Your partner hugs you, holds your hand, comes to see you, and says, “I love you.”
Then the brain calls an emergency meeting anyway:
“But what if they do not really mean it?”
The annoying part is that this question has no perfect falsification test.
They say they love you—maybe they are lying.
They text every day—maybe it is obligation.
They hug you—maybe they simply enjoy physical touch.
They stay for decades—maybe they were secretly enduring it for decades.
At that point, the investigation has stopped being an investigation and become an infinite dungeon.
People understandably want rules such as “this kind of hug means true love.” Unfortunately, the human brain has no public API. There is no endpoint where a five-second frontal hug returns love_score: 73.4.
That does not mean we know nothing. We cannot directly observe another person’s inner state, but we can make increasingly good estimates from what they say, how they touch us, what they repeatedly do, whether those patterns stay consistent, and how they repair damage after conflict.
The practical conclusion is simple:
Romance is not mind reading. It is observation, updating, and repair.
1. Start with the hierarchy: words are first-party data, behavior is the log, anxiety is a hypothesis
When trying to understand a partner’s feelings, separate three kinds of information.
- Explicit statements: “I love you,” “I want to see you,” “I dislike this,” “That made me happy.” These are first-party reports about their own state.
- Repeated behavior: showing up, caring, keeping promises, helping when things are hard, considering your interests. These are longitudinal logs.
- Your inference: “Maybe they secretly dislike me,” “Maybe they are only doing this from duty.” These are hypotheses.
Anxiety becomes costly when the third category silently outranks the first two.
A meta-analysis of romantic-partner perceptions found that people do track their partners to some degree, but projection is also substantial. Average tracking accuracy was β=.24, while projection was β=.42.[1]
So “I feel this, therefore my partner probably feels it too” is more powerful than we tend to notice.
A useful operating rule is:
Take what your partner says at face value unless sustained behavior seriously contradicts it. Update confidence with repeated behavior. Keep unsupported fear labeled as a hypothesis, not a fact.
2. What a hug can tell you—and what it cannot
Touch carries information.
In experiments where people communicated emotions only through touch, without seeing one another’s faces, emotions such as love, gratitude, and sympathy were identified above chance. In some touch-communication studies, accuracy for certain emotions ranged roughly from 48% to 83%.[2]
That does not mean “50% is basically a coin toss.” These were not simple love-versus-not-love tasks. Participants chose among many emotions; chance performance could be below 8%.[2]
So touch can communicate a surprising amount.
But the message is closer to:
“This touch seems affectionate or intimate.”
It does not decompose the sender into “love 80, comfort 10, dependency 5, sexual desire 5.”
Behavior can change probabilities. A hug plus kissing or clearly sexual touching raises the probability of a sexual motive. That is ordinary Bayesian updating.
But a higher posterior probability is not the same thing as reading a neural log.
If someone touches a partner’s butt, “sexual meaning became more likely” is a reasonable update.
“Sexual desire is 68.2%” is not.
The brain API is still returning 404.
3. A hug can change the body—but oxytocin is not a love meter
Affectionate touch is not only “feeling nice because I consciously love this person.” Skin stimulation, pressure, warmth, context, trust, and perceived safety can all participate in physiological regulation.
In everyday-life research, moments of affectionate touch were associated with better momentary well-being, lower stress, and measures including salivary oxytocin.[3]
But the internet-friendly sentence “hugging releases the love hormone” is too clean.
Directly measuring oxytocin inside the human brain during ordinary hugging is difficult. Much human research relies on peripheral measures such as saliva or blood. A 2025 review emphasized that direct evidence connecting subtle everyday social interactions to central oxytocin release in humans remains limited.[4]
The safer statement is:
Pleasant social touch interacts with several regulatory systems, including oxytocin-related pathways, but an oxytocin concentration is not a measurement of true love.
No blood test is returning “genuine love: 82/100.”
4. Wanting to cling to a partner does not mean “looking for a substitute mother”
Adults may want to cuddle a partner, relax in their arms, or seek support after a difficult day without wanting anything remotely similar from a parent.
