What to expect from a friend: the six standards people actually hold

Reliably: loyalty, trustworthiness and mutual regard. Hall (2012) analysed ideal friendship expectations across a series of studies and found they sort into six dimensions, with symmetrical reciprocity — that cluster of loyalty, trust, and being held in genuine positive regard — rated highest and most consistently. Everything else is real but variable. Some people expect a friend to be practically useful in a crisis; others expect deep self-disclosure; others mostly want to enjoy themselves. Expecting all six at full intensity from every friend is the fastest route to chronic disappointment, because almost nobody delivers on all six.

Hall (2012) identified: symmetrical reciprocity (loyalty, trust, mutual regard, support), communion (self-disclosure, emotional intimacy, vulnerability), agency (the friend has resources, status, intelligence or humour worth being around), enjoyment (fun, companionship, unstructured time together), instrumental aid (practical and material help when needed), and similarity (shared interests, values and background). The first is the near-universal standard. The other five are where individual people differ enormously, and where two friends can each be behaving impeccably by their own definition while disappointing the other.

Symmetrical reciprocity, and it is not close. In Hall's (2012) data it was the highest-rated dimension and the one that most clearly separates a close friend from a pleasant acquaintance. This is intuitive once stated: you can enjoy someone's company enormously, share every interest, and still not consider them a friend if you suspect they would not hold a confidence. The reverse rarely holds — a person you trust completely and rarely see is still filed as a friend. Trust does the categorising; everything else adjusts the closeness.

On average, yes, though the overlap is large. Hall's (2011) meta-analysis found women held higher overall expectations for same-sex friends, with the largest differences on symmetrical reciprocity and communion — loyalty, intimacy and self-disclosure. Men placed proportionally more weight on agency and shared activity. Two cautions. These are group averages with heavy individual variation, so they predict nothing about a specific friendship. And in cross-sex friendships, expectations tend to converge on general friendship norms rather than tracking gender stereotypes.

Yes, if 'always' means available. No, if it means reliable. The realistic standard is responsiveness over time, not availability on demand: a friend who misses three messages during a hard month at work has not broken anything, and a friend who is consistently unreachable exactly when you are struggling has. Argyle and Henderson (1984) catalogued the informal rules people actually hold for friendship, and standing up for the friend in their absence and keeping confidences ranked far above frequency of contact. Presence is the wrong metric. Direction of effort under load is the right one.

Because friendship has no ceremony. Romantic relationships get an explicit conversation about exclusivity, commitment and often cohabitation; friendships get nothing. There is no moment at which two people agree on terms, so the terms stay implicit until one is breached — at which point the aggrieved party is arguing for a rule the other person never signed. This is also why friendships end by fade rather than by conversation. Nothing was ever formally agreed, so there is nothing to formally end.

Run three checks. Does the expectation require mind-reading — would a reasonable person have known without being told? Does it require more than one dimension at full strength from a single friend, such as deep intimacy and constant availability and practical help? And does it hold your friend to a standard you are not currently meeting yourself? An expectation failing any of these is likely a preference that has been quietly promoted. The reciprocity standard, by contrast, survives all three, which is part of why it is the one worth defending.

A standard is what you will accept, with the action clause belonging to you: 'I need to be able to trust what a friend tells me, and I pull back from people I cannot.' An expectation is a rule assigned to them, usually unannounced: 'she should have checked in after my interview.' Standards create clean decisions; unspoken expectations create grievances. Most friendship resentment is expectation-shaped rather than standard-shaped. The full distinction is in standards vs expectations.

Almost never, outside of safety. Ghosting bypasses the reciprocity standard directly: it withholds the one thing — clear communication and basic regard — that people rate as the core of friendship. Freedman and colleagues (2019) found beliefs about ghosting track people's implicit theories of relationships, with destiny-oriented people rating it more acceptable. That explains the behaviour without excusing it. The exception is genuine: where a relationship has become unsafe or coercive, disappearance is a protective strategy, not a courtesy failure. Ending things badly and ending things safely are different problems.

Say the level out loud rather than delivering less and hoping they adjust. 'I am not good at frequent contact and I never will be, but I will show up when it matters' is a real offer, and most people would rather have an accurate small one than an inaccurate large one. The damage in mismatched friendships rarely comes from the size of the gap; it comes from one person continuing to expect a level the other stopped providing months ago. Naming the dial position ends the guessing.

For a while, and then it converts into something else. Sustained one-sidedness violates symmetrical reciprocity, which is the dimension people rate most highly, so the friendship gradually gets refiled as an acquaintance in the mind of whoever is doing more. That refiling is usually silent. Before concluding anything, check whether the imbalance is situational — new baby, illness, a brutal work period — because those recover, and permanent-seeming asymmetries often are not. Our one-sided friendship quiz is a structured way to work out which one you are in.

Very few, which is the point. Full symmetrical reciprocity plus communion is expensive in time and attention, and most people sustain it with a handful of people at most — Dunbar's (1992) layered model puts the innermost circle at around five. The practical implication is not to lower the standard but to apply it selectively: hold the full bar for the small group you actually want it with, and let the wider circle be exactly what it is, which is enjoyable and shallower. We work through the sizing in how many friends you can actually maintain.