Week 9: Escalation Rates

Suppose you’ve had three good conversations with someone from your lab. Really good ones. And yet it doesn’t feel like a friendship, and you can’t tell what’s missing.

Or: someone you’ve met twice told you about their parents’ divorce in the café queue, and you didn’t know what to say back.

Both are the same undocumented system, running at different speeds. This week is about its pace.

Friendship costs hours#

Hall asked adults who had recently moved about a new acquaintance, and followed first-year students over nine weeks. In both studies, hours spent together were associated with becoming closer. Time spent on leisure activities predicted closeness, and so did everyday talk (Hall, 2019).

We couldn’t read the full paper, only its abstract, so this page quotes no hour counts from it. It gives no timetable for your particular friendship. If three good conversations have left you wanting more, another chance to spend time together is a place to start.

Disclosure is the engine#

Hours are the clock. Disclosure is the engine. Collins and Miller combined the research on self-disclosure and liking, and found three effects, all significant (Collins & Miller, 1994):

1. People who make more intimate disclosures tend to be liked more.
2. People disclose more to those they already like.
3. People come to like those they have disclosed to.

The third gives the listener a role too: the other person is disclosing to you. Listening is a move, not a pause between moves.

The effects vary with the kind of disclosure and the setting. “Disclose more, get liked more” is not a rule you can run in a café queue.

Which side gains, and for how long#

Sprecher, Treger and Wondra asked how the two roles affect liking. Pairs of unacquainted undergraduates did a structured disclosure task. One was randomly assigned to disclose; the other to listen. Then they swapped and did it again (Sprecher et al., 2013).

After the first interaction, the listeners reported more liking and other positive impressions than the disclosers did. Listening was also associated with greater perceived similarity.

Those differences disappeared after participants switched roles.

In this task, the difference between assigned roles disappeared after both people had listened and disclosed. That is part of the evidence under the rule in §4: pay attention to both sides of the exchange. It does not validate the course’s five levels or its plus-or-minus-one window.

The scope is narrow and worth saying: strangers, a structured task, two interactions. It tells you nothing about how depth paces over months.

Escalation works in steps#

Aron and colleagues found a way to generate closeness in a lab. Pairs of strangers spent 45 minutes on self-disclosure tasks that gradually escalated in intensity, and ended up closer than pairs who spent the time on small talk. Making “get close” an explicit goal made no significant difference. Neither did matching people on attitudes (Aron et al., 1997).

The word to take from that study is gradually. The procedure moves both people up together, a step at a time. It is not evidence that jumping straight to depth works outside a lab, where nobody has agreed to the steps.

The rate limit#

Here is the course’s model of the undocumented pace. The levels and the rule are ours, not the literature’s; test them in your journal.

For voluntary escalation: the model suggests the other person’s last level, plus or minus one. Step up one level at a time, and wait for them to step before stepping again.

There is no minimum disclosure obligation. You can hold a lower level, step down, change topic or pass, including outside the suggested window. Someone else’s level 4 does not require your level 3.

In the café-queue example, the model marks a jump from level 0 to level 3, with no step from the listener in between. For the three good conversations, it gives you a question: did either person offer anything beyond level 1?

Signals you’ve gone past a rate limit#

The course’s working list, for testing in your journal:

• the other person’s replies get shorter;
• the topic comes back to level 0 (“anyway, did you do the reading?”);
• a laugh or a joke where a response to the content would go;
• they physically start to leave.

In the room — one turn each (9 min)#

One minute, in pairs. Read the prompt: “What took longer than you expected this week?” Use a minor real example, or this fictional one: Alex waited for a late bus, missed a library booking, and found another place to work. You may pass on personal content and use Alex, or observe.

Three minutes. A answers; B listens and may ask follow-up questions without matching the disclosure. Spend the last thirty seconds privately noting your own feeling towards your partner: warmer, unchanged, cooler, or unsure.

Three minutes. Swap roles and repeat, including the private note.

Two minutes, privately. Compare your own two notes. What changed, if anything? You do not know your partner’s rating unless they choose to tell you.

This is practice in the two roles, not a replication of Sprecher et al. Neither an individual feeling nor a fictional exchange can establish the study’s group result. “Unchanged” and “unsure” are useful answers.

The step-down#

Week 5’s repair, applied to depth. Name it in one turn and go down a level:

“Sorry, that was a lot for a Tuesday. How was your weekend?”

The step-down tells the other person you noticed, and that they don’t have to match level 3 in a queue.

Did it land?#

The ladder has rungs and no receipt. It tells you how far a turn went; it says nothing about whether the person you said it to came away feeling heard.

Laurenceau, Feldman Barrett and Pietromonaco had people fill in a record immediately after every social interaction — one week in the first study, two in the second. For each one: what you disclosed, what they disclosed, how responsive they seemed, and how intimate it felt (Laurenceau et al., 1998).

