – Converging Topography and Dynamics

Yasir Çatal, M. Günay, Chunbo Li, Jijun Wang, HuiRu Cui, Wei Li, Georg NorthoffbioRxiv 2022 · DOI

Abstract

The brain exhibits an intrinsic hierarchy in its topography that converges with its dynamics. This study investigates whether such hierarchical organization applies to the brain’s inner core and higher-order functions like the "self." Using multiple fMRI datasets, we demonstrate a three-layer topography of self (internal, external, mental) that persists across resting and task states. This topographic hierarchy aligns with dynamic changes in neural signals, suggesting an intrinsic spatiotemporal organization of the self.

Introduction

Neuroscience has long recognized a "core-periphery" structure in the brain, where sensory regions (periphery) process fast, incoming data, while the default-mode network (core) handles slower, high-level cognition. However, it remained unclear if this same hierarchical logic applies to the "self." Previous meta-analyses suggested the self is processed in three nested layers: Internal (interoceptive), External (exteroceptive/proprioceptive), and Mental (self-referential/cognitive). This study sought to determine if this three-layer model is an intrinsic feature of the brain—meaning it exists even when we aren't actively performing a task—and whether the "where" (topography) of these layers converges with the "how" (dynamics).

Methods

The research team utilized fMRI data from two large, independent datasets (Shanghai, N=50; UCLA, N=130) to analyze the relationship between brain regions and signal patterns. They defined specific Regions of Interest (ROIs) to represent the three layers of the self based on prior meta-analytic data.

Table of Regions of Interest (ROIs) defining the three layers of self: Internal, External, and Mental.

To characterize neural dynamics, five distinct measures were calculated:

  • Power-Law Exponent (PLE): Measures the strength of slow vs. fast frequencies.
  • Autocorrelation Window (ACW): Measures the "neural timescale" or how long a region holds onto information.
  • Median Frequency (MF): The midpoint of the power spectrum.
  • Sample Entropy (SE) & Lempel-Ziv Complexity (LZC): Measures of information complexity and predictability.

Key Results

Spatiotemporal Convergence

The findings revealed a clear convergence between topography and dynamics. In both datasets, the Mental layer (the core) showed the highest PLE and ACW—indicating the slowest, longest integration—and the lowest SE and LZC (most predictable). Conversely, the Internal layer showed the exact opposite. This hierarchy was "carried over" from rest to task states regardless of the task type, confirming its intrinsic nature.

Task-Specific Modulation

While the hierarchy is stable, it is also malleable. Self-specific tasks (like memory retrieval) significantly altered the "gradient" or the steepness of the hierarchy compared to non-self tasks. Network analysis revealed that PLE, MF, and ACW form a central hub of high centrality (the "dynamic core"), while information-processing measures (SE, LZC) are more peripheral and flexible "foreground" measures.

Discussion

These findings support the "Spatiotemporal Neuroscience" hypothesis—the idea that time and space are the "common currency" that allows neural activity to transform into mental and psychological experiences. The existence of an intrinsic topographic hierarchy of self and its convergence with dynamics suggests that the self is a fundamental organizational principle of the brain.

Future research could explore how this intrinsic hierarchy is disrupted in clinical conditions. In disorders like depression or schizophrenia, where the "sense of self" is often fragmented, we might expect to see a breakdown in the convergence between the brain's topography and its dynamics.

Summary updated

Abstract

Abstract The brain can be characterized by an intrinsic hierarchy in its topography which, as recently shown for the uni-transmodal distinction of core and periphery, converges with its dynamics. Does such intrinsic hierarchical organization in both topography and dynamic also apply to the brain’s inner core itself and its higher-order cognitive functions like self? Applying multiple fMRI data sets, we show how the recently established three-layer topography of self (internal, external, mental) is already present during the resting state and carried over to task states including both task-specific and -unspecific effects. Moreover, the topographic hierarchy converges with corresponding dynamic changes (measured by power-law exponent, autocorrelation window, median frequency, sample entropy, complexity) during both rest and task states. Finally, analogous to the topographic hierarchy, we also demonstrate hierarchy among the different dynamic measures themselves according to background and foreground. Finally, we show task-specific- and un-specific effects in the hierarchies of both dynamics and topography. Together, we demonstrate the existence of an intrinsic topographic hierarchy of self and its convergence with dynamics.
BibTeX
@misc{catal2022intrinsic,
  author = {Yasir Çatal and M. Günay and Chunbo Li and Jijun Wang and HuiRu Cui and Wei Li and Georg Northoff},
  title = {{The Intrinsic Hierarchy of Self – Converging Topography and Dynamics}},
  year = {2022},
  doi = {10.1101/2022.06.23.497287},
  howpublished = {bioRxiv},
}
Figure