A Visual Explainer

What Is Compassion?

An Evolutionary Analysis and Empirical Review

Informal summary based on the research of Jennifer L. Goetz, Dacher Keltner & Emiliana Simon-Thomas. For exact context, refer the original paper. 

The one thing to take from this paper

Compassion and empathic distress look similar from the outside — both arise when we witness suffering — but they are fundamentally different states, with opposite effects on the body and opposite effects on behavior. Distress turns attention inward and leads to avoidance. Compassion turns attention outward and motivates approach.

Science now shows that compassion is a biologically distinct emotion with its own evolutionary origins, its own physiological signature, and its own motivational logic — and that it can be deliberately cultivated.

Compassion has been central to major spiritual and philosophical traditions — from Buddhism and Christianity to Confucianism. And yet, until recently, science has barely studied it. It has been treated as a form of distress, a variant of sadness, or a subtype of love — rarely as its own thing. The evidence says otherwise.

In this landmark 2010 review, Jennifer Goetz, Dacher Keltner, and Emiliana Simon-Thomas set out to answer a deceptively simple question: What is compassion? Drawing on evolutionary theory, appraisal research, emotion science, and neuroscience, they make the case that compassion is a distinct emotion — with its own origins, its own triggers, its own signals, and its own physiological signature — that evolved specifically to motivate care for those who suffer.

Their definition: compassion is the feeling that arises when witnessing another's suffering, and that motivates a desire to help. That definition may sound simple. But the depth beneath it — and its difference from emotions we often confuse it with — is what the evidence reveals.

A Portrait of Compassion

Why Compassion Evolved: Three Evolutionary Arguments

Darwin himself called sympathy "the strongest of humans' evolved instincts." Early evolutionary thinkers were skeptical — how could an emotion that motivates costly care for others survive natural selection? Three converging lines of reasoning explain how.

The first argument centers on vulnerable offspring. Human babies are born more prematurely and remain dependent longer than any other mammal. This extraordinary dependency created evolutionary pressure for a caregiving system — and compassion, in this view, is the emotional engine of that system. The feeling that arises when we witness a crying infant, a wounded companion, or a suffering stranger is, at its core, an adaptation for protecting the fragile and the dependent. Across radically different cultures, caregiving behavior — soothing touch, skin-to-skin contact, specific vocalizations — has been reliably observed, and nonhuman primates most closely related to us show similar caregiving toward vulnerable conspecifics.

The second argument comes from mate selection. Compassionate individuals make better reproductive partners — more likely to invest resources in offspring, to remain in cooperative long-term bonds, and to provide physical care and protection. Research supports this: across cultures, warmth and kindness rank among the most desired qualities in a mate. Individuals high in trait compassion correlate with secure attachment styles, which in turn predict healthier child development. Over generations, the mate preferences of both males and females likely increased compassionate tendencies in the gene pool.

The third argument involves non-kin cooperation. In a world where survival depends on reciprocal alliances with people who are not relatives, compassion serves as a signal of trustworthiness and prosocial disposition. Compassionate individuals are more likely to be chosen as allies, more likely to cooperate, and more likely to enforce norms of fairness within groups. Children high in compassion-related traits enjoy richer friendship networks; adolescents high in agreeableness — which correlates strongly with compassion — are more accepted by peers. Compassion, then, is not merely a private feeling: it is a signal of trustworthiness that shapes who we choose as allies, mates, and cooperators.

What Triggers Compassion: The Appraisal Process

Emotions don't arise from events themselves — they arise from how we appraise events. The same situation can trigger very different emotions depending on how we evaluate it. Compassion has a distinctive appraisal profile shaped by three judgments, each constrained by evolutionary logic.

The first is self and goal relevance: we feel more compassion toward those who matter to us — family, friends, members of our group, people who share our values. This doesn't mean compassion is selfish; it means it is structured by the relational proximity that defined survival in our evolutionary past. Crucially, though, compassion also requires maintaining a clear self-other distinction — awareness that the other's suffering is "not one's own." Without this distinction, witnessing another's pain collapses into empathic distress rather than compassion.

The second appraisal concerns deservingness. Compassion is most likely when the sufferer is not to blame for their situation. A meta-analysis of 39 helping studies found that targets perceived as having high control over their suffering elicited less sympathy (r = -.45) and more anger (r = .52), while sympathy toward less controllable suffering was positively correlated with helping behavior (r = .42). This is not indifference to moral complexity — it is evidence that compassion is a finely calibrated response attuned to undeserved harm.

The third appraisal is coping ability — the individual's sense that they have the psychological resources to help. When we feel capable of responding, compassion arises. When we feel overwhelmed and unable to cope, we are more likely to experience distress or anxiety instead. This explains why emotional regulation capacity is so strongly associated with compassion: people who can regulate their own emotions can remain steady enough in the face of another's suffering to feel concern rather than alarm.

How Compassion Shows Up: The Body as Signal

Every emotion has a signal function — it communicates something to others. Anger signals a boundary has been crossed. Fear signals threat. Compassion signals: I see your suffering, and I am oriented toward you.

The facial expression of compassion involves a characteristic lowering and furrowing of the brows, a forward lean of the head, and a soft, sustained gaze toward the sufferer. This expression is recognizable, but it is also easily confused with sadness — the two share oblique eyebrow movements. Recognition studies find that compassion is identified from the face only about 30% of the time, compared to approximately 82% for sadness and 76% for happiness. Compassion's face, it seems, is subtler — and more context-dependent.

