Attraction & Connection curriculum

Scores from the two members of a couple are nonindependent: they are more alike, or more different, than scores from two unrelated people. Analyzing each partner as a separate case ignores this and can bias the estimates. Dyadic models such as the actor-partner interdependence model treat the overlap as information to study rather than noise to remove.

In the actor-partner interdependence model, each person's predictor may predict both their own outcome and their partner's. In one study of 141 infertile couples in Iran, each spouse's anxiety was linked to their own marital satisfaction, and wives' anxiety was also linked to husbands' satisfaction. The data were cross-sectional, so the direction of influence stayed unknown.

Sentiment override is the tendency for reports of concrete behaviors, such as how often a partner gave a hug, to reflect overall satisfaction more than the specific behavior. One review of 771 relationship studies found self-report in 96 percent, with observation or informant ratings rare. In a study of mothers of preschoolers, some with mental disorders, self-reported relationship quality and coded behavior correlated only weakly.

A daily diary asks for a report once a day; ecological momentary assessment prompts people several times a day about how they feel right now. Looking back invites recall bias, and people can overestimate or underestimate how intense past experiences were. One study of 100 Korean couples used six prompts a day for three days. Both methods still rely on self-report, gathered at many time points.

Data gathered at one time point show that two things go together, not which drives which. Following couples over time can show whether one measure predicts a later one, and researchers can test the reverse path as well. Even two waves of data fall short of proof: a randomized experiment can establish cause, yet many relationship conditions cannot be assigned at random.

Comparing couples by how long they have been together can mislead: lower-quality relationships may end sooner, so long-lasting couples can look happier partly for having survived. Longitudinal studies face the same problem when less satisfied couples drop out more. A couple study also needs both partners to volunteer together, which makes a representative sample of pairs harder to recruit.

Researchers often treat trust, commitment and satisfaction as related but separate constructs. In two studies of about 3,400 US adults in relationships, one general factor explained most of the shared variance in items drawn from dozens of relationship scales, consistent with sentiment override. The authors note they used one-time questionnaires, and distinctions might appear in diary or longitudinal data.

Pooling 148 studies of about 309,000 people, a 2010 meta-analysis found 1.5 times the odds of survival during follow-up for people with stronger social relationships. The authors judged this comparable to quitting smoking and larger than risks such as physical inactivity. The studies followed people over time rather than assigning relationships, and most participants lived in North America or Europe.

Structural measures capture integration in networks, such as marital status, network size or living alone; functional measures capture support received and support believed to be available. These typically correlate only about 0.20 to 0.30 and may affect health differently. In a 2010 mortality meta-analysis, structural, functional and combined measures each predicted survival, with odds ratios of 1.44 to 1.57.

In a 2010 meta-analysis, complex measures of social integration predicted survival more strongly than simpler ones, and a yes-or-no question on living alone predicted least, possibly because someone living alone may still have a large, supportive network. In a 10-year US study of adults aged 57 to 85, having a large network did not predict self-rated health, but lower negative support did.

Under the buffering model, relationships supply informational, emotional or practical resources that soften the health effects of stressors such as illness or life transitions. The main-effects model instead proposes direct benefits, for example by encouraging healthy habits and giving people meaningful roles. These are presented as two general theoretical models of how ties could affect health.

Relationships are linked to health habits and to stress and depression, but those routes do not fully account for their link with health, and reviews point to several intertwined biological pathways. Studies tie social support to better immune function. In 94 couples, a person's ambulatory blood pressure was lower when their spouse had more supportive ties and higher with more aversive or ambivalent ones.

Most studies in the 2010 mortality meta-analysis treated every tie as positive, though research links negative relationships to higher mortality and finds that marriage's effects depend on its quality. A large network could even be a liability for some people through role overload. In a 10-year US study of older adults, negative support from family predicted poorer self-rated health, perhaps partly through mental health.

Researchers cannot randomly assign people to be isolated, married or in a poor relationship, so cause is hard to establish, as with many lifestyle risks. Studies adjust for different sets of confounders, and in the 2010 meta-analysis such controls reduced effect sizes only for structural measures. Most of the data came from North America and Western Europe.

The belongingness hypothesis names two requirements: frequent, pleasant interactions with a few people, and a stable bond in which each cares about the other's welfare. Lacking such attachments is linked to poorer health, adjustment and well-being. Measuring the motive is unsettled: in one study of 372 people, eight existing models of belonging and acceptance failed to fit the data.

Social pain is the distress of damage to one's social connection or value, and researchers proposed that it engages regions of the pain network, the dorsal anterior cingulate and anterior insula. Later data questioned both the size and the specificity of this overlap. In one brain-imaging study of 40 adolescents, these regions responded to being judged by others whether the verdict was acceptance or rejection.

In Cyberball, a participant in an online ball-tossing game stops receiving the ball. A meta-analysis of 120 such studies with about 11,900 people found a large average effect that held across group size, duration, gender, age and country. Afterward, people may respond prosocially or antisocially, and replications show that some seek solitude, which may deepen isolation.

Schachter noted as early as 1959 that uncertainty favors contact with similar others, and is described as showing that such contact reduces uncertainty and related anxiety. Seeking contact is also common in emergencies. In a French lockdown survey, people reporting more threat-related emotions also reported more affiliative activity, in women as well as men; only among men did more affiliation go with lower reported compliance.

The social brain hypothesis holds that network size is limited partly by cognition and partly by the time relationships take, with layers near 5, 15, 50 and 150. Two UK surveys found Facebook networks similar in size to offline ones. In one diffusion model, gains in access to information diminished beyond a community of about 50, a modeled rather than observed result.

Granovetter argued that close friends' networks overlap, so ties reaching into other circles are better placed to bring news; one Twitter study found new information traveled preferentially along links between groups. In US and Mexican communication networks, stronger bridging ties, rather than weak ties as such, went with greater economic prosperity, a cross-sectional link that may run both ways.

In high-mobility settings people have more chances to pick new partners and drop old ones; in low-mobility settings they are more firmly bound by obligation. It is often measured by asking how easily people around you make and change relationships. Where it is higher, people report more intimacy and self-disclosure with a close friend, a correlation replicated in Brazil.

That proximity predicts who interacts and forms ties is described as one of sociology's most robust findings. In a study of 235 children aged 8 to 11, classmates seated near each other were more likely to be friends, and seat proximity predicted new friendships over about 13 weeks. Seats were set by teachers rather than at random, and the effects were relatively small.

The mere exposure effect is a rise in liking for something, such as a shape, word, odor or face, just because it has been met repeatedly. In visual studies it did not require recognizing the item as seen before, and subliminal presentation produced larger effects. It works best for things that start out neutral, and familiarity from repeated contact is one proposed reason proximity breeds liking.

Proximity can be measured in several ways, such as sitting side by side, sharing a table, or the number of seats between people. In two companies whose headquarters moved to open-plan offices, wearable sensors showed face-to-face interaction fell by about 70 percent while email and messaging rose. Open layouts appeared to prompt withdrawal from officemates rather than more contact.

