Social Intelligence

Social Intelligence

Linas Juozenas
Intelligence Unleashed · Minds in relationship

The intelligence between minds

Social intelligence is the capacity to notice social information, reason about perspectives and relationships, communicate clearly, cooperate, respect boundaries and repair mistakes. It can grow through experience and deliberate practice. It complements—never replaces—general intelligence, deep expertise or wisdom.

AttentionPerspectiveCommunicationCooperationBoundariesRepair
SelfAttention, emotion, motive and regulation
OtherWords, perspective, needs and uncertainty
RelationshipTrust, history, reciprocity and boundaries
ContextCulture, power, role, setting and medium
The essential idea

Understanding is a disciplined process

People do not broadcast their inner lives as simple codes. A pause might mean thought, uncertainty, fatigue, caution or nothing unusual. A skilled response notices the cue, considers several explanations and checks instead of declaring certainty.

Social intelligence is therefore not mind-reading. It is a cycle of attention, inference, conversation, feedback and correction.

What it is not

Charm is not the same as understanding

Extroversion, popularity, rapid speech, eye contact and confidence can shape impressions, but none proves accuracy, compassion, integrity or deep thought. Quiet, direct, introverted, autistic or culturally unfamiliar communication can be socially valuable without resembling a conventional performance of charisma.

01 · A useful working definition

Social intelligence is several abilities working together

No single test or biological marker defines it. The term is most useful for perceiving social information, understanding relationships and choosing effective, ethical action.

Edward Thorndike used the term in 1920 for understanding people and acting wisely in human relations. Later work separated social understanding, memory, behavioural flexibility, emotional skill and cultural adaptation.2 Modern instruments sample different slices, not one pure “social quotient.”

01

Perceive

Attend to words, timing, tone, action, context and change. Notice when a channel is incomplete or a familiar cue may mean something different.

02

Understand

Represent possible knowledge, intentions, feelings and constraints without confusing inference with fact. Track relationships, norms, roles and power.

03

Act and update

Choose language, timing and boundaries; notice impact; accept correction; and repair when result differs from intention.

Social skill is not automatically moral character

The same ability to read an audience can comfort, coordinate, deceive or exploit. Ethical SI joins accuracy with consent, honesty and respect for another person’s freedom to disagree or leave.

How can it be measured?

Self-reports describe perceived skill; performance tests sample problem-solving; informants report recurring behaviour; observations sample action in context. Every method adds information and bias.

The 21-item Tromsø scale reports perceived social information processing, skills and awareness.1 It is not an objective meter of a whole person. Confidence may also diverge sharply from skill-test performance.3 Ask: which capacity, in which setting, measured by whom, for what purpose?

02 · Related, not interchangeable

IQ, emotional intelligence, social intelligence and wisdom answer different questions

Labels overlap, yet flattening them into one ranking obscures what a person can actually do. Strong social reasoning can amplify intellectual work; it cannot manufacture the knowledge or abstraction that difficult work requires.

What each construct emphasizes—and what it does not establish by itself
Construct Central question Often assessed through What it does not prove
General intelligence / IQ How effectively does a person learn, reason, detect structure and solve novel problems across domains? Standardized cognitive tasks sampling several abilities Kindness, integrity, every talent, all knowledge or social judgement
Emotional intelligence How well does a person perceive, understand and manage emotion-related information? Ability tasks, self-report or observer ratings A single agreed trait, virtue or complete social competence
Social intelligence How well does a person understand social situations and respond constructively? Self-report, situational tasks, informants and observed behaviour Popularity, conformity, universal effectiveness or moral intent
Cultural intelligence How effectively does a person learn and adapt across culturally unfamiliar settings? Knowledge, metacognition, motivation and behavioural-flexibility measures Perfect knowledge of any group or permission to stereotype
Personality What relatively stable patterns characterize feeling, motivation and behaviour? Self- and observer-report inventories Maximum ability or inevitable behaviour in every situation
Wisdom How are knowledge, uncertainty, values, consequences and human welfare integrated in judgement? Reasoning scenarios, life narratives and multidimensional scales Omniscience or freedom from bias and error

General intelligence matters—profoundly

General cognitive ability is associated with learning, reasoning and performance across many complex tasks.4 It reflects both biology and development, and education can produce meaningful gains in measured intelligence.5 Scores are estimates whose relevance depends on task and context. Still, stronger performance in what a valid test measures can be unusually important when society needs someone to comprehend complexity, learn faster, anticipate consequences or create a solution that does not yet exist.

