Empathy and compassion

Understanding another person, sharing their feelings, and wanting to help them are different psychological responses. “Empathy” is used for several of these, so a study must state which response it measures. None alone demonstrates effective helping.

This distinction is central to the SAGE Lab’s work on contemplative and other prosocial interventions.

Define the response

TermWorking meaning in this primerWhat it does not establish
Perspective-takingDeliberately considering another person’s point of viewThat the inferred perspective is accurate.
Affective empathySharing an emotion while distinguishing another’s experience from one’s ownConcern, altruistic motivation, or action.
Empathic concernOther-oriented feelings such as sympathy and tenderness toward someone in needA sustained or impartial response.
Personal distressSelf-oriented discomfort when encountering another’s sufferingWhether relief comes through helping or disengagement.
CompassionConcern for suffering with a wish to alleviate itCompetence or successful benefit.

Terminology overlaps across traditions. Singer and Klimecki (2014) distinguish emotional sharing, distress, and compassionate concern. Goetz, Keltner, and Simon-Thomas (2010) examine compassion’s proposed affective and evolutionary profile. These are useful distinctions, not a universally adopted vocabulary.

Illustration: someone can understand a colleague’s worry without feeling it, feel distressed by it without knowing how to respond, or offer thoughtful assistance without sharing the same emotion.

What is the explanatory problem

Different accounts ask different questions. Empathy–altruism concerns whether other-oriented concern can produce a goal of benefiting another. Appraisal accounts examine how perceived need, responsibility, and coping shape emotion. Training accounts ask which responses can change and whether those changes affect behavior.

A correlation between reported concern and helping cannot identify the ultimate motive or prove that increasing concern will increase help. The altruism primer explains experimental comparisons between other-oriented and egoistic accounts.

How to measure it

Davis (1983) developed a multidimensional self-report approach that separates perspective-taking, empathic concern, personal distress, and fantasy. Subscales should not be treated as interchangeable indicators of one response.

Use separate items or tasks for the proposed processes. Distinguish a report about the present episode from a trait questionnaire about typical reactions. Check whether a “compassion” outcome actually asks about concern, distress, intentions, or past behavior.

Pair psychological outcomes with a relevant behavioral opportunity. Record what participants could do, its cost, and the beneficiary’s outcome. Physiological or brain measures can inform a process hypothesis, but do not replace behavioral or recipient evidence.

Eyal, Steffel, and Epley (2018) distinguish imagining another’s perspective from obtaining information directly. For perspective-taking, distinguish trying to understand from accurate understanding. When a recipient’s report is available, compare the inference with that report rather than assuming that imagining their perspective is successful.

Selected training evidence

Weng et al. (2013) found greater redistribution to a victim after compassion training than after reappraisal training in healthy young adults. The result concerns a consequential task and a specified comparison; it does not establish lifetime change or all forms of helping.

Böckler et al. (2018) found different effects of mental-training modules on task-based and self-reported prosocial outcomes. This supports examining the content of training and the outcome separately.

Kreplin, Farias, and Brazil (2018) found that conclusions about meditation’s prosocial effects depended on outcome and methodological features. Training labels, comparator quality, and follow-up matter when interpreting apparent benefits.

Questions that remain open

Does training broaden concern beyond selected recipients? Does it change distress, motivation, behavioral choice, or their relationship? Are effects retained outside the task and after practice support ends? When does concern lead to unsuitable assistance?

Avoid concluding that empathy is always helpful or always harmful. Also avoid treating mindfulness, loving-kindness, compassion, and perspective-taking as interchangeable interventions. Specify the practice, the proposed mechanism, and the outcome.

Use in SAGE Lab studies

For a proposed intervention, write a prediction that could fail: “this practice changes reported concern and actual helping under these conditions.” Include an appropriate comparison and distinguish the effect on each outcome. A mediation estimate by itself does not establish the causal psychological pathway.

To connect this work to robust prosociality, add the specific follow-up, recipient variation, or challenge needed for the robustness claim.

Essential reading: Singer and Klimecki for distinctions; Goetz et al. for compassion theory; Weng et al. and Böckler et al. for primary training studies; Kreplin et al. for methodological qualifications.

Next: intention to action · Important concepts