In one odor study, repeated exposure made neutral and mildly pleasant odors more pleasant but left unpleasant and very pleasant ones unchanged. An earlier word study suggested that disliked words grew more disliked, yet other evidence suggests exposure can ease dislike of angry faces and caged snakes. In two experiments, exposure increased approach to threatening stimuli shown alone but not when mixed with neutral ones.

Twelve-month-old infants and 3-month-old macaques looked longer at faces closer to their species' average. Neither showed this preference for an unfamiliar category of faces, suggesting it depends on experience. Two accounts, not mutually exclusive, treat averageness either as a cue to mate quality or as a by-product of the easier, more fluent processing of typical faces.

In one lab study, faces were rated more positively when shown with a pleasant odor and more negatively with an unpleasant one, compared with no odor. In another, a noxious smell that induced disgust led people to report less warmth and liking toward gay men, though not toward heterosexual men. A 2026 review of 21 such studies calls odors contextual cues that bias social judgment, while urging tests in more realistic settings.

The beauty-is-good stereotype credits attractive people with more desirable personality traits. A meta-analysis by Eagly and colleagues found strong effects on ratings of social competence and intelligence, weak ones for moral integrity and none for concern for others. In a study of 100 student faces, judgments of competence tracked actual academic performance more closely once attractiveness was statistically controlled.

In one sample, men's rated attractiveness rose with symmetry and averageness, while women's rose with femininity and fell with body fat. When raters in ten countries judged Czech faces, averageness and femininity predicted attractiveness but asymmetry had no effect, whereas a Spanish study linked facial asymmetry to women's rated attractiveness. In the first sample, links from these traits to measured health were weak.

When raters in ten countries judged the same Czech faces, consensus was high, yet raters from countries lower in human development agreed less with Czech raters. Agreement is generally greater within a culture than between cultures. Ideals also change: the waist-to-hip ratio of women depicted as beautiful in Western art has fallen since the 15th century.

If more attractive people are just more often accepted as partners, simulations produce couples matched on attractiveness even though nobody looks for similarity. Network models show the match is looser when people have fewer potential partners, because less attractive people who do pair up often land a more attractive partner. These are models of a mechanism, not observations of real couples.

Pooling three speed-dating studies with follow-ups, researchers found that a date's consensual desirability, the appeal most people agree on, predicted later contact and romantic interest, along with a person's unique liking for that date. Suitors also tend to message attractive online daters. These results cover the days and weeks after a first meeting, not how long relationships last.

Some studies, mostly using speed dating, report that stated preferences have little if any power to predict actual choices, and some argue this reflects the method. Across three couple samples, partners sat close to people's preferences overall, but newlyweds' partners were closer to their preferences than only about 61 percent of alternatives. In 116 French couples, men's preferred and actual partner body size differed, but women's did not.

Mock-juror studies linked attractive defendants to fewer convictions and lighter sentences, an attraction-leniency effect later seen in the sentences real judges imposed, and recruiters are reported to rate attractive applicants more favorably. A 2023 study using video clips and many different defendants found no clear effect on judged guilt; any effects were small, uncertain and varied by type of crime.

People who learn that a stranger shares their attitudes feel more attracted to that person, and people who are very similar are described as having smoother, more enjoyable interactions. Across four studies with 428 people, attitude similarity predicted attraction, but whether that attraction became willingness to befriend the person depended on how strongly the shared attitude was held.

The complementary hypothesis says people are more satisfied with partners who complement them, yet most findings favor similarity. Dominance is a named exception: pairing a dominant with a submissive partner is thought to avert constant conflict. And in one study of Chinese dating couples, partners with more dissimilar personality profiles reported better relationship quality.

Similarity-attraction effects are described as driven mainly by perceived rather than actual similarity. The influence is also described as running both ways: perceived similarity is both a cause and an effect of liking, since seeing a liked partner as similar is cognitively consistent. In political studies, people also liked confident, coherent targets beyond what perceived similarity explained.

In a story task, Chinese 5- and 6-year-olds picking friends among unfamiliar peers favored those who liked them as well as those they liked. Among adults who knew each other well, people assumed others returned their attraction; actual reciprocity existed but was much weaker than assumed. Mutual interest is described as likely necessary for a relationship to form.

Speed-dating research separates a person's general desirousness from their unique desire for one partner. Pooling three studies with follow-ups, people showed somewhat more later interest in a date when that date had uniquely desired them, and less when the date was generally desirous toward everyone. The authors read this as people noticing and valuing selectivity, though these effects were modest.

Models built on a couple's current interaction have predicted break-up well, but only for existing relationships. For traits measured in advance, information about the pair adds little: in 192 German partners followed for four years, partner traits and similarity added nothing beyond each person's own traits, and a machine-learning study of 208 singles found individual traits predicted romantic interest poorly.

The social relations model splits attraction into a perceiver's general tendency, a target's general appeal, and a relationship effect: one person's unique response to a specific partner. After speed dating, appeal and unique liking both predicted later dating, appeal somewhat more. In one machine-learning study, judgments about a specific partner, such as finding them exciting, predicted romantic interest better than individual traits.

Breadth is the amount and variety of what is shared; depth is how close a topic sits to the innermost self. The theory expects breadth and depth to rise together, but offline studies found that depth, rather than breadth, explained differences in outcomes such as liking. Disclosure is often described as deepening step by step as a relationship grows closer.

People who receive a disclosure tend to respond by revealing a comparable amount about themselves. In one online forum, the length of women's replies closely tracked the length of the posts they answered. Disclosure is described as mainly a two-person process driven by reciprocity. Word counts capture how much is said, not how intimate it is, so on their own they cannot show whether depth is matched.

One view holds that very personal information shared on first meeting can seem inappropriate. In experiments where a confederate mentioned a same-sex dating partner to heterosexual participants, disclosing early rather than later led to more liking and closer approach. The authors set this against earlier studies that found benefits of delayed disclosure, so timing effects are not uniform.

Luft and Ingham's model has four panes: open (known to self and others), blind (known to others only), hidden (known to self only) and unknown (known to neither). Receiving feedback openly is described as one way to shrink the blind pane. One qualitative study of nursing and physiotherapy students used the model to interpret how peer feedback in team learning may have widened their open area.

A meta-analysis found the link runs both ways: people reveal more to those they already like, and they like others more after having disclosed to them. People are also said to like those who reveal more about themselves. A sense of connection may grow even without an exchange: earlier studies suggest people feel more connected to a constructed profile after merely reading its Facebook posts.

Under this theory, people see information about themselves as theirs to control, held inside a privacy boundary. Telling someone makes that person a co-owner who also controls its further spread, so owners set rules for co-owners to follow. When a co-owner breaks them, as when relatives post a child's photo, privacy turbulence follows and the boundaries must be renegotiated.