Social intelligence adds a transmission system: it helps insight be heard, challenged, protected, coordinated and translated into shared action. It should not be used to belittle IQ, and IQ should not be used to excuse cruelty. The strongest human achievements often join deep individual cognition with trustworthy cooperation.

03 · From cue to correction

Use a social-reasoning loop, not a snap verdict

Cues are evidence, not certainty. The most reliable social thinker remains willing to learn that the first interpretation was wrong.

  1. NoticeWords, behaviour, context and change
  2. Interpret lightlyGenerate more than one explanation
  3. Ask or actChoose the least presumptive useful response
  4. Observe impactWatch what clarifies, helps or harms
  5. Update and repairRevise the model and correct mistakes

Observation

“You became quiet after the schedule changed.” This describes something available to both people. It leaves room for correction.

Unverified story

“You are quiet because you are angry with me.” This may be right, but grammar can disguise an inference as knowledge.

Perspective-getting

“I may be reading this incorrectly. What changed for you?” Across a large series of experiments, simply imagining another person’s viewpoint did not reliably improve accuracy; obtaining information from them did.7

Calibrated response

“Would you like me to listen, help solve it, give you space—or something else?” A choice can be caring without assuming that one style fits everyone.

The socially intelligent sentence is often not “I know exactly how you feel,” but “Here is what I think I heard—what have I missed?”
Accuracy grows through contact with the person, not confidence alone
04 · Context changes meaning

Read the interaction at four levels

A behaviour can mean one thing between friends, another inside a hierarchy and something else in a public digital space. Social judgement improves when attention widens beyond the individual.

Person to personGoals, attention, emotion, history, communication preferences and the immediate task.
GroupRoles, alliances, status, norms, audience effects, inclusion and who receives airtime or credit.
CultureLearned expectations about directness, time, hierarchy, privacy, emotion and obligation—without treating a group average as an individual identity.
System and mediumRules, incentives, unequal power, interface design, surveillance and which cues a channel preserves or removes.

Culture should generate questions, not verdicts

Cultural intelligence research emphasizes metacognitive awareness, knowledge, motivation and behavioural flexibility across unfamiliar settings.6 It does not justify reducing whole countries to “high-context” or “low-context” personalities. Within every society, communication varies by region, family, profession, generation, language, neurotype and situation.

  • Name your own norm. “I tend to value direct feedback; how do you prefer to receive it?”
  • Ask about meaning. Do not assume that silence, eye contact, disagreement or punctuality carries one universal message.
  • Adapt without impersonating. Respect does not require caricature or abandoning your own boundaries.
  • Notice power. Apparent agreement may be compliance when refusal carries a cost.

Power is part of the data

A manager and employee, clinician and patient, adult and child, platform and user do not meet with equal freedom. The person with more power carries more responsibility to invite dissent, explain consequences, avoid retaliation and make “no” genuinely possible. Reading compliance as trust is a serious social error.

05 · The social brain

No single region contains empathy or social intelligence

Social understanding emerges from distributed systems that represent other minds, bodily feeling, language, memory, value and control. Their contributions overlap, change with the task and cannot be read as a personality scan.

MentalizingInferring beliefs, knowledge, intentions and perspectives
Affect and salienceDetecting significance and representing bodily or emotional states
Action observationRepresenting seen movements, goals and familiar actions
Control and regulationHolding context, inhibiting impulses and changing strategy
Language and learningCommunicating, remembering outcomes and updating predictions

Mentalizing

Sometimes called theory of mind or cognitive empathy, mentalizing means working out what someone may know, believe, intend or feel. Meta-analyses find recurring participation of medial prefrontal and temporoparietal systems, but different tasks recruit partly different patterns.9

Affective empathy

Sharing or resonating with another person’s affect is not identical to inferring their belief. People can be stronger in one than the other, and both differ from compassion: concern for welfare plus motivation to help.10

Live interaction

Real social exchange is reciprocal. Each person changes in response to the other, so studying passive photographs or stories captures only part of what brains do during conversation, coordination and mutual attention.11

Mirror neurons are not an empathy switch

Neurons that respond during both an action and observation of an action were first recorded cell by cell in macaque premotor cortex. A small human study later recorded some neurons with mirror-like properties in clinical participants.12 Most human imaging measures activity across much larger action-observation systems; fMRI cannot identify individual mirror neurons by itself. These systems may contribute to action understanding, but they do not explain empathy, morality, compassion or social intelligence on their own.

