A Distinct Communication Strategies Model of the Double Empathy Problem

The double empathy problem recasts the difficulty of forming empathy bonds in social interactions between autistic and neurotypical individuals as a bidirectional problem, rather than due to a deficit exclusive to the person on the spectrum. However, no explicit mechanism to explain such a phenomenon has been proposed. Here we build a feedback-loop mathematical model that would theoretically induce the empathy degradation observed during communication in neurotypical-autistic pairs solely due to differences in communication preferences between neurotypical and neurodivergent individuals. Numerical simulations of dyadic interactions show the model, whose mechanism is based solely on communication preferences, can illustrate the breakdown of empathic bonding observed clinically. Stability analysis of the model provides a way to predict the overall trajectory of the interaction in the empathy space. Furthermore, we suggest experimental designs to measure several parameters outlined here and discuss the future directions for testing the proposed model.

💡 Research Summary

The paper presents a novel mathematical framework to explain the “double empathy problem,” which describes the mutual difficulty autistic (A) and neurotypical (NT) individuals experience when trying to empathize with each other. Rather than attributing the breakdown solely to a deficit in autistic people, the authors model the interaction as a two‑dimensional feedback loop that operates on verbal (V) and non‑verbal (NV) channels of empathy. Each agent perceives the partner’s V and NV signals through a perception function p_i, weighted by individual parameters ϕ_i that reflect how much importance the agent places on each channel. The perceived empathy (RE_i) is compared to an expected empathy level (EE_i), producing an empathy gap Δ_i = EE_i – RE_i. This gap drives an internal “defensivity” state D_i via an increment function y_i, while a decay term λ_i models forgetting or emotional regulation. Defensivity then determines the next round’s output through separate output functions z_i,V and z_i,NV, closing the loop.

A key assumption is that autistic agents heavily weight verbal information and down‑weight non‑verbal cues, whereas neurotypical agents treat both channels more equally or even favor non‑verbal cues. This asymmetry creates a large initial Δ, which raises defensivity, reduces subsequent empathy outputs, and amplifies Δ in the partner—a positive feedback spiral that can lead to an “empathy collapse.”

The authors formalize the system’s dynamics in discrete time using the defensivity vector D = (D_A, D_NT). Linearizing around a candidate fixed point yields a Jacobian matrix J whose diagonal entries are (1‑λ_A) and (1‑λ_NT), and off‑diagonal entries are products of the sensitivities of perception to the partner’s defensivity (S_A←NT, S_NT←A) and the slopes of the increment functions (y′_A, y′_NT). The loop‑gain product L = y′_A·S_A←NT·y′_NT·S_NT←A quantifies the propensity for instability. Applying Jury’s criteria for discrete‑time systems, the authors show that stability requires 1 ± tr(J) + det(J) > 0 and 1 − det(J) > 0, where det(J) = (1‑λ_A)(1‑λ_NT) − L. If L exceeds the product of the damping terms, the determinant becomes negative, the eigenvalues leave the unit circle, and the system diverges, manifesting as rapid empathy degradation. Conversely, sufficiently low L or higher λ values produce a stable fixed point where empathy levels plateau at a functional level.

To bridge theory and data, the paper proposes experimental protocols for estimating the model’s parameters. Verbal weighting (ϕ) can be derived from linguistic analyses of literalness and precision; non‑verbal weighting from facial‑expression coding, prosody analysis, and eye‑tracking. Defensivity parameters (λ, ψ) could be inferred from physiological markers (skin conductance, heart‑rate variability) and self‑report scales of hostility or anxiety. The increment function y_i could be fitted using trial‑by‑trial changes in these measures as participants engage in controlled dyadic conversations. By fitting the model to empirical trajectories, researchers could predict whether a given pair will converge to a stable empathic state or spiral into collapse.

Overall, the study offers a mechanistic, quantitative account of the double empathy problem, demonstrating that divergent communication preferences alone can generate the observed breakdown in mixed‑neurotype dyads. The stability analysis provides a clear criterion for when interventions—such as training to increase attention to non‑verbal cues in autistic individuals or techniques to reduce defensivity in neurotypicals—might shift the system from an unstable to a stable regime. This work thus opens a pathway for integrating mathematical modeling with clinical research on neurodiversity, offering testable predictions and a framework for designing targeted empathy‑enhancement interventions.