Hyperpersonal theory argues that anonymity, invisibility and time to compose give people more control over how they come across, and these features may also encourage deeper disclosure. In mock Facebook studies, more disclosure on a widely seen timeline went with lower romantic intimacy and satisfaction, while offline disclosure went with higher. Other work reports that posts made in a highly emotional state are often regretted.

In the Fast Friends Procedure, two strangers take turns answering a series of questions that invite steadily more personal disclosure. In recent studies, self-rated closeness rose from before to after one session, including a 15-minute text version with pre-written partner answers, where a small-talk version raised it too. These studies measure closeness right after one session, not whether a lasting relationship follows.

In the Fast Friends Procedure, the question sets begin with superficial small talk and move toward personal values, attitudes and beliefs. In a 2026 study of 124 people in same-gender stranger pairs, closeness rose from each stage to the next except between the first two intimate sets. The authors flag a confound: because the order was fixed, deeper disclosure was mixed up with time effects such as habituation or fatigue.

In the Fast Friends Procedure, partners take turns answering the same questions, so disclosure runs both ways. In a 2026 text study using pre-written partner answers, participants who saw more disclosing answers felt closer and disclosed more themselves. Their own disclosure was not significantly tied to how close they felt, and all of these links were correlations.

In studies of short conversations between strangers, people rate how much their partner liked them and enjoyed their company lower than the partner actually does. A 2021 review files this under meta-accuracy, how well people know what others think of them. Most studies use brief encounters; a 2025 study found preschoolers assumed their best-liked classmates liked them less in return.

Behavioral confirmation is the process by which people come to act in ways that confirm someone else's expectations of them. A 2015 review summarizes evidence that people who expect acceptance behave warmly and are accepted more, while those expecting rejection act coldly and are accepted less. In a classic intercom study, a woman whom a man believed attractive behaved in a more appealing way; warmth is the presumed link.

In one 2026 study, stranger pairs randomly given deep prompts, such as times they had cried in front of others, felt more connected than pairs given shallow topics like the weather. In another 2026 study, a deep-talk version of a getting-acquainted task produced no more closeness overall than a small-talk version. Both measured connection by self-report right after one conversation.

Bowlby used the idea of a behavioral system, a biologically evolved program, to explain why infants seek out a stronger, protective figure, especially in danger. Under threat the system activates behaviors such as crying and clinging; once the child feels secure it quiets, freeing energy for exploration. Later theorists ask whether attachment also differs by relationship and context.

As a safe haven, an attachment figure offers comfort and reduces stress, helping a person regain a sense of security. As a secure base, the same figure offers support from which to explore and take on the world. The two can be split across people: in a study of 198 committed couples, partners were relied on most for every function except secure base, where mothers were relied on about as much.

In Ainsworth's Strange Situation, a child goes through eight three-minute episodes, alone or with a caregiver or a stranger, and the child's behavior is sorted into patterns such as secure, avoidant, ambivalent or disorganized. Reunions are telling: an avoidant child may turn or lean away from the returning parent, while an ambivalent child fails to settle and usually keeps crying.

Ainsworth and colleagues identified three patterns, secure, insecure-avoidant and insecure-ambivalent, and a disorganized category was added later. In one account, avoidance reflects a deactivated attachment system and ambivalence a hyperactivated one, both tied to the adequacy of care, while disorganization is tied to a frightening caregiver. Researchers also argue a person can hold distinct attachment models for different figures.

Ainsworth developed maternal sensitivity to explain which interactions lead to secure attachment, defining it as accurate reading of an infant's signals plus prompt, appropriate responses. Studies consistently find a link with security, though a meta-analysis of more than 65 studies called it important but modest. A 2018 critique argues later versions, such as an emphasis on play, may be culturally specific.

Internal working models represent the attachment figure, the self and the relationship between them. They work like predictive maps that help a child anticipate a caregiver's behavior, and mostly positive experience is thought to make secure people more optimistic about social interactions. They are not one fixed script: alternative models of one figure may coexist, and people may hold different models for different relationships.

A meta-analysis by van IJzendoorn and Sagi, comparing samples from Africa, China, Israel, Japan and Indonesia with Western ones, found similar patterns in every study and a secure majority of infants. In a 2022 study of adults, secure was the most frequent style overall, but the most common insecure style differed between Japan and Spain or Italy. Critics argue sensitivity measures carry culturally specific assumptions.

Bartholomew proposed two dimensions, a person's image of self and image of others, and Brennan and colleagues linked the self image to attachment anxiety and the other image to avoidance. High anxiety goes with overvaluing relationships and fearing separation; high avoidance with downplaying them and fearing commitment. Researchers still disagree whether adult attachment is better treated as types or dimensions.

Secure means low anxiety and low avoidance; preoccupied, high anxiety and low avoidance; dismissing, low anxiety and high avoidance; fearful, high on both. Treating these as four separate boxes is contested: in a 2021 study of 515 Spanish adults using one attachment questionnaire, only two profiles, insecure and normal range, could be clearly told apart.

When attachment feels threatened, people higher in anxiety tend to push for closeness and support, closely monitor whether others are available, and attend to cues of rejection. Research summaries link this vigilance to jealousy, rumination and an ironic undermining of a partner's commitment. These are correlates of less satisfying relationships, not proof that the strategy causes the outcome.

When attachment needs are triggered, people higher in avoidance tend to suppress dependency, dampen emotion and keep distance, which can include compulsive self-reliance. It often shows up as reluctance to self-disclose and downplaying relationship problems. Suppression may carry a cost: studies of thought suppression find that pushing thoughts away tends to make them more accessible and raise arousal.

Emotionally focused couple therapy describes a demand-withdraw cycle: one partner escalates to meet attachment needs, the other pulls back, and the cycle becomes entrenched. Findings are mixed, and research often using undergraduates suggests both anxiety and avoidance may relate to several conflict behaviors. In one study of expectant couples, avoidant men reported more emotional flooding when their partner was high in anxiety.

In a study that followed 574 emerging adults and 463 parents over a year, their attachment to each other was very stable, with earlier measurements explaining more than 60% of the reliable variance, though situational effects built up over the year in the young adults, not their parents. Life and relationship contexts can alter adult attachment, yet few life events have been linked to change, and those effects were small and temporary.

The term earned secure is used for people who experienced poor parenting or adversity but attain secure attachment later in life. Factors discussed include an alternative caregiving figure, a supportive adult relationship and psychotherapy. The evidence is mostly correlational, and a 2019 clinical review warns that retrospective studies cannot settle it and that current mood symptoms may bias memory.

Companionate love is felt less intensely, often follows a period of romantic love, and merges intimacy with commitment. One proposed timeline has passion holding steady for months to years while intimacy and commitment grow, then passion subsiding as they peak. That timeline rests on a single study of mostly university students, and in some people romantic love lasts for years or decades.