Neurochemistry is contextual, not a set of social potions

Oxytocin is often marketed as a “love hormone,” yet human effects vary with the person, relationship and situation; it does not simply make everyone trusting or benevolent.13 Dopamine participates in learning, motivation and the updating of value. Social approval and rejection can engage valuation and learning systems, but there is no single dopamine reading for social intelligence. Similar caution applies to claims about endorphins, synchrony and bonding: an indirect proxy is not proof of one molecule causing a complex relationship.

Plasticity is real; transfer is the hard question

Training in socio-affective or socio-cognitive practices has been accompanied by skill-specific behavioural and structural changes in adults.14 That does not mean every workshop “rewires” a person, that a trained task automatically improves real relationships, or that change is unlimited. Durable growth depends on the precise skill, feedback, context, motivation, practice and opportunity to use it.

06 · Learnable components

Train behaviours you can observe, not a vague “SQ”

Specific interpersonal skills can improve through instruction, modelling, deliberate practice, feedback and use in real relationships. Progress is more credible when it appears across settings and is visible to more than the learner.

A

Attend fully

Remove avoidable distraction, let the person finish and notice when you begin rehearsing a reply. Attention is not staring: eye contact, camera use and stillness have different meanings and accessibility needs.

B

Ask openly

Use questions that permit an unexpected answer: “What matters most here?” “How did you reach that view?” “Do you want listening, ideas or action?”

C

Paraphrase and verify

Summarize content in your own words and ask whether it is accurate. In an experimental initial interaction, active-listening responses helped speakers feel more understood than advice or simple acknowledgement; the effect was not a magical doubling of empathy.8

D

Regulate before responding

Notice activation, slow the pace and postpone high-stakes decisions if needed. Regulation is not suppression; it creates enough room to choose rather than discharge an impulse.

E

State a clean boundary

Name the behaviour or request, your limit and what you will do: “I can discuss the decision, but I will pause the conversation if insults continue.” A boundary guides your participation; it is not a threat to control another person.

F

Give usable feedback

Describe a specific action, its observable effect and a possible next step. Separate evidence from interpretation, and give the recipient space to explain context or disagree.

G

Repair precisely

Name what you did, recognize impact, apologize without demanding forgiveness, correct what can be corrected and change the repeated behaviour. Intention matters, but it does not erase effect.

H

Switch channels wisely

A short message suits logistics; ambiguity or conflict may need a richer, calmer medium. Offer options when voice, video, synchronous conversation or written language creates an access barrier.

I

Review real outcomes

Ask: Did interruptions decrease? Are decisions clearer? Can people safely correct me? Did the change last? Combine self-reflection with behaviour, feedback and consequences.

A meta-analysis of randomized empathy-training studies found a moderate average improvement, with variation among programs and signs that publication bias may inflate the estimate.15 This supports teachable components—not every branded exercise, a universal dose or permanent improvement in every relationship.

A charter for intellectual respect

Use social intelligence to recognize and protect exceptional minds

Intelligence is not decoration or a product from a shelf. It develops through inherited potential, development, education, effort, experience and years of learning. Preserving it can require health, safety, autonomy, challenge and protection from environments that punish difference.

Exceptional ability and work can matter enormously to a problem, community, generation or humanity. No score captures every form of importance or should ration dignity and safety. Yet society should not deny the value of rare reasoning, foresight, knowledge and creation. Some insights cannot simply be replaced by more tools, more people or louder confidence.

Listen beyond polishJudge the evidence and idea, not charisma, small talk, accent, status or conformity.
Protect deep workAllow uninterrupted time, privacy and freedom from constant social performance.
Protect the personTake health, safety, overload, bullying, coercion and exploitation seriously.
Supply real supportProvide education, tools, resources, fair compensation and complementary collaborators.
Preserve independenceDo not demand obedience, self-erasure or convenient opinions in exchange for belonging.
Credit and celebrateName the creator, reward the work and help a valuable contribution reach people who need it.

