deciqAI Reciprocity
Overview
The rule of reciprocity — if someone gives you something, you owe them something back — recurs across virtually every documented human culture (Mauss 1925; Gouldner 1960). The rule operates below deliberation, scales asymmetrically, and fires even when the favor was unrequested or from someone you dislike. Regan (1971): a 10¢ Coke produced ~50¢ in compliance, and liking stopped predicting behavior once a favor was in play. Three operating properties: (1) asymmetric exchange — repayment routinely exceeds the favor; (2) override of liking — the obligation does the work; (3) "no obligation" disclaimers are part of the install, not an exception.
Composes with social-proof, anchoring (door-in-the-face combines reciprocity with concession-anchoring), and repeated-games-reputation.
When to Use
Apply when: accepting a gift/sample/concession where the giver has a future ask; designing sales, fundraising, or partner-development sequences; a negotiation counterpart just conceded and you feel pulled to match; evaluating cumulative small gifts in regulated relationships; weighing free AI credits / a generous free tier / a free vendor pilot where a paid contract or lock-in is the eventual ask; someone asks "why are they so generous?" or "should I concede back?"
When NOT to use: exchange is explicitly priced; gift is a normal friendship pattern with no pending ask; favor is too small to constrain any future action; bright-line legal/ethical rules apply — follow the rule, skip the analysis.
Coaching Novices (Adaptive Front Door)
- Engine mode: user has a concrete situation → run The Process directly.
- Coach mode: user is unfamiliar or has no concrete case → guide step by step.
In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.
- One-line what-it-is: if someone gives you something — even small, even unwanted — you'll feel a strong pull to repay, and the repayment tends to be bigger than what you got, even when you don't like the giver.
- Check fit against When to Use / When NOT to use. If it's commerce or normal friendship, point elsewhere.
- Elicit their real situation — a concrete current case. Never run the analysis on hypotheticals.
[WAIT — do not advance until user responds]
- Run The Process one step at a time — name the favor, cost to giver, eventual ask, counterfactual test.
[WAIT — do not advance until user responds]
- Close by naming the one concrete move — accept, decline, restructure, or counter — with the reason.
[WAIT — do not advance until user responds]
The Process
Run the Reciprocity Analysis.
- Name the favor. What exactly is being given? Be specific — vague favors cannot be analyzed.
- Estimate cost to giver. Real dollars, time, risk. Pull is proportional to perceived cost to giver, not value to you.
- Estimate value to you. The gap between cost and value is where calibration goes wrong.
- Identify the eventual ask. What does the giver want — now or later? If you can't articulate it, either there isn't one (analysis stops) or the install is still being staged.
- Run the counterfactual. Would you say yes if asked directly, absent the gift? If yes — install is benign. If no — the rule is converting a "no" into a "yes." Decline the install, not the eventual ask.
- Check for door-in-the-face. Large ask → decline → smaller ask → you feel pulled to match the concession. Naming it weakens the pull.
- Decide: accept and reciprocate at par / decline up front / accept with explicit frame reset / accept then evaluate the ask on its own merits in writing.
- Audit cumulative install. Analyze the whole relationship, not just the latest favor — especially in regulated professions.
Output: Reciprocity Analysis
Favor: [what / cost to giver / value to me]
Eventual ask: [probable ask / time horizon / explicit or unstated]
Counterfactual: [would I say yes cold? yes → benign; no → decline install]
Structure: [door-in-the-face? concession-anchoring? cumulative?]
Decision: [move + why + how to execute]
→ Method in Action: Dennis Regan's Coke Experiment, Cornell, 1971 → 2026 lens: Free AI credits, freemium tiers & open-weight releases (2023–2026) — reciprocity as product-led growth, and the buyer's defense.
Pack: Reciprocity in Practice
- B2B sales: give weeks before asking; thoughtful > generic. Buyer: counterfactual test — would I sign absent this relationship?
- Fundraising: free-gift enclosures lift response rates; evaluate cause on merits, not the unsolicited gift.
- Negotiation: door-in-the-face = open extreme → retreat → counterparty feels pulled to match. Defend by anchoring on merits, not concession magnitude.
- Lobbying/regulation: cumulative install across meals and invitations. Bright-line "no gifts" rules exist because per-event analysis underestimates aggregate effect.
- Founder-investor: write comparison criteria before value-add is delivered; don't update them based on subsequent generosity.
Applying It Well
- Give only when you're willing to give regardless of return — genuine value, not a manufactured trigger
- Signal no-obligation explicitly when you mean it: "no expectation — just thought you'd find this useful"
- In regulated relationships, institutional rules override per-event analysis
→ Primary sources: references/sources.md
Common Rationalizations
[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.
| Fake move | Reality |
|---|---|
| [D] "It's just a small gift, it doesn't influence me" | Regan 1971: a 10¢ Coke produced ~50¢ of repayment. Small gifts are individually deniable, statistically reliable. |
| [D] "I'll accept but I won't be influenced" | The rule operates automatically; deliberate intent to resist provides only partial protection. If the obligation matters, decline up front or evaluate the ask in writing before it arrives. |
| [D] "I genuinely like them, that's why I'm helping" | Regan's finding: in the favor condition, liking dropped out as a predictor. If liking has stopped predicting your behavior, obligation has taken over. |
| [D] "It's only a problem if the gift is large" | Wazana 2000, Dana & Loewenstein 2003: even small gifts produce significant behavioral effects. Bright-line rules exist because per-event analysis underestimates cumulative impact. |
| [D] "They said no obligation, so it's genuine" | The disclaimer is part of the staged install — it prevents refusal; the ask comes later. |
| [D] "Door-in-the-face isn't real, it's just negotiation" | Cialdini et al. 1975 (JPSP): one of the most replicated influence findings. Naming the technique reduces but does not eliminate its pull. |
| → Add [O] entries here after each real use — paste the actual failure pattern | What went wrong and why |
Red Flags
- Someone you barely know is unusually generous and you can't articulate why
- "No obligation" precedes a favor that precedes a significant ask
- Negotiation opens extreme, then "concedes" to a moderate position
- You're defending a vendor's behavior more than the merits warrant
- Small gifts have accumulated over time and you've never run the cumulative analysis
Verification
- Favor and cost-to-giver are named specifically
- Eventual ask is articulated explicitly
- Counterfactual test performed: would I say yes approached cold?
- If counterfactual fails: decision is decline/refuse, not accept-and-rationalize
- Structural patterns (door-in-the-face, concession anchoring, "no obligation") were checked
- Cumulative install across many small favors was audited
- If sender: intent is named; receiver is not someone whose deliberation you wish to subvert
Part of deciqAI Knowledge Skills — 237 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. See it run → https://www.deciqai.com/s/reciprocity · Built by deciqAI · github.com/deciqAI · Contributions welcome.
Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/reciprocity.json