Intimacy covers warmth, communication and connectedness; passion covers excitement, desire and physical arousal; commitment is the more deliberate decision to maintain the relationship. A 15-item short form of Sternberg's scale was validated across 37 language versions, supporting the three-part structure. It is one framework among several, alongside love styles and the passionate-companionate split.

Lee proposed three primary styles, eros (passionate), ludus (game-playing) and storge (friendship-based), which combine into pragma (practical), mania (possessive) and agape (altruistic). Hendrick and Hendrick's Love Attitudes Scale scores everyone on all six instead of sorting people into one type. In one survey of 1,101 adults, married or cohabiting people scored higher on eros and agape than singles, a correlation only.

The model starts from a proposed basic motive to expand the self, and close relationships are a main route: each partner comes to experience some of the other's resources, perspectives and identities as partly their own. This inclusion of the other in the self is often measured by choosing among pairs of overlapping circles. In the model's account, inclusion happens fast at first, and the excitement slows as partners get used to it.

In an early experiment, couples assigned activities both partners rated exciting and pleasant, at least 90 minutes a week for 10 weeks, reported higher satisfaction than couples doing pleasant but unexciting ones or a control group. Later tests are mixed: in a 2025 virtual reality study with couples, a novel shared activity brought less boredom and more closeness than a mundane one, but not more satisfaction or self-expansion.

Limerence involves intrusive preoccupation with one person, a longing for the feeling to be returned, and often idealization of that person. Accounts describe it as driven by uncertainty and possible rejection and as easing once certainty arrives. Some researchers see falling in love as related to, yet distinct from, limerence, while a 2021 review notes that seminal work called romantic love itself limerence.

A 2021 review calls romantic love universal or near-universal, yet arranged marriages are the norm in 80% of 200 forager societies in one ethnographic database. Across five non-Western societies, love scores did not differ significantly overall between arranged and free-choice marriages, with differences both ways inside some. In a survey spanning 90 countries, people on average were unwilling to commit long term without love.

In female prairie voles, oxytocin infused into the brain can induce a partner preference and an oxytocin blocker stops it forming; blocking vasopressin V1a receptors does the same in males. Yet in one recent study, voles lacking working oxytocin receptors still pair-bonded. Human work is harder to read: blood levels track brain levels weakly, assays have known problems, and candidate-gene links have replication concerns.

Voles were useful because monogamous and non-monogamous species exist in the same genus, so differences in oxytocin and vasopressin receptor maps could be linked to pair bonding. Yet prairie voles themselves vary in mating tactics and receptor patterns, and a 2021 study of lemur brains found little evidence of a comparable pair-bonding circuit. Unique features of human neurobiology add further doubt.

Brain imaging shows intense romantic love engaging dopamine-rich reward regions, including the ventral tegmental area, that also respond in drug and behavioral addiction. One group calls early love a natural, often positive addiction, citing craving, withdrawal and relapse. Researchers have not classified it as an addiction, since addiction is a harmful disorder in a subset of people while love is common and often positive.

In a lab study, women who briefly embraced their romantic partner before a stress test showed a smaller cortisol response than women who did not; men showed no such effect, and blood pressure and mood did not differ. In a 14-day diary study of 404 adults, receiving a hug on conflict days went with smaller mood changes, a correlational result that did not differ by relationship status.

In a survey across 37 countries and a second study, people who reported more love for their partner also reported more affectionate touch such as hugging and kissing, an association rather than a measured effect. In the UK and Japan, the stronger the emotional bond, the larger the body area people allowed someone to touch, though UK participants found social touch more pleasant.

Critics question whether oxytocin sprayed into the nose reaches the brain in amounts that matter, and some early attempts to replicate its effects failed. One 2020 imaging study found a drop in amygdala blood flow was fully explained by oxytocin entering the bloodstream, yet also found evidence the nasal route can target specific brain regions. Detecting a small two-group effect at 80% power needs 394 people per group.

A 2025 review notes that passionate love is often presented as innate and biological, yet how often people experience it may vary with cultural context, social structures and beliefs. Even a single marker is unsettled: studies of cortisol in new love point in different directions, which the review traces partly to differences in methods and in what is measured.

In exchange and interdependence accounts, people weigh a relationship's rewards against its costs. Outcomes at or above the comparison level, what a person expects to receive, bring satisfaction; staying depends on the comparison level for alternatives, what they believe they could get elsewhere, including being single. So a dissatisfied person may still stay, try to change things, or lower their expectations instead of leaving.

Commitment is predicted to rise with satisfaction, with poorer perceived alternatives and with investments, the time, shared plans and other resources tied to a relationship. In two meta-analyses, satisfaction had the largest association with commitment, while investment and alternatives had smaller, roughly comparable ones. Most tests involve relationships with at least some face-to-face contact.

Committed people tend to rate attractive alternatives as less attractive and attend to them less. In one study of 109 heterosexual Chinese college men, those in exclusive relationships largely ignored attractive women in an attention task, while single men primed with love attended to them more. Evidence centers on heterosexual people, and in one set of studies lower attraction ratings appeared mainly under a gain-seeking, or promotion, focus.

Responses to a partner's hurtful act are sorted on two axes, active or passive and constructive or destructive: discussing the problem, quietly waiting it out, criticizing or threatening to leave, or ignoring the partner. Early work found both constructive kinds protected satisfaction, but a later study linked only the active kind to closeness and satisfaction, and passive constructive responses were noticed less.

Exit means actively leaving, such as breaking up; voice means openly discussing the problem to resolve it; loyalty means staying and expecting things to improve; neglect means pulling away and going cold. Voice and loyalty count as constructive and exit and neglect as destructive, while exit and voice are active and loyalty and neglect passive. The same scheme is also used to study job dissatisfaction.

A positive illusion is a tendency to see one's own relationship as better than other people's, or one's partner more favorably than the partner sees themself. Higher levels have been associated with more satisfaction, less conflict and lower breakup risk. Whether the bias helps is debated, and researchers argue bias and accuracy can coexist, so a person can idealize a partner and still be accurate about them.

Clark and Mills distinguished exchange relationships, where a benefit creates a debt to repay promptly in kind, from communal ones, where people give to meet the other's needs without expecting return. Most people follow communal norms with close others, though some apply exchange rules even there. In a study of new parents, a drop in women's perceived equity of support, either way, went with more depressive symptoms in both partners.

In Reis and Shaver's model, intimacy builds through two moves: one person discloses something about the self, and the partner responds. Whether the exchange feels intimate depends on whether the discloser feels understood, validated and cared for. One multi-study test of the model in couples was correlational, so it could not rule out effects running the other way.

Researchers measure responsiveness by asking how far a person feels their partner understands, validates and cares for them; one scale uses 18 such statements. Support research treats the same three feelings as a necessary ingredient of effective help. Because each statement records the receiver's view, the scale captures how responsive a partner seems, not what the partner actually did.