Do not romanticize suffering

Originality does not need deprivation, humiliation, addiction, chaos or neglect. Avoidable harm destroys concentration, memory, health and unfinished work. Protecting an exceptional mind protects a life, hard-earned knowledge and future work.

Do not confuse friction with failure

High IQ does not guarantee empathy or ethics. Yet low social polish does not negate intelligence, and unconventional communication does not prove arrogance or incapacity. Socially intelligent groups translate across styles and make room for quiet, exacting and neurodivergent contributors.

Universal dignity sets the floor. Recognition of rare ability raises our responsibility to protect what a mind has built—and what it may still give.
Equality of protection does not require denial of excellence
07 · Originality and belonging

Solitude can protect the idea; community can help it live

Healthy social intelligence does not demand constant company. It allows thought to step away and exchange to return when either is wanted.

Chosen solitude

Space without immediate direction

When other people continually instruct, evaluate or redirect attention, their priorities can occupy the space where a different question might emerge. Time alone can protect curiosity, dissent and unfinished associations.

An undemanding interval conducive to mind-wandering can sometimes support creative incubation on a prior problem.18 Brainstorming research also identified production blocking: waiting to speak can interrupt idea generation.19 Neither proves that isolation creates genius; both show why continuous interaction is not always best for original thought.

Chosen return

People who test, build and celebrate

Others may contribute before, during or after an idea forms. They can supply knowledge, detect error, challenge assumptions, build, fund, protect and carry an insight farther than one person could.

Returning to trustworthy people should be a celebration, not a surrender. A mature community welcomes difference and helps an original mind stay intact. The aim is freedom to move between reflection and relationship.

  1. Withdraw
  2. Explore
  3. Return
  4. Test
  5. Build & share
  6. Rest

Chosen solitude is not unwanted loneliness

Solitude can be restorative when it is voluntary, safe and connected to the possibility of return. Unwanted isolation, exclusion and chronic loneliness are different. Prospective evidence links stronger social relationships with better survival, although such findings do not prove that one social-skills exercise causes longer life.17 Balance must be personal: some minds need more quiet, some need more contact, and the same person may need both at different times.

08 · Relationships and teams

Trust is accumulated evidence, not instant chemistry

Trust often grows through repeated experiences of reliability, competence, care, honesty and repair. There is no universal intimacy sequence, and closeness should develop at a pace chosen by everyone involved.

Reliability

Do words and actions match over time? Are commitments renegotiated before they are broken?

Competence

Can this person perform the relevant task, know their limits and seek help when needed?

Care

Do they consider effects on others, protect vulnerability and respond without exploitation?

Integrity

Are principles consistent when honesty is inconvenient or power could hide a violation?

Psychological safety is candour without humiliation

In teams, psychological safety is a shared belief that interpersonal risk—asking a question, reporting an error or challenging an assumption—is permitted.16 It is not comfort at all times, automatic agreement or low standards. High standards become more useful when people can admit uncertainty early, question a popular leader and protect the quiet expert or inconvenient truth-teller.

When conflict begins

  • Reduce speed and audience pressure.
  • Describe observable behaviour and its effect.
  • State the need, decision or boundary clearly.
  • Invite the other account without surrendering facts.
  • Agree on a concrete next action and time to revisit it.

When repair is possible

  • Acknowledge what happened without minimizing.
  • Separate explanation from excuse.
  • Ask what correction would be meaningful.
  • Restore resources, credit or safety where possible.
  • Let repeated changed behaviour carry the apology.

Belonging must not require self-harm or self-erasure

No friendship, team, celebration or initiation should require intoxication, unwanted touch, private disclosure, constant availability, humiliation or abandoning a clear boundary. Alcohol is a psychoactive drug; cultural familiarity does not make pressure to use it harmless. Social intelligence respects an unembellished “no.”

Repair is valuable when there is safety and genuine accountability. It is not an obligation to remain in a coercive, abusive or dangerous relationship. A person may pause or end contact without presenting a case persuasive enough for the other party.