The most reliable channel for compassion, however, is touch. Soothing tactile contact — gentle, moderate-pressure touch of longer duration — appears to be the primary medium through which compassion is communicated and received. In studies where participants communicated twelve distinct emotions via touch on another person's forearm, observers identified sympathy/compassion at above-chance levels 48–57% of the time. Notably, compassion was recognized from touch far better than from the face (see chart in the original paper). This finding aligns with evolutionary theory: touch is the most developed sensory modality at birth, and soothing touch is central to the caregiving behaviors that compassion is thought to have evolved alongside.

The voice also carries compassion. Brief, non-word vocal bursts expressing compassion are recognized at above-chance levels, and identified as compassion, love, or gratitude — prosocial vocal states — about 47% of the time. Together, these findings suggest that compassion is not primarily a "face emotion." It is communicated most powerfully through movement, proximity, and touch — the same channels that evolved to soothe, protect, and connect.

What Compassion Feels Like — and Does to the Body

Factor analyses of subjective reports consistently reveal that compassion loads on a cluster of words including compassionate, sympathetic, moved, tender, warm, and softhearted — and that this cluster is entirely separate from distress-related words like alarmed, upset, disturbed, and perturbed. Sadness-related words load on a third, separate factor. In subjective experience, these are genuinely distinct states.

Crucially, compassion motivates approach, not avoidance. In a paradigm where participants receive an appeal for help but are offered an easy escape route, greater compassion leads to more helping even when escape is easy — whereas distress leads to less helping when escape is easy (because the distressed person simply removes themselves from the uncomfortable situation). Compassion keeps us oriented toward the other; distress pulls us back toward the self.

The physiological signature of compassion is perhaps the most striking finding. When people witness suffering and feel compassion, their heart rate decelerates. This is the signature of the parasympathetic nervous system — associated with outward attention, social engagement, and the calm that enables caregiving. Children who showed heart rate deceleration during compassion-inducing films were subsequently more willing to help and donate. In contrast, distress — and sadness — are associated with heart rate acceleration and increased skin conductance, reflecting sympathetic arousal.

This parasympathetic signature is linked to the vagus nerve, a branch of the nervous system thought to have evolved uniquely in mammals to support attachment and caregiving behaviors. Higher vagal tone — measured by respiratory sinus arrhythmia (RSA) — is positively related to trait-like compassionate responding, and elevated RSA during exposure to suffering predicts the self-reported experience of compassion. The body, it appears, has an ancient system specifically calibrated for care.

Not Distress. Not Sadness. Not Love.

Compassion vs. Distress: Distress is self-focused — it motivates the reduction of one's own discomfort. When witnessing suffering overwhelms our capacity to cope, attention turns back toward the self. Compassion is other-focused — it motivates reducing the other's suffering. The physiological signature makes this distinction visceral: distress accelerates the heart; compassion slows it.

Compassion vs. Sadness: Sadness arises from personal loss — something bad has happened to us. Compassion arises when something bad has happened to another. The appraisal processes are structurally different: sadness involves self-relevance of the negative outcome; compassion involves appraisal of an other's suffering as relevant to the self while maintaining awareness that it is not the self's own experience.

Compassion vs. Love: Love responds primarily to positive events — to the presence, affection, and good qualities of a cherished person. Compassion responds to suffering and negative events. Love has been associated with reduced amygdala activation and engagement of the orbital frontal cortex, fitting its more positively valenced core; the appraisal model of compassion, by contrast, predicts engagement of regions involved in detecting suffering, appraising deservedness, and coping — a structurally different neural profile, though direct comparisons remain to be studied. One intriguing possibility: love may modulate compassion — overriding normal blame appraisals in extreme cases of need, so that we would save a sibling even when we judge them responsible for their situation.

Compassion as a Moral Force — and a Trainable One

Compassion has been central to moral philosophy from Aristotle to Buddhism, and the empirical evidence supports its importance in moral life. Research shows that compassionate individuals endorse policies that reduce suffering for the vulnerable, show reduced punitive impulses toward wrongdoers, and are powerful drivers of volunteerism and altruistic action — including costly altruism that benefits non-kin with no expectation of reward. Compassion functions, in the authors' phrase, as a "guardian" of the moral domain of unjustified harm.

At the same time, compassion is not unlimited or unconditional. It is shaped by appraisals of blame, deservingness, and coping — and these appraisals have both individual and cultural dimensions. Cultures differ in how prominently compassion features in daily emotional life, who is considered most deserving of it, and how it is appropriately expressed. While some features of compassion appear universal (the caregiving response to vulnerable others, the role of blame appraisals), the forms it takes are culturally inflected.

Perhaps the most encouraging finding is that compassion is a state that can become a trait — and a trait that can be deliberately cultivated. Loving-kindness meditation practices, which involve systematically extending feelings of warmth and concern first to close others and gradually to all beings, have been found to shift resting brain lateralization toward the left frontal lobes (associated with approach motivation), increase overall well-being, and build social connection. Compassion appears to be both a state and a trainable trait — one whose cultivation has measurable effects on brain function, well-being, and social connection.

What this paper establishes is that compassion is not a soft feeling on the margins of human psychology. It is a biologically distinct, evolutionarily grounded, physiologically measurable state that evolved to motivate care for those who suffer — and one that has demonstrable effects on behavior, health, and moral judgment. Understanding what it actually is, how it arises, and how it differs from related states is the foundation for understanding how it can be strengthened.

Based on: Goetz, J.L., Keltner, D., & Simon-Thomas, E. (2010). Compassion: An evolutionary analysis and empirical review. Psychological Bulletin, 136(3), 351–374.

Prepared for the Born to Flourish community.