Support research separates perceived support, the belief that help is there, from received support, the help actually given. Perceived support is linked to benefits, while received support shows mixed and sometimes negative links with well-being. In a 148-study meta-analysis of mortality, perceived support was tied to survival, but received support, measured in just 9 of the pooled estimates, could not be told apart from no effect.

A multi-study paper on romantic couples found that people who saw their partner as more responsive were more likely to touch that partner affectionately, in a lab session and in daily life. In one diary study, touch on one day also predicted the partner seeing the toucher as responsive the next day. All of the evidence was correlational, so the direction is not settled.

A bid for connection is any attempt by one partner to engage the other, such as a text, a kiss or a greeting at the door, and the other can turn toward it or away. Gottman's method lists bids among the predictors of relationship success it identifies, and turning toward each other is one of the seven parts of its Sound Relationship House. Trials test whole programs, so they cannot isolate what answering bids does.

Couple researchers define validation as accepting a partner's point of view, which brings respect into the exchange. Validating couples are described as keeping eye contact, paraphrasing and asking questions even in disagreement. In a review of couples living with chronic pain, empathic and validating partner responses went with better pain and relationship outcomes in most studies and invalidating ones with worse.

Research on unhelpful support lists quiet failures: not really listening, quickly bringing up one's own problems, minimizing how serious the problem is, and advice offered before any acknowledgment, which can make it seem premature. Partner phone use is another candidate; in a survey of 772 Chinese young adults, lower perceived responsiveness helped explain its link with poorer relationship quality for women but not men.

Emotional support shows love, care and empathy; informational support offers facts or advice; instrumental support is practical help through materials or tasks; appraisal support affirms that what someone did or said was appropriate. In a two-study paper on oncology patients rating their providers, the four formed related factors, and in the second study informational and emotional support each uniquely predicted trust in the doctor.

The model pairs controllable problems with informational support that helps solve them, and uncontrollable ones with emotional support that helps a person feel cared for. In one test of the matching idea, three workplace vignette experiments found that support moderated the effect of change demands in only one, and the authors argued that a match also depends on what the receiver finds personally relevant.

Received support sometimes goes with worse well-being, perhaps because explicit help draws attention to the recipient's difficulties. Bolger and colleagues proposed invisible support: help given outside the recipient's awareness, such as quietly doing extra housework, or not coded as support. Studies suggest benefits, yet in one study of couples with a smoker trying to quit it went with less distress but more cigarettes.

In a week-long diary study of couples coping with rheumatoid arthritis, patients' pain tended to rise across a day after more negative support and more partner-reported solicitous help, while esteem support went with pain easing; pain may also draw solicitous help. In experiments on help between groups, help that handed over a full solution, rather than empowering recipients, made helpers seem less warm and competent.

In a diary study of 55 married couples, explicitly telling a partner about daily stress predicted both partners' responsiveness in coping with it together, and that responsiveness in turn linked to satisfaction; explicit communication had no direct link to satisfaction. Asking has another side: in one study, Americans who tried to give more requested support reported more negative emotion.

A review of 27 studies of military spouses whose partners had mental health problems found caregiver burden and strain on the intimate relationship as recurring themes. In a national sample of 549 adults in Switzerland, emotional support was more often exchanged both ways in widely dispersed families, which the authors read as one way families sustain long-distance ties. Neither study tests whether reciprocity prevents burden.

In a survey of 887 adults in Spain aged 65 and over, support from a spouse, grandchildren, siblings and friends went with lower social and emotional loneliness, but support from neighbors went with more emotional loneliness and support from children with more social loneliness. In a study of 31 autistic adults, other autistic people gave distinct support yet were rarely counted as close ties.

Many studies tie adults' ability to resolve and recover from conflict to relationship satisfaction and stability. But a three-study longitudinal analysis found negative communication mainly rose and fell with satisfaction at the same time rather than predicting later change, and positive communication rarely tracked it. In 80 adolescent couples, observed conflict resolution and recovery did not predict breakups.

One partner pushes through requests, criticism or complaints while the other withdraws by changing the subject, falling silent or leaving. The pattern appears more often in dissatisfied couples. In a questionnaire study of 175 Spanish couples, one partner's withdrawal paired with the other's demands or aggression went with lower satisfaction in both, and avoidant attachment was tied more to withdrawing than to demanding or aggression.

Gottman and colleagues coded couple conversations in 10-second segments and found that unhappy couples answered negative affect with negative affect more often than happy couples did. Looking back over observational studies since the 1970s, a 2016 paper judged that greater negativity and longer chains of negative reciprocity signal distress reliably but weakly, and that no such feature foretells how a marriage turns out.

A harsh startup is entering a discussion by blaming or criticizing the partner, and Gottman's theory names it, alongside flooding and failed repairs, among the patterns it used to predict divorce. One Polish couple study measured it with five self-report statements adapted from a work by Gottman and Silver, such as whether arguments seem to come out of nowhere, so the score reflects impressions rather than coded openings.

Gottman's framework lists solvable problems and perpetual issues among the factors linked to whether marriages last. A deep rift such as one partner changing religion is the kind of conflict it might frame as a perpetual problem, while integrative behavioral couple therapy calls such issues differences. Some family therapists argue both treat the conflict as the couple's alone, when it also involves children, relatives and communities.

Couples with a constructive communication style are described as repairing an interaction when it turns negative, and Gottman's framework lists failed repair attempts among factors linked to dissolution. Recovering positive feeling after a conflict talk, a related idea, predicted lower divorce risk over four years in Gottman and Levenson's work, yet observed recovery did not predict breakups among 80 adolescent couples.

Gottman described flooding as emotional overwhelm in response to a partner's negative behavior, and later work treats it as a state of overload that disrupts thinking and blocks problem-solving. In 96 expectant couples, men's attachment anxiety was tied to observed withdrawal and women's to observed hostility, each through their own flooding. That study did not test whether taking a break helps.

Couples discuss a disagreement on video, and trained coders label what happens; Gottman's team classified 10-second stretches as positive, negative or neutral. One adapted version of the Specific Affect Coding System separates codes such as criticism, defensiveness, stonewalling and contempt. Reliable coding takes training: one coder reached high agreement after 12 hours of training, 4 of supervision and 60 of practice tapes.

Gottman named four conflict behaviors the four horsemen of the apocalypse and used his communication patterns to predict divorce. One account holds that defensiveness often follows perceived criticism and contempt and can set off a cascade toward stonewalling. An adapted coding system scores each behavior separately, but the strong prediction claims come from studies whose accuracy critics dispute.

Gottman suggested that a five-to-one ratio of positive to negative feeling during conflict marks constructive couples, and studies of affect ratios found partners more likely to separate below it. Some papers repeat it as a minimum couples must meet. Other ratios, such as three to one, also circulate, and after a critique of Fredrickson and Losada's ratio, how much negativity outweighs positivity is debated.