09 · Digital settings and AI

Choose technology for the relationship—not the reverse

Digital channels can extend connection, access and reflection. They also remove context, preserve messages indefinitely and give platforms unusual power over attention, visibility and inference.

Match medium to meaning

Use text for records, asynchronous thought and accessible pacing. Use voice or video when tone and rapid clarification help. Use in-person contact when shared physical context matters. Offer alternatives: no channel is universally richer for every body, language or neurotype.

Do not diagnose from metadata

Response time, camera use, typing speed, punctuation, emoji choice and gaze can reflect workload, disability, language, technology, safety or preference. They are not transparent measures of interest, status, deception or emotion.

Make remote work intentional

In one large company during the pandemic, firm-wide remote work was associated with more static and siloed collaboration patterns.21 Another set of paired tasks found videoconferencing reduced idea generation but not idea selection.20 These results identify design problems, not a universal verdict against remote work.

AI can assist a conversation; it cannot know a private mind

A language model may help someone rehearse, summarize or try alternative wording. It can still miss history, culture, irony, power and the meaning known only to the people involved. Treat output as a draft or hypothesis, disclose consequential use when appropriate and keep humans responsible for decisions.

Emotion recognition is a high-risk promise

Facial movements do not map reliably onto specific inner emotions across people and contexts.22 Systems that label a worker, student or applicant from face, voice or behaviour can turn uncertain inference into surveillance, discrimination and punishment—especially when refusal is impossible.

Governance must be concrete

Ask what data are collected, why, for how long, who can access them, how error is tested and whether a person can appeal. NIST’s AI Risk Management Framework emphasizes validity, reliability, transparency, privacy, fairness and accountability.23 In the EU, the AI Act prohibits AI systems that infer emotion from biometric data in workplaces and education institutions, with limited medical or safety exceptions.24

Persuasion at scale changes the ethical stakes

Personalized interfaces and generated messages can adapt pressure to a person’s traits, timing and vulnerabilities. Ethical design makes sponsorship and intent visible, permits refusal, avoids fabricated urgency, minimizes data and does not hide the path to leave. Social intelligence should protect autonomy from dark patterns—not optimize their power.

10 · Learning and difference

Build reciprocal understanding, not compulsory sameness

Education can teach attention, emotion knowledge, cooperation and conflict skills. The goal should be greater agency and mutual understanding—not a polished performance that hides distress.

Structured learning can help

A major meta-analysis of 213 school-based social and emotional learning programs reported improvements across skills, attitudes, behaviour and academic performance, including an average 11-percentile-point gain in achievement.25 Newer synthesis across countries also finds benefits alongside substantial variation in setting, implementation and outcome.26

Difference is not a failed norm

Autistic and non-autistic people may misread one another in both directions—the “double-empathy” problem.27 Support should be voluntary, co-designed and aimed at the person’s own goals: access, clearer mutual communication, safer relationships or less exhausting participation—not forced masking.

Context can disable or enable

Language, sensory load, trauma, hearing, vision, mobility, anxiety, unfamiliar rules and unequal power all shape social behaviour. Offer written options, processing time, predictable agendas, explicit expectations and ways to participate without performance theatre.

Clinical care is individualized

Changes in social cognition can accompany many developmental, neurological or psychiatric conditions, but a score or awkward encounter is not a diagnosis. Assessment requires history, context, multiple methods and qualified interpretation. Exercises in this article are education, not treatment.

11 · Direction matters

Ethical influence preserves another person’s agency

Communication always influences. The ethical question is whether it helps people understand and choose—or secretly engineers compliance.

Respectful influence

  • Makes purpose and relevant interests visible
  • Uses reasons a person can examine
  • Welcomes questions and informed disagreement
  • Accepts refusal without hidden punishment
  • Protects privacy and minimizes collected data
  • Seeks mutual or openly stated benefit

Manipulation

  • Conceals the real aim or sponsor
  • Targets fear, dependency or private vulnerability
  • Creates false urgency or fabricated consensus
  • Extracts disclosure, closeness or consent through pressure
  • Uses surveillance the person cannot meaningfully refuse
  • Penalizes resistance while pretending choice was free

Mirroring, “reading” micro-expressions or memorizing scripts does not grant authority over another mind. Use any conversational scaffold transparently. Vulnerability is voluntary, intimacy is not a prize for technique, and the other person remains the primary source on what they mean.

