[System Dynamics] C-EMO: Modeling the "Heartbeat" of Collaborative Networks through Organizational Emotions

TL;DR

Can a company feel "frustrated" or "excited"? This paper argues that while organizations aren't sentient, their interactions within a Collaborative Network (CN) produce emotional dynamics that mirror human psychology. By combining Agent-Based Modeling (ABM) and System Dynamics (SDM), the authors introduce the C-EMO framework to simulate and monitor the emotional health of organizations, providing a "human-tech" friendly way to ensure long-term collaboration sustainability.

Problem & Motivation: The Hidden Social Friction in CNs

Collaborative Networks—groups of independent organizations like SMEs working together for a common goal—often fail not because of technical glitches, but due to social and organizational "complexity."

Current management systems treat these nodes as cold, rational entities. However, these organizations are run by humans whose expectations, satisfaction, and stress levels affect the entire network's performance. The challenge addressed here is: How do we quantify these "collaborative emotions" without being intrusive or violating the privacy of the participating members?

Methodology: The Architecture of Feeling

The authors treat each network member as an Individual Member Agent (IMA). The "brain" of this agent is divided into four systems:

1. Perception System: Gathers stimuli (e.g., net income, VO invitations).
2. Internal System: Keeps track of the agent's history.
3. Emotion System: The core engine that calculates valence and arousal.
4. Behavior System: Translates emotion into actionable responses.

The Russell's Circumplex Model DNA

Instead of complex, subjective adjectives, the model relies on Valence (pleasure vs. displeasure) and Arousal (activation vs. apathy).

• Excitement: High Valence + High Arousal
• Contentment: High Valence + Low Arousal
• Frustration: Low Valence + High Arousal
• Depression: Low Valence + Low Arousal

The logic is implemented using System Dynamics (SDM) causal loops, where successful events (like high performance evaluation) feed into a "Stock" of positive valence, while time-delays simulate how emotions naturally decay or persist.

Experiments: Simulating a 100-Day Corporate Lifecycle

To test C-EMO, the authors simulated "Company A" over 100 days. They fed the agent different scenarios:

• Days 0-15: High performance and met expectations led the agent into a state of Contentment.
• Days 15-30: Receiving invitations to form new Virtual Organizations (VOs) boosted arousal, pushing the state into Excitement.
• Days 50-70: A drop in satisfaction and performance evaluation plummeted the agent into Depression.

The simulation proved that organizational "emotions" could be modeled as dynamic trajectories rather than static labels, providing a predictive look at when a member might become uncooperative or leave the network.

Critical Analysis & Conclusion

The C-EMO framework moves beyond the "black box" of organizational behavior. Its strongest contribution is the non-intrusive nature of the data collection—using existing business metrics (net income, communication frequency) to infer emotional states.

Takeaways

• Hybrid Power: Combining ABM (for entity representation) with SDM (for continuous state change) is a superior way to model social-technical systems.
• Quantifiable Soft Skills: By mapping emotions to a 2D coordinate system, "soft" factors like satisfaction become "hard" data for administrators.
• Future Work: The authors plan to expand this into a "Collective Emotional State" for the entire network, effectively creating a "mood ring" for global supply chains or industry clusters.

Limitations: The model currently lacks real-world calibration data—most parameters are based on logical assumptions rather than empirical historical data. Furthermore, the "decay rate" of an organizational emotion (how long a company stays "angry" after a bad deal) needs more rigorous psychological grounding.