Some papers still say Gottman's team can predict divorce from the first three minutes of a discussion. Critics call such accuracy rates often misleading, since models were mostly fit to the same data rather than tested on new couples, and cite thin validation and scarce data. In a 2023 study of 101 couples with 17 breakups, emotions in a conflict talk did not predict separation two years later, though patterns in a positive talk did.

The Sound Relationship House has seven parts: building love maps, sharing fondness and admiration, turning toward each other, keeping a positive perspective, managing conflict, making life dreams come true and creating shared meaning. Gottman built it to show how relationships work or fail and uses it in psychoeducation and therapy. Two trials delivered such programs whole, over eight or ten sessions, so they cannot show which parts work.

An eight-session online program for 72 Iranian couples improved self-reported communication against a control group sent parenting e-books, with gains still reported ten months after it ended. A 16-couple study reported better adjustment and intimacy, and a 60-couple study of infertile couples found higher intimacy in women but not men. The online trial's authors note a homogeneous sample, self-report measures and possible confounds.

Maintenance behaviors are the things couples do to keep a relationship from declining. One early set has five: positivity, openness, assurances, shared social networks and sharing tasks, studied as markers of stability. Others group relationship behaviors differently, for example as emotional closeness, social companionship and positivity, and one team argues that no current measure of romantic behavior is comprehensive.

Couple researchers describe rituals as symbolic actions that make up a couple's bonding, and partners maintain their shared order as a couple through symbols such as joint anniversaries and gifts. In interviews with 17 long-married German spouses facing illness and care needs, earlier rituals were adapted rather than dropped, and earlier work stresses keeping daily routines such as shared meals. The evidence is qualitative and small.

Leslie Baxter's theory names tensions such as integration versus separation, stability versus change and expression versus nonexpression, and treats them as natural and unavoidable in any relationship. Related work adds autonomy versus control and openness versus closedness. Recent uses extend to caregiving, such as parents of teens and young adults with blood cancer who describe constantly navigating competing pulls.

The coming-together stages are initiating, experimenting, intensifying, integrating and bonding; the coming-apart stages are differentiating, circumscribing, stagnating, avoiding and terminating. On the way down, partners first put their own interests ahead of the relationship, then limit talk to avoid arguments. It is one of several stage models of how relationships build and dissolve.

A turning point is any event associated with a change in a relationship, such as a first big fight or a marriage proposal. Turning point analysis, an established method in relationship science, asks people to retell a relationship's history and mark these moments. Accounts come from each partner's own point of view; in one study of therapist and patient pairs, the two sides named partly different kinds.

Long-distance partners can face limited face-to-face contact, shifting roles, and less sexual intimacy and social support, while often working to keep closeness and commitment. The separation-reunion cycle can lift mood briefly as partners avoid conflict to protect scarce time together. After a military deployment, reunion can be stressful because partners must readjust their expectations and routines.

Stage models of dissolution describe a slide from private doubts through emotional disengagement to separation rather than a single blow. In a British national survey, people whose live-in partnership had ended in the past five years most often cited growing apart, ahead of arguments and infidelity, for marriages and cohabitations alike. Reasons were self-reported, and more than one could be given.

People share positive events with others more often than negative ones, so couples get frequent chances to capitalize. Work by Gable and colleagues found that responses to good news predicted relationship well-being better than responses to bad news. That research has drawn mostly on Western samples, a gap one Malaysian survey set out to address.

Active-constructive replies are enthusiastic, passive-constructive ones quietly pleased, active-destructive ones find a problem, and passive-destructive ones pay no attention. Among 139 Malaysians in dating relationships, perceived active-constructive replies had the strongest tie to satisfaction. Quiet support is less settled: that survey found it helpful for good news, while in 151 cohabiting couples it went with lower sexual satisfaction.

Studies tie expressed gratitude to bonding, intimacy, commitment and relationship quality, though most evidence is correlational. In one study, a partner's felt gratitude predicted the other's satisfaction while expressed gratitude did not. In a five-week field experiment, couples randomly assigned to plan gratitude expressions spent more time together in daily life, an effect carried by their rise in expressed gratitude.

Find-remind-and-bind theory casts gratitude as a detection-and-response system: noticing a thoughtful act can flag a potential high-quality partner, remind people of an existing partner's worth, and keep the person thanked engaged in the relationship. It is a proposed function that researchers use to interpret findings, and evidence that gratitude builds bonds is mostly correlational.

Approach relationship goals have been linked to more satisfying, less conflictual sacrifices, while avoidance goals in general go with falling satisfaction. A review of 112 studies of young people linked sacrifice aimed at benefiting the relationship with well-being. Researchers studying couples coping with low sexual desire propose that sacrifice driven by fear of conflict or loss may end up tied to the very outcomes feared.

One model defines forgiveness as a change in motivation toward someone who caused harm: less urge to retaliate, less avoidance, more goodwill. Most researchers distinguish it from condoning, which treats the offense as justified, and from reconciliation, which restores the relationship. Its desired endpoint can range from reconciliation to detachment, and definitions still vary.

Gottman's Sound Relationship House ends with creating shared meaning, described in one counseling program as building a common culture, including its spiritual side. In 148 Hungarian couples followed for a year, coordinating personal projects predicted better progress on them, which went with higher life satisfaction. The direct link from coordination to satisfaction was not significant.

Friendship is a chosen tie, and network studies find it weakens unless kept up through contact and shared activity, while kin ties are steadier. In a study that followed 30 students in England for 18 months as they left school, friendships lost closeness while family ties proved more stable, and feeling close to old friends did not stop the slide. A 2025 review reports closeness can drop within months without meeting.

One account of social networks describes nested circles of about 5, 15, 50 and 150 people, with time spent per person falling at each outer layer. It sets minimum contact rates for a tie to stay stable: about weekly for the closest 5, monthly for 15 and yearly for 150. A 2021 reanalysis of the primate data behind Dunbar's number of 150 found the estimates too uncertain to fix any single figure for human group size.

Keeping a friendship close takes upkeep, mainly communicating and doing things together. When 30 students in England were tracked through the move from school to university or work, friendships that got more effort declined less, but the route differed: more talking protected the women's friendships, more shared activities protected the men's. The authors call their sample modest, so the split needs confirming.

In a year of calls from one European phone operator, the number of people a user contacted each month peaked around age 25, fell until about 45, held steady for a decade, then fell again after 55. Socioemotional selectivity theory expects friends to matter more with age, yet a 20-year Swiss panel of adults aged 55 to 90 found friends losing ground to neighbors, even for emotional support.

Research on young adulthood lists moving, changing jobs, starting or finishing school, marrying and having children as transitions that can cost friendships, and longitudinal work suggests fewer chances to meet in person is the main reason ties lapse. In one study of students leaving school, distance itself did not drive the decline in friendships; time spent making new friends appeared to come at the expense of old ones.

When adults in one university course rated one another, those who named someone a friend expected the feeling to be returned in 94 percent of cases. Across surveys from several experiments, only about half of such ties were mutual. In the same team's fitness experiment, peer pressure changed behavior when the tie was mutual or ran from the persuader to the target, but not when only the target counted the persuader a friend.