That is not a contradiction.
In adult close relationships, a romantic partner can function as a safe haven—someone with whom vigilance can drop and support can be sought. Feeling that a partner understands, validates, and cares about you is strongly connected to intimacy and partner-specific attachment security.[5][6]
The partner is not a copied parent.
They can become a new safe person inside an adult relationship.
The same caution applies to attachment labels. “Likes hugs = anxious attachment” and “likes being alone = avoidant attachment” are crude shortcuts. Attachment varies across individuals and relationships, and touch preferences depend heavily on context and the identity of the toucher.[6][7]
Average sex differences can appear in some studies, but they are weak tools for diagnosing an individual. Sex differences in the communication of emotion through touch are not uniform either.[2]
The more useful question is not “What do men do?” or “What do women do?” It is:
Who does this person want touch from, in what relationship, and under what conditions does that touch feel safe and welcome?
5. The boundary is not “zero nervousness”—it is whether approach wins over defense
Early physical intimacy is often nervous.
Nervousness is not the same thing as rejection.
“I am nervous to hold hands, but I want to” and “I am scared and want distance” can both involve a racing heart while meaning opposite things.
A study of 1,368 people mapped where different people were allowed to touch the body. The permitted area expanded strongly with emotional closeness; emotional bonds explained about 54% of variation on average. Strangers were largely restricted to hands, while romantic partners had much broader permitted areas.[8]
Humans therefore seem to have a graded touchability map, not a single universal switch.
Stranger: keep distance.
Acquaintance: this zone is fine.
Person I like: maybe the hand.
Person I trust: a hug is welcome.
Intimate partner: I may initiate contact myself.
When moving a boundary, however, “they did not run away” is weak evidence of enthusiasm.
They squeeze your hand back, hug back, move closer themselves, or verbally say yes—those are much more informative signals of reciprocity.
Silence, freezing, or passive nonresistance should not be treated as consent.
The useful distinction is not “Are they perfectly relaxed?” It is “Are they nervous but moving toward me, or nervous and trying to get away?”
6. The anxiety-generated internal board meeting: sometimes you are reading yourself more than your partner
Humans dislike leaving ambiguous spaces blank.
A late reply becomes:
“Maybe they are busy.”
Or, much faster:
“Maybe they are losing interest.”
But that second sentence is not a download from the other person’s brain.
The meta-analysis above found substantial projection.[1] A 2025 daily-diary study of 327 couples followed for 35 days likewise found that partners could track some day-to-day changes in each other’s relationship emotions, but they relied heavily on their own emotions when estimating the other person’s state. Accuracy improved when partners actually verbalized and socially shared their feelings.[9]
So the most powerful mind-reading technology currently available is:
Ask the owner of the mind to speak.
Decades of science eventually rediscovered conversation. Brutal, but useful.
7. Say the love out loud—and when it is said, actually receive it
If you love someone, saying “I love you,” “I missed you,” “I feel safe with you,” or “That made me happy” is useful information transfer.
The receiver has a job too:
Do not automatically invalidate explicit information with an unsupported theory.
“I love you.”
“Maybe that is a lie.”
“I want to see you.”
“Maybe it is duty.”
“I want to stay together.”
“Maybe only for now.”
Every piece of evidence can be given an escape hatch. Even after thirty years, a brain determined to worry can say, “Maybe they were pretending for thirty years.”
At that point, suspicion has become unfalsifiable and therefore unbeatable.
This is not an argument for blind trust. If words and long-term behavior seriously diverge—chronic disrespect, broken promises, ignored boundaries—reassessment is sensible.
But if words and actions broadly align, “they probably mean what they say” is a reasonable best estimate.
Feeling understood, validated, and cared for by a partner is associated with greater partner-specific security.[6]
Occasional reassurance is normal. Repeated, excessive reassurance seeking, however, has been associated with attachment anxiety and relationship processes that can perpetuate distress.[10]
A good relationship can provide a safety net. The safety net does not have to become a bottomless pit.
