Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Identifying the Hierarchical Emotional Areas in the Human Brain Through Information Fusion

Zhongyu Huang, Changde Du, Chaozhuo Li, Kaicheng Fu, Huiguang He

TL;DR

The paper addresses the brain basis of emotion by arguing that emotions arise from distributed interactions among multiple brain regions rather than isolated loci. It introduces a theoretical framework and a novel method, the Hierarchical Emotion Network (HEmoN), which identifies hierarchical emotional areas by extracting a brain tree from fMRI-derived networks and maximizing information fusion along longest shortest paths. The approach combines trunk-based hierarchical area identification with LSTM-based trunk representations to enable cross-dataset emotion decoding, outperforming several baselines. These findings offer a more nuanced, multi-regional view of emotional processing aligned with the psychological constructionist hypothesis, with potential implications for affective neuroscience and applied emotion decoding.

Abstract

The brain basis of emotion has consistently received widespread attention, attracting a large number of studies to explore this cutting-edge topic. However, the methods employed in these studies typically only model the pairwise relationship between two brain regions, while neglecting the interactions and information fusion among multiple brain regions$\unicode{x2014}$one of the key ideas of the psychological constructionist hypothesis. To overcome the limitations of traditional methods, this study provides an in-depth theoretical analysis of how to maximize interactions and information fusion among brain regions. Building on the results of this analysis, we propose to identify the hierarchical emotional areas in the human brain through multi-source information fusion and graph machine learning methods. Comprehensive experiments reveal that the identified hierarchical emotional areas, from lower to higher levels, primarily facilitate the fundamental process of emotion perception, the construction of basic psychological operations, and the coordination and integration of these operations. Overall, our findings provide unique insights into the brain mechanisms underlying specific emotions based on the psychological constructionist hypothesis.

Identifying the Hierarchical Emotional Areas in the Human Brain Through Information Fusion

TL;DR

The paper addresses the brain basis of emotion by arguing that emotions arise from distributed interactions among multiple brain regions rather than isolated loci. It introduces a theoretical framework and a novel method, the Hierarchical Emotion Network (HEmoN), which identifies hierarchical emotional areas by extracting a brain tree from fMRI-derived networks and maximizing information fusion along longest shortest paths. The approach combines trunk-based hierarchical area identification with LSTM-based trunk representations to enable cross-dataset emotion decoding, outperforming several baselines. These findings offer a more nuanced, multi-regional view of emotional processing aligned with the psychological constructionist hypothesis, with potential implications for affective neuroscience and applied emotion decoding.

Abstract

The brain basis of emotion has consistently received widespread attention, attracting a large number of studies to explore this cutting-edge topic. However, the methods employed in these studies typically only model the pairwise relationship between two brain regions, while neglecting the interactions and information fusion among multiple brain regionsone of the key ideas of the psychological constructionist hypothesis. To overcome the limitations of traditional methods, this study provides an in-depth theoretical analysis of how to maximize interactions and information fusion among brain regions. Building on the results of this analysis, we propose to identify the hierarchical emotional areas in the human brain through multi-source information fusion and graph machine learning methods. Comprehensive experiments reveal that the identified hierarchical emotional areas, from lower to higher levels, primarily facilitate the fundamental process of emotion perception, the construction of basic psychological operations, and the coordination and integration of these operations. Overall, our findings provide unique insights into the brain mechanisms underlying specific emotions based on the psychological constructionist hypothesis.
Paper Structure (23 sections, 2 theorems, 22 equations, 4 figures, 1 algorithm)

This paper contains 23 sections, 2 theorems, 22 equations, 4 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related work
  3. Preliminaries
  4. Graph theory
  5. Psychological constructionist hypothesis
  6. Hierarchical emotional areas: Theoretical analysis and methodology
  7. Extracting the brain tree
  8. Theoretical analysis: Interactions and information fusion in the brain tree
  9. Identifying the hierarchical emotional areas
  10. Practical application: Emotion decoding
  11. Experiments
  12. Dataset description
  13. Dataset 1
  14. Dataset 2
  15. Implementation details
  16. ...and 8 more sections

Key Result

Theorem 1

Given a tree $T$, the node influence $I_T(u, v)$ of node $v$ on node $u$ after $k = \text{dist}(u, v)$ steps of the random walk is where the only path between nodes $u$ and $v$ is denoted as $(u, v_1, v_2, \cdots, v_{k-1}, v)$; $a_{v_iv_j}$ is the edge weight between nodes $v_i$ and $v_j$, and $\mathcal{N}(u)$ is the set of $u$'s 1-hop neighbors.

Figures (4)

  • Figure 1: The pipeline of identifying hierarchical emotional areas in the human brain. To investigate the brain mechanisms underlying emotion (i.e., emotion encoding), we first measure and collect the brain signals of subjects when they experience audio-visual multi-source emotional stimuli. Next, we construct a brain network using the collected fMRI signals and extract a brain tree from this brain network. Then, we decompose the entire brain tree into trunks at different levels (i.e., hierarchical trunks), with each trunk facilitating information fusion among brain regions. Following the decomposition, we revert these hierarchical trunks in the brain tree back to hierarchical emotional areas in the human brain. As a result, we complete the identification of hierarchical emotional areas on a given dataset. Finally, we use the proposed model, HEmoN, which builds on these identified hierarchical emotional areas, to perform cross-dataset emotion decoding on other challenging datasets.
  • Figure 2: The illustration of the identified hierarchical emotional areas, including (a) all basic emotions, (b) happiness, and (c) sadness. The legend at the top lists all 13 functional systems and one uncertain system, as proposed by the Power Atlas power2011functional. Each subfigure shows the corresponding brain tree and presents the emotional areas at the first three levels. The nodes are colored according to the legend, and the internal connections within the 1st-level, 2nd-level, and 3rd-level emotional areas are highlighted in green, blue, and red, respectively.
  • Figure 3: Emotion decoding results (measured by MAE, lower is better) on Dataset 2. The black error bars represent the standard deviation.
  • Figure 4: A brain tree (left, adapted from Figure \ref{['fig:pl']}) and the process of identifying its hierarchical trunks (middle and right).

Theorems & Definitions (6)

  • definition 1: Node Influence
  • Theorem 1
  • definition 2: Path Information
  • Theorem 2
  • proof
  • proof