In a series of studies, most people reported losing touch with an old friend yet showed little interest in reconnecting, and fewer than a third sent a message even when they wanted to, had contact details and were given time to write. Participants were no more willing to message an old friend than to talk to a stranger. A warm-up of practice messages to current friends and acquaintances raised the number who reached out by about two-thirds.

Family systems thinking, brought into therapy by pioneers such as Bateson, Bowen and Minuchin, sees members as emotionally linked parts of one unit. A change in one person, such as a serious diagnosis, ripples through the others, and the family either settles into a new balance or struggles to readjust. A third person multiplies the ties: a child relates to each parent and also to the parents' relationship.

Minuchin's structural family theory ties well-being to clear authority, subsystems such as parents and children, and boundaries that keep members connected yet separate. Rigid boundaries leave members disengaged with weak bonds; loose ones produce enmeshment that crowds out privacy. Among 348 Ohio college students, disengagement went with poorer family communication and, through it, worse self-rated mental health; enmeshment did not.

Triangulation happens when a child becomes entangled in the parents' conflict, either pulled in by them or feeling caught in the middle and pressed to take sides. Bowen's theory describes a related pattern: anxiety in a two-person bond often leads one or both to draw in a third. In a cross-sectional study of 75 school-aged children, triangulation correlated with emotional and behavioral problems, as did threat appraisals and self-blame.

The theory, refined by Ritchie and Fitzpatrick from earlier work by McLeod and Chaffee, has two dimensions. Conversation orientation is how far a family encourages members to share feelings and ideas on many topics; conformity orientation is how far it stresses shared values and beliefs. A family can score high or low on both, though a meta-analysis of 32 studies suggests the two tend to be inversely related.

Sibling bonds typically outlast ties to parents and to spouses, which makes them some of the longest relationships in a life. Researchers describe them on three dimensions: warmth, conflict, and rivalry, meaning competition for parents' love and attention. Most studies find all three run less intensely in adulthood, as siblings build lives outside the family, with warmth still outweighing conflict and rivalry.

Using 38,000 UK life histories, one study found parents' separation was tied to a much higher risk that their children's own unions would dissolve, whatever the parents' social class. Parenting also carries across generations, but in a UK cohort of 146 mothers and 146 fathers the links were modest. A small interview study in India found harsh parenting dropped in the next generation where outside support was present.

A family of choice is people, such as close friends and partners, who are not biologically related but are treated as family; Kath Weston introduced the term. In an online survey of 432 LGBTQ+ adults aged 50 and older in Israel, most had a chosen family of about five people. Those ties were less stable than family-of-origin ties but more accepting and less negative, and having more of either kind went with better mental health.

Peer acceptance is how well a child is liked by classmates as a group, while friendship is a tie with particular children. Among 650 Romanian children aged 9 to 12, reporting more friends was linked to life satisfaction, and peer acceptance to school grades. Earlier work holds that a good best friendship can buffer peer rejection, but in 189 children aged 9 to 11 it did not significantly buffer peer problems' link to emotional problems.

First romances typically start around ages 14 to 15 and tend to be short, growing more stable with age; one study found average length rose from about 5 months at 15 to 21 months at 21. Researchers see them as a setting for building skills used in later relationships. In a German panel of 2,457 young people, those who never dated reported lower life satisfaction and more loneliness than moderate daters did.

Arnett named the years from about 18 to 25 emerging adulthood: a self-focused period before marriage, children or a settled career, marked by identity exploration, feeling in between, and instability in housing, relationships and work. Writers on US young adulthood describe marriage and children moving later. Its questionnaire was built on mostly white, female US students, and a Flemish sample did not reproduce all its factors.

A 2022 meta-analysis of 49 studies found marital satisfaction drops by a medium amount from pregnancy to a year after birth and by a small further amount in the second year; non-parents declined slightly over a similar span. Partners' declines tracked each other. The drop did not differ significantly between mothers and fathers or by age, and in the second year it was smaller in studies using one older scale.

Family life-cycle accounts place many couples around ages 40 to 50 between raising teenagers and dealing with their own parents' aging, followed by children leaving home and partners readjusting to life as a pair. Parents now give grown children more support, on average, than parents did in the recent past. A Portuguese survey of 331 married people found slightly lower satisfaction in families with young children or teenagers.

Socioemotional selectivity theory holds that older adults keep fewer but closer relationships to meet emotional needs, though studies of its core ideas have not given consistent results. In Australian panel data on 749 widowed people, a spouse's death was strongly associated with more loneliness, yet also with more contact with friends and family outside the home and no change in social support.

Kubler-Ross's five-stage model is still widely repeated, but researchers criticize using it as a guide to grieving; one study found well-being swung back and forth after a loss rather than moving through steps. Older grief theory held that mourners must detach from the dead, and Klass and colleagues proposed continuing bonds, an ongoing inner relationship with the deceased, instead. Whether such bonds always aid adjustment is debated.

Rusbult's Investment Model holds that satisfaction, investment and the perceived quality of alternatives shape commitment, which in turn predicts whether a relationship lasts; a 2019 survey of 287 US adults replicated the links to commitment. Satisfaction alone does not map one to one onto staying together: in an exploratory study of 101 couples, emotion patterns in a positive talk, not a negative one, predicted separation two years later.

Duck's 1982 model treats a breakup as a gradual, socially embedded process rather than a single event. In the intrapsychic phase one partner privately weighs the costs and imagines alternatives; in the dyadic phase grievances are aired and repair or separation is decided; in the social phase others are told and recruited for support; in grave-dressing partners build an account that protects self-esteem and seek closure.

Endings range from a face-to-face talk to ghosting, where one person cuts all contact without warning or explanation. It is described as sparing the leaver the discomfort of giving bad news while leaving the other in silence. Several studies link being ghosted to loneliness and lower self-esteem, yet a survey of 626 adults found no significant link between being ghosted in the past year and life satisfaction, loneliness or helplessness.

Forecasting research describes an impact bias: people expect events, breakups among them, to affect them more strongly and for longer than they do. Critics say it depends partly on how accuracy is scored, and one study found forecasts understated how intense events felt, especially bad ones. In two retrospective breakup surveys, more distress went with more self-reported growth; in one, reflection and brooding both carried that link.

In a Dutch cohort of 2,230 children followed from ages 10 to 12 into their early twenties, emotional and behavioral problems rose after a parental divorce but not before it, by a small amount that persisted and grew over the follow-ups. Divorce researchers name children's exposure to conflict between their parents as the most damaging part of divorce for children's adjustment and well-being.

When a couple with children divorces, one of the hardest tasks is building a good co-parenting relationship, whose quality divorce researchers tie to how well children adapt. In two cross-sectional samples of Dutch divorced parents, co-parenting conflict was higher when parents felt their friends and family disapproved of the ex-partner, and lower forgiveness of the ex partly explained that link.