8. Watch the long-term log, not one dramatic event
Do not judge the entire relationship from one hug, one gift, or one argument.
Watch the accumulation:
- Do they want to see you?
- Do they care about your state?
- Do they help when things are difficult?
- Do they keep promises?
- Do they consider your interests as well as their own?
- Do they sometimes choose the relationship over an immediate selfish benefit?
- Do they apologize after mistakes?
- Do they attempt repair rather than simply abandoning the problem?
Pro-relationship behaviors such as accommodation and sacrifice can act as diagnostic cues that a partner values the relationship, helping trust grow over time.[11]
A person with hundreds of generally reliable actions plus one stupid mistake is not equivalent to a chronically unreliable person who delivers one spectacular bouquet.
Relationships deserve a moving average, not a verdict from the latest candlestick.
And even the average is not everything. What happens after a crash matters too.
9. Do not uninstall the relationship after every bug—repair capacity matters
Partners are humans, so they will mismatch.
They phrase something badly when tired. They forget a message. They misread an expectation. They argue over chores.
If every ordinary conflict becomes proof that “this cannot be my destined person,” eventually being human itself becomes a red flag.
Relationship researchers use the term accommodation for inhibiting an immediately destructive response to a partner’s negative behavior and responding more constructively. Self-control and commitment have been linked to this process.[12]
A two-year study of 904 couples also found that stronger growth beliefs—the belief that relationships can develop through effort and experience—were associated with a slower decline in relationship satisfaction.[13]
That does not mean “every problem can be fixed if you try hard enough.”
It means this cycle can be healthy:
small mismatch → discuss → adjust → improve → accumulate new trust.
By contrast, a tendency to keep maximizing—believing there must always be an even better partner available—has been associated with lower satisfaction, investment, and commitment and with perceiving alternatives as more attractive.[14]
If you reroll the partner gacha forever, the moment you pull an SSR you can still wonder whether a UR exists.
There is no stopping rule.
10. Forgiveness is not the same as leaving the bug unfixed
“Look at the long term” does not mean tolerating everything.
Forgiveness is not universally beneficial. In newlywed research, when a partner frequently behaved negatively, being more forgiving could be associated with steeper declines in satisfaction.[15]
Other longitudinal work found a useful distinction: forgiveness combined with direct requests for constructive change predicted less later offending behavior and more compromise, whereas forgiving without regulation could move in the opposite direction.[16]
A practical ladder is:
Small mistake: discuss and adjust.
Repeated problem: state a clear boundary and a change requirement.
Improvement: let new evidence rebuild trust.
No willingness to change, or repeated serious boundary violations: reassess the relationship itself.
Violence, coercion, chronic humiliation, or severe controlling behavior do not need to be absorbed under the label of love.
Love is not diplomatic immunity.
Repairability and painting a red light green are different operations.
11. This belongs in school: start with “You cannot directly read another person’s mind”
Much of this is less a dating trick than basic human-information literacy.
A useful middle- or high-school curriculum could include:
- You cannot directly read another mind. Separate feelings from facts.
- The person’s own words are first-party information. Do not erase them with an unsupported guess.
- Look at longitudinal behavior. Care, promises, support, investment, and repair matter.
- Say love, happiness, discomfort, and boundaries out loud. Stop waiting for telepathy support.
- Physical touch requires reciprocity and consent. Nonresistance is not the same as active agreement.
- Nervousness and refusal are different. “Nervous but moving closer” is not “afraid and trying to leave.”
- Conflict is not automatic disqualification. Learn how repair works.
- Serious violations are a separate category. Endurance is not automatically a virtue.
A systematic review of 27 studies involving 26,212 young people found relatively consistent improvement in knowledge about healthy romantic relationships, while effects on attitudes and actual behavior were more mixed.[17]
Japan has been moving in this direction. In March 2026, the Ministry of Education revised its “Life Safety Education” materials to explicitly include learning about sexual consent in light of legal and social changes.[18]
For FY2023, 45.3% of public schools reported providing education aimed at preventing sexual crime and sexual violence; among those implementing such education, 32.8% used the Life Safety Education materials.[18]
Progress is real, but the national operating system for “consent, emotion inference, communication, and repair” is not exactly installed everywhere yet.