12 · Deliberate practice

A four-week experiment in better interaction

Choose one real behaviour and one setting. “Interrupt less in project meetings” can be observed and improved; “become socially brilliant” cannot.

ObserveTrack one situation. Separate cue, interpretation, action and result. Write two alternative explanations before judging.
Listen and verifyAsk an open question, paraphrase the answer, then ask what you missed. Do not add advice unless it is wanted.
Boundary and repairPractise one clear limit and one precise apology. Role-play, receive behavioural feedback and retry.
TransferUse the skill in a second context. Compare your view with outcomes and trusted observers; keep what genuinely helps.
  • Measure behaviour and impact, not only confidence.
  • Practise with consent; people are not training equipment.
  • Expect context-specific progress before broad transfer.
  • Preserve your own limits while learning flexibility.
13 · Evidence and perspective

Keep the construct useful by keeping it honest

Social intelligence is multidimensional, cognitively demanding and context-bound. Specific skills can improve, but no universal “SQ,” brain image or app can rank a whole relationship. Confidence is not calibration; look for behaviour, transfer across settings, durable outcomes and willingness to be corrected.

The complete picture

Understand people without shrinking the mind

Social intelligence turns separate capacities into relationship: ask rather than presume, coordinate without domination, protect boundaries, repair harm and recognize value beyond charisma.

Give an original mind room to wander. Give cultivated work recognition proportionate to what it contributes. Then build communities capable of testing ideas, correcting error, sharing work and celebrating excellence—while preserving every person’s dignity and freedom.

Evidence library

Sources and further reading

Foundational studies, major reviews and official guidance supporting the distinctions and cautions in this guide.

  1. Silvera, Martinussen & Dahl. The Tromsø Social Intelligence Scale (2001).
  2. Weis & Süß. Reviving the search for social intelligence (2007).
  3. Heck et al. A robust negative relationship between self-reports of social skill and performance (2024).
  4. Deary. Intelligence (2012).
  5. Ritchie & Tucker-Drob. How much does education improve intelligence? (2018).
  6. Ang et al. Cultural intelligence: measurement and effects (2007).
  7. Eyal, Steffel & Epley. Perspective mistaking: getting perspective, not taking perspective (2018).
  8. Weger et al. The relative effectiveness of active listening (2014).
  9. Schurz et al. Toward a hierarchical model of social cognition (2021).
  10. Kanske et al. Are strong empathizers better mentalizers? (2016).
  11. Redcay & Schilbach. Using second-person neuroscience to elucidate social interaction (2019).
  12. Mukamel et al. Single-neuron responses during action execution and observation in humans (2010).
  13. Bartz et al. Social effects of oxytocin: context and person matter (2011).
  14. Valk et al. Structural plasticity of the social brain (2017).
  15. van Berkhout & Malouff. The efficacy of empathy training: a meta-analysis (2016).
  16. Edmondson. Psychological safety and learning behavior in work teams (1999).
  17. Holt-Lunstad, Smith & Layton. Social relationships and mortality risk (2010).
  18. Baird et al. Mind wandering facilitates creative incubation (2012).
  19. Diehl & Stroebe. Productivity loss in brainstorming groups (1987).
  20. Brucks & Levav. Virtual communication curbs creative idea generation (2022).
  21. Yang et al. Remote work and collaboration among information workers (2022).
  22. Barrett et al. Emotional expressions reconsidered (2019).
  23. NIST. Artificial Intelligence Risk Management Framework (2023).
  24. European Commission. AI Act, Article 5.
  25. Durlak et al. The impact of enhancing students’ social and emotional learning (2011).
  26. Cipriano et al. The state of evidence for social and emotional learning: a contemporary meta-analysis (2023).
  27. Mitchell, Sheppard & Cassidy. Autism and the double empathy problem: implications for development and mental health (2021).

Educational and assessment note: This article provides general education, not a diagnosis, treatment plan or character test. Social behaviour is shaped by culture, neurotype, disability, trauma, language, context and power. Use exercises voluntarily, respect boundaries and do not infer a private state from one cue or score.

Continue exploring
Return to the Intelligence Unleashed hub
Back to blog