Loneliness is a painful sense of being alone or cut off, while social isolation means having few contacts and few people to talk with regularly. The two are related but not the same: living alone does not guarantee loneliness, and a person can feel lonely in company. A conceptual review notes that loneliness, isolation, living alone and solitude are often treated as interchangeable when they are not.

Cacioppo's evolutionary model treats loneliness as a signal that social connections need repairing or replacing, a kind of social hunger that tilts attention toward social cues the way hunger tilts it toward food. The model adds that loneliness also flags missing protection, so it can raise self-protective caution alongside the urge to reconnect. In two surveys, lonelier people preferred more distance from others within intimate space.

Theoretical accounts propose that loneliness raises vigilance to social threat, which might show as reading a neutral remark as an attack or withdrawing from a group, and such shifts are thought to help keep loneliness going. The NIA notes chronic loneliness can leave a person feeling threatened and mistrustful. Yet a pooled analysis of four datasets found no link between loneliness and attention captured by negative social words.

Social loneliness is the felt lack of a wider circle of contacts, while emotional loneliness is the felt lack of a close, intimate relationship. In a two-wave study of 3,275 adults, emotional loneliness was linked in both directions with every mental health measure, while social loneliness tracked social isolation and predicted only social anxiety. The authors suggest emotional loneliness may matter more for symptoms.

The NIA lists higher risks of high blood pressure, heart disease, weakened immune function, depression, cognitive decline, dementia and death among people who are lonely or isolated. These are associations, and the arrow can point both ways, since poor health can itself make isolation more likely. A review of 12 longitudinal studies of older adults noted that some suggest a two-way link with cognitive decline.

A 2024 umbrella review of 29 systematic reviews of randomized trials grouped interventions as social, such as contact and support; psychological, such as therapy and social skills training; and digital. Psychological approaches looked most promising for loneliness and social ones for isolation, but effects were small to moderate, varied widely and came from reviews of mixed quality. Digital approaches showed mixed or no effect.

The NIA's list of risk factors for loneliness and isolation includes conditions beyond the person: lack of transportation, financial struggles, living alone, and living in a rural, unsafe or hard-to-reach neighborhood. In a survey of 917 New Zealanders aged 60 to 100, views of housing and of neighborhood access, safety and cohesion helped explain both social and emotional loneliness.

Dating apps let users browse a large pool of profiles before any meeting. In some studies, mostly speed-dating ones, stated preferences showed little power to predict real choices, though some blame that method. In 2018 Swiss survey data, couples who met on apps were about as satisfied as others and more often crossed education levels or distance. A review of 70 app studies found all were cross-sectional.

In a study of 192 German adults already in relationships, models built from both partners' personality explained 37 percent of the variance in relationship quality four years later, mostly from each person's own traits; partner and similarity effects added no further prediction. Speed-dating data show later romantic interest tracks both compatibility, one person's unique liking for a partner, and that partner's broad appeal.

Dating apps can show users very large pools of potential partners, and research on online dating found that more profiles led to more searching, more time on poor fits and a lower chance of choosing the best one. In a survey of 804 Greek speakers, believing they had more options went with being single less often overall, but also with more choice overload, which in turn went with a higher chance of being single.

Walther's social information processing theory argues that without nonverbal cues, people lean on words and can reach similar intimacy. His hyperpersonal idea goes further: selective self-presentation, time to edit and freed-up attention may yield better relational outcomes than meeting in person. Prior work finds online daters embellish some details, yet interviews with men on dating apps found an expected meeting curbs deception.

The displacement hypothesis holds that time online replaces face-to-face contact and so lowers well-being. A 2024 review argues social media time probably comes out of TV and web browsing instead, and that already social people use it to keep in touch. In abstention studies people mostly filled the time with browsing, work and chores, and some had fewer social interactions. Experiments that directly test displacement are rare.

Partner phubbing means ignoring a romantic partner in favor of a phone. A 2025 meta-analysis of 52 studies found it went with lower satisfaction, intimacy and closeness and with more conflict and jealousy, but most studies were cross-sectional, so phubbing may harm relationships or distress may invite it. A close replication of a claim that a phone merely lying nearby lowers strangers' rapport did not find the effect.

A parasocial relationship is a lasting, one-sided bond with a media figure a person has never met, felt as something like friendship; it differs from parasocial interaction, the sense of exchange during a single viewing. Research studies their role-model function and a compensating role when real-life ties are lacking. Most of this work uses correlational surveys of young, mostly female samples, which cannot trace cause and effect.

Researchers commonly describe sexual orientation as linked dimensions: whom a person is attracted to, their sexual behavior, and how they identify. These need not match; some people who identify as heterosexual report same-sex attraction or behavior. Newer models separate romantic from sexual attraction, which overlap without being identical, and cited research finds behavior and identity shift more over time than attraction.

Sexual desire discrepancy is when one partner's desire is higher or lower than the other's, and it is among the most common reasons couples seek therapy. One study of coping strategies treats it as a normal part of relationships, asking how couples manage it rather than how to restore lost desire. In a representative German sample of couples, about half of partnered adults said they wished for more sex than they had.

Spontaneous desire appears unprompted, as sexual thoughts or fantasies; responsive desire arises in reaction to erotic cues or stimulation, and ICD-11 counts a lack of either as a sign of low desire. Basson's model, used in treatment research, describes some women with low desire starting from neutral and developing desire in a welcome encounter. Yet most desire questionnaires treat desire as a stable trait measurable in any context.

In a representative German study of 964 couples, the quality of sexual communication predicted sexual satisfaction in both sexes, while how often couples had sex predicted it only for women. An online survey of 369 partnered people, mostly women, linked better sexual communication to a smaller perceived desire gap, via higher satisfaction. Both are associations, and frequency is a poor gauge of desire, since people have sex for many reasons.

Sexual consent is generally defined as agreement by someone able to agree, given by choice and without threat, force or coercion. Affirmative-consent definitions in law and from advocacy groups describe it as conscious, voluntary agreement and an ongoing yes rather than a one-time event: it can be withdrawn at any point and cannot be assumed from past sex, a private setting or being in a committed relationship.

Relationships can be monogamous or consensually non-monogamous, meaning partners openly agree that one or more of them may have other sexual or romantic partners. In such arrangements exclusivity is negotiated rather than assumed. A 2024 review found the two kinds function comparably, but a survey of 1,528 LGBTQ+ adults in Quebec found lower well-being in a non-monogamous group, and findings overall are inconsistent.

The interpersonal exchange model of sexual satisfaction predicts that sexual satisfaction rises as relationship satisfaction does, and newer longitudinal studies suggest the reverse also occurs, so the link may run both ways. In a two-wave study of 1,001 young newlywed couples, those high in sexual satisfaction but poor at managing money did not differ significantly in later marital satisfaction from couples doing well on every measure.