Abstract:Key PointsObservational and experimental studies dating back to the 1950s demonstrate that mammals spontaneously help distressed conspecifics. Research emphasizes the untrained, unrewarded nature of this behaviour, which is also biased towards familiar individuals, thus arguing against explanations that are exclusively based on associative learning or conditioning.The perception–action model extends an existing motor theory on overlapping representations to emotional phenomena; it states that observers who attend to a target's state understand and 'feel into' it through personal distributed representations of the target, the state and the situation. Easily observed manifestations of this mechanism are emotional contagion and motor mimicry, which have been demonstrated in many animals. In cognitive forms of empathy, the same representations are accessed from the top-down.Experiments on two common mammalian expressions of empathy — the consolation of distressed individuals and spontaneous assistance to those in need — support the crucial role of caught distress and arousal because these behaviours are suppressed by anti-anxiety medication and engage the same neuropeptide system that supports social attachment.The Russian-doll model seeks to arrange forms of empathy into layers that are built on top of each other — with the layers ranging from emotional contagion to more cognitive forms of empathy — in a functionally integrated whole based on perception–action processes. Perspective-taking is well developed in some non-human species, as manifested by theory-of-mind and targeted helping.One can segregate emotional and cognitive empathy (as well as felt and observed states) in the brains of observers, but all forms require some initial access to the observer's distributed, shared, personal representations of the target's state. At least in the initial phase of processing, this access helps to decode the target's state and provide subsequent processing with content and meaning, even if the shared state is not experienced, or is incomplete or inaccurate.Empathic pain does not usually include the peripheral sensation of the target's injury, but it can include sensory information when the stimuli and task instructions emphasize the specific nature of the feeling at the location of the injury.

Empathic Coupling of Homeostatic States for Intrinsic Prosociality

Artificial Empathy: A New Perspective for Analyzing and Designing Multi-Agent Systems

Reaching Cooperation Using Emerging Empathy and Counter-empathy.

Empathy Structure in Multi-Agent System with the Mechanism of Self-Other Separation: Design and Analysis from a Random Walk View

Promoting Constructive Interaction And Moral Behaviors Using Adaptive Empathetic Learning

Empathy: Its ultimate and proximate bases.

Modeling adaptive empathy based on neutral assessment: a way to enhance the prosocial behaviors of socialized agents under the premise of self-security

Social decision-making in a large-scale MultiAgent system considering the influence of empathy

EMOTE: An Explainable architecture for Modelling the Other Through Empathy

Empathy: A motivated account.

Building Altruistic and Moral AI Agent with Brain-inspired Affective Empathy Mechanisms

Reciprocity and Its Neurological Correlates in Human-Agent Cooperation

Brain-Inspired Affective Empathy Computational Model and Its Application on Altruistic Rescue Task

Adaptive Empathy: Empathic Response Selection as a Dynamic, Feedback-Based Learning Process

Empathy, mirror neurons and SYNC

An Integrated View of Empathy: Psychology, Philosophy, and Neuroscience

Emotional Empathy as a Mechanism of Synchronisation in Child-Robot Interaction

Learning to Balance Altruism and Self-interest Based on Empathy in Mixed-Motive Games

Mammalian empathy: behavioural manifestations and neural basis

Emergence of integrated behaviors through direct optimization for homeostasis