If a teacher wrote only one sentence on the board, this would be a good start:
“What you feel the other person thinks is not the same as what the other person actually thinks.”
Humanity has spent an impressive amount of time mixing those two fields.
12. Conclusion: not certainty, but trust supported by enough evidence
There is no device that proves true love at 100%.
Not a hug. Not heart rate. Not oxytocin. Not texting frequency.
But that does not make relationships unknowable darkness.
There are words.
There are actions.
There are patterns over time.
There is reciprocal touch.
There is behavior under stress.
There is repair after conflict.
Each signal is imperfect. When dozens or hundreds of independent signals keep pointing in the same direction, however, the estimate becomes strong.
Healthy security is not “all logical doubt has been destroyed forever.”
It is closer to:
“I know I cannot access your mind directly, but I have enough consistent evidence to trust that you care about me.”
Say what you feel.
Receive what the other person says.
Before promoting anxiety to fact, check it.
Repair small mismatches.
Draw lines around serious violations.
And judge the long log, not one cinematic moment.
Romance is not telepathy.
It is a very long Bayesian update run by two humans together.
Sources
- LaBuda, J. E., & Gere, J. — A meta-analytic review of accuracy and bias in romantic partner perceptions. Psychological Bulletin pubmed.ncbi.nlm.nih.gov
- Hertenstein, M. J., et al. / Hertenstein & Keltner — Emotion communication through touch; gender and touch communication pmc.ncbi.nlm.nih.gov
- Affectionate touch and diurnal oxytocin levels: an ecological momentary assessment study pubmed.ncbi.nlm.nih.gov
- Frontiers in Psychiatry (2025) — Review discussing human oxytocin measurement and limits of direct evidence for central release during everyday social interaction frontiersin.org
- Murray, S. L., et al. — Risk regulation in relationships: safety cues, trust, and connection pubmed.ncbi.nlm.nih.gov
- Rice, T. M., et al. — Perceived partner responsiveness and partner-specific attachment security pmc.ncbi.nlm.nih.gov
- Social touch: a systematic review of individual differences and contextual factors pubmed.ncbi.nlm.nih.gov
- Suvilehto, J. T., et al. — Topography of social touching depends on emotional bonds between humans. PNAS pmc.ncbi.nlm.nih.gov
- Kotiuga, J., et al. — Empathic accuracy in couples: A daily diary study of relationship-related emotions. Emotion (2025) pubmed.ncbi.nlm.nih.gov
- Shaver, P. R., et al. — Attachment style, excessive reassurance seeking, relationship processes, and depression pubmed.ncbi.nlm.nih.gov
- Wieselquist, J., et al. — Commitment, pro-relationship behavior, and trust in close relationships pubmed.ncbi.nlm.nih.gov
- Finkel, E. J., & Campbell, W. K. — Self-control and accommodation in close relationships pubmed.ncbi.nlm.nih.gov
- Gander, F., et al. — The role of relationship beliefs in predicting relationship satisfaction over time (904 couples, two years) pubmed.ncbi.nlm.nih.gov
- Relational maximization / maximizing relationship possibilities and relationship outcomes pubmed.ncbi.nlm.nih.gov
- McNulty, J. K. — Forgiveness in marriage: putting the benefits into context pubmed.ncbi.nlm.nih.gov
- Russell, V. M., et al. — “You’re forgiven, but don’t do it again!” Direct partner regulation buffers the costs of forgiveness pubmed.ncbi.nlm.nih.gov
- Hielscher, E., et al. — Healthy romantic relationship interventions for young people: systematic review (27 studies, 26,212 participants) pubmed.ncbi.nlm.nih.gov
- 文部科学省 — 「生命（いのち）の安全教育」教材及び指導の手引き（令和8年3月改訂版）／令和5年度実施率 mext.go.jp