The findings linked intimacy in a single interaction to both people’s disclosure, with perceived partner responsiveness as a partial mediator. What was shared and how the partner seemed to receive it both mattered.

So pacing needs more than a rung number. Ask how the other person seemed to receive the turn, without treating a personal disclosure in return as owed.

Emotion outranked facts#

The detail that changes the advice is in their second study. Disclosing emotion predicted intimacy better than disclosing facts and information.

Read that against the ladder and something useful falls out. A step up does not have to mean a more private subject. It can be the same subject with the feeling attached:

The subject stays the same; the right column adds what it means to the speaker. That feeling may itself be private. Use the extra clause when you want to share it, and leave it out when you don’t.

The move nobody expects to work#

Huang and colleagues studied three sets of live conversations between pairs: two online get-acquainted conversations and a field study of face-to-face speed dating. People who asked more questions, and follow-up questions in particular, were liked more by the person they were talking to. People instructed to ask more questions were rated higher in responsiveness — the construct the paper glosses as listening, understanding, validation and care — and responsiveness is what explained the effect (Huang et al., 2017).

A follow-up question picks up what the other person just said. A switch question starts a new topic. Both are questions; they are not the same move:

Them: “I’ve been trying to get the stats assignment finished all weekend.”

Switch: “Have you chosen your elective yet?”

Follow-up: “Which part of it is taking the time?”

A follow-up lets you take a turn — which §4 says to make room for — without disclosing something of your own. It is one option when the other person has gone to level 3 in a café queue and you do not want to match that depth.

Then the last line of the abstract:

people do not anticipate that question-asking increases interpersonal liking.

In the room — the follow-up (8 min)#

Three minutes, on your own. Take a sentence someone said to you this week — any sentence, dull is fine — and write three follow-up questions to it. Not three questions. Three questions that could only have been asked of that sentence.

If you have no example to hand, use: “I tried finding a quiet place to work, but the library was full.” Real examples should contain no identifying details.

Three minutes in pairs: read yours out. Your partner says which words in the original each one picks up. If they can’t find the words, it’s a switch question wearing a follow-up’s clothes.

Two minutes, the whole room: compare at most two examples. Which question followed something the speaker said, and which introduced a new topic?

The part where the research argues with itself#

In 2025 the journal published a correction to it. An independent audit found several minor errors in how some results were reported, and stated that all the conclusions in the paper are valid, that the substantive result of every hypothesis test is unchanged, and that the corrected results and the audit materials are posted publicly.

In 2019 Kluger and Malloy published a reanalysis of the speed-dating study. Their argument: being offered a second date is not the same thing as liking; follow-up and switch questions weren’t distinguishable as latent scores; and about a third of the variance in how much someone asks questions belongs to the particular pair, not to the person. Huang’s group published a reply saying the reanalysis tested the wrong variables.

This course is not going to adjudicate a social relations model it hasn’t run. For this week’s practice, keep three points in view:

1. the effect in the two lab conversations, which the reanalysis doesn’t touch;
2. a second date is a proxy for liking, not liking — worth holding onto, because the depth scale above is also a proxy;
3. the dyad point, as a claim you can test: how much you ask questions may depend less on you than on who you are with.

And then notice the shape of all that. A numbered correction. A signed objection. A reply. Anyone can read the lot and decide.

Nobody has ever issued an erratum for “how are you”.

Choosing your own rate#

The rate limit isn’t a speed you’re obliged to reach. You can keep someone at level 1 for a whole semester, on purpose. Some people run the protocol with a high limit and some with a low one, and neither is the correct setting.

Declining escalation costs something, and it belongs in your week 6 ledger: the other person may read a steady level 1 as distance. If you want them to read it differently, say so, once: “I’m slow to talk about personal stuff, but I like these conversations.” That’s a hold token for depth.

In the room — follow, switch or step down? (8 min)#

Two minutes, in pairs. Read this fictional turn: “I’m still sorting out the new flat. Anyway, what time is Thursday’s lab?” Compare three replies:

• Follow-up: “Which part of the move is taking the time?”
• Switch: “Have you chosen your elective yet?”
• Step-down: “14:05. We can leave the flat stuff there.”

Three minutes. Choose a reply, or write a fourth. State which words in the turn support your choice, what your reply asks of the other person, and what you still do not know. None of the three grants access to a private subject.

Two minutes. New turn: “Actually, I would like to talk about the move for a bit, if you have time.” What changes? You can accept, limit the time, or decline; write the sentence rather than merely naming the choice.

One minute. Record the clue that changed your response, or why your own limit stayed the same. A topic invitation and an obligation are different messages.

This week’s lab starts the Interoperation Report#

On Thursday you meet your Interoperation Report partner. Over the labs of weeks 9, 10 and 11, the two of you negotiate a shared protocol for how you work together, and report what each of you conceded and what held. The rate limit is a good first thing to agree on.

If time remains#

How could two people make room for closeness without treating disclosure as a debt the other person has to repay?

Optional discussion. Finish by 15:55.