The flexible strain sensors based on the textile substrate have natural flexibility, high sensitivity, and wide-range tensile response. However, the textile's complex and anisotropic substructure leads to a negative differential resistance (NDR) response, lacking a deeper understanding of the mechanism. Therefore, we examined a graphene textile strain sensor with a conspicuous NDR tensile response, providing a requisite research platform for mechanism investigation. The pioneering measurement of single fiber bundles confirmed the existence of the NDR effect on the subgeometry scale. Based on the in situ characterization of tensile morphology and measurement, we conducted quantitative behavior analyses to reveal the origin of tensile electrical responses in the full range comprehensively. The results showed that the dominant factor in generating the NDR effect is the relative displacement of fibers within the textile bundles. Based on the neural spiking-like tensile response, we further demonstrated the application potential of the textile strain sensor in threshold detection and near-sensor signal processing. The proposed NDR behavior model would provide a reference for the design and application of wearable intelligent textiles.