A mechanistic model of threat-driven respiration and its homeostatic recovery in anxiety and depression

Abstract

Anxiety and depression change how people breathe, often early in the illness, and wearable sensors now allow breathing to be monitored continuously. This work presents, to our knowledge, the first mechanistic model of how the brain's threat response sets breathing rate and depth and how fast breathing recovers after stress. A threat signal from the amygdala, shaped by stimulus magnitude and by cognitive appraisal (coping capacity and perceived stress), drives the breathing-control center that receives lung feedback: stretch receptors slow breathing and airway receptors speed it up. A simple electrical circuit stores the threat signal and expresses recovery as one time constant. We prove the model's key properties, test them under noise and parameter variation, and compare it with chest-belt recordings from 15 adults in the Wearable Stress and Affect Detection (WESAD) study. Breathing rate rose from 18 to 28 and 31 breaths per minute from normal to moderate to severe, an order that held in every test. Recovery took about 3, 10, and 50 seconds, paced by how long the airway-receptor signal lingers in the brainstem, and the circuit followed breathing closely (correlation 0.96-0.98). In WESAD, a spoken stress task raised anxiety and breath depth (p<0.001); speech set the breathing rhythm during the task, and breathing then returned to rest, as the model predicts for healthy people. A compact, provable model links clinical stress measures to the breathing signals that wearables record. It offers measurable targets for monitoring anxiety and depression: how strongly threat drives respiration and how quickly homeostatic regulation restores it.