5D Library › Relationship Rupture Survival Analysis, No-Contact Memory Decay and Reconciliation Game Windows · 中文
Chapter 3. The Reactance Ratchet: Why Early Pressure Backfires
Pressure while the other person is defensive
When people feel that a freedom is being taken from them, they push to restore it, and the thing being pressed on them becomes less attractive. This is psychological reactance (Brehm 1966). After a rupture, repeated messages, calls and visits that the other person did not ask for can read as exactly such pressure, and they arrive when defensive arousal is at its highest. p05 adds this to the model of Chapter 2. Each unwanted contact raises negative arousal, by more if it arrives while the negative side still outweighs the positive, and part of each rise turns into a resistance that fades only over a year: a ratchet.
Calculation
# Chapter 3. p05: The reactance ratchet. The receiving partner starts
# with N = 1.5 (tn 17.5 days) and P = 1.0 (tp 75 days). Each unwanted
# contact adds 0.25 to N if it arrives while N > P, and 0.05 otherwise;
# a share 0.4 of each addition goes instead into a lasting resistance R
# that fades with a time constant of 365 days (the ratchet). Contacts
# come on random days in days 0 to 30 at f a week; then silence. On day
# 60 one calm message is sent; the chance it is welcomed is
# 1 / (1 + exp(-6 (P - N - R) + 1)). 5,000 runs per row, seed 92.
# The last row moves 3 contacts a week to days 31 to 45 instead.
import random
from math import exp
TN, TP, TR, K = 17.5, 75.0, 365.0, 6.0
def run(rng, per_week, first_day=0, last_day=30, ask_day=60):
n, p, r = 1.5, 1.0, 0.0
for day in range(ask_day):
if first_day <= day < last_day and rng.random() < per_week / 7:
add = 0.25 if n > p else 0.05
n += 0.6 * add
r += 0.4 * add
n *= exp(-1 / TN)
p *= exp(-1 / TP)
r *= exp(-1 / TR)
return 1 / (1 + exp(-K * (p - n - r) + 1)), r
if __name__ == "__main__":
rng = random.Random(92)
print(" contacts a week welcome chance (day 60) lasting R")
base = None
for f in (0, 0.5, 1, 2, 3, 7):
out = [run(rng, f) for _ in range(5000)]
w = sum(o[0] for o in out) / len(out)
base = w if base is None else base
print(" {:15.1f} {:23.3f} {:9.3f}".format(
f, w, sum(o[1] for o in out) / len(out)))
late = [run(rng, 3, 31, 46) for _ in range(5000)]
print(" 3 a week but only on days 31 to 45: {:.3f}".format(
sum(o[0] for o in late) / len(late)))
# contacts a week welcome chance (day 60) lasting R
# 0.0 0.803 0.000
# 0.5 0.672 0.098
# 1.0 0.488 0.236
# 2.0 0.140 0.667
# 3.0 0.017 1.119
# 7.0 0.000 2.649
# 3 a week but only on days 31 to 45: 0.582With no contact in the first month, a single calm message on day 60 is welcomed with chance 0.803. Half a contact a week lowers this to 0.672 and one a week to 0.488; two a week give 0.140, three a week 0.017, and daily contact 0.000. The lasting resistance tells why: after daily contact it stands at 2.649 on day 60, larger than anything the positive memory can offset, and it will still be there months later. Even three contacts a week placed later, on days 31 to 45 when net pull is already positive, lower the chance to 0.582: smaller rises, but they still accumulate.
When to send one message
If pressure hurts, when is a single, calm, low-demand message best? Waiting lets anger fade, but the other person may move on meanwhile. p06 multiplies the chance of a welcome on day by the chance that the other person has not yet moved on, .
Calculation
# Chapter 3. p06: When to send one calm message. Same partner as p05
# but no earlier contact. If the message is sent on day d, it is
# welcomed with chance w(d) = 1 / (1 + exp(-6 (P(d) - N(d)) + 1)),
# but only if the other person has not yet moved on, which happens
# with hazard 1 / tm a day (tm = 120 or 240). The chance of a welcome
# is w(d) exp(-d / tm). Which day is best, and what is lost by
# writing on day 3 or waiting until day 180?
from math import exp
TN, TP = 17.5, 75.0
def welcome(d):
n, p = 1.5 * exp(-d / TN), exp(-d / TP)
return 1 / (1 + exp(-6 * (p - n) + 1))
if __name__ == "__main__":
print(" tm (days) best day chance day 3 day 42 day 180")
for tm in (120, 240, 10 ** 9):
val = [welcome(d) * exp(-d / tm) for d in range(366)]
best = max(range(366), key=lambda d: val[d])
lab = "never" if tm > 10 ** 6 else str(tm)
print(" {:>9} {:8d} {:6.3f} {:5.3f} {:6.3f} {:7.3f}"
.format(lab, best, val[best], val[3], val[42], val[180]))
print(" welcome chance alone peaks on day {}".format(
max(range(366), key=welcome)))
# tm (days) best day chance day 3 day 42 day 180
# 120 29 0.625 0.055 0.587 0.087
# 240 34 0.713 0.056 0.700 0.183
# never 42 0.834 0.056 0.834 0.388
# welcome chance alone peaks on day 42If no one ever moved on, the best day would be the peak of the welcome chance itself, day 42, with chance 0.834. When the other person moves on with a mean time of 240 days the best day comes earlier, day 34 (chance 0.713); with 120 days, day 29 (0.625). A message on day 3 is welcomed with chance only about 0.055 in every case, while one on day 180 has fallen to 0.087 when people move on quickly and 0.388 when they never do. The best moment is thus earlier than the peak of feeling, because the world does not wait.
Hypothesis · not established
The author’s interpretation (not yet tested): unwanted contact in the first weeks after a rupture lowers the chance that a later, calm approach is welcomed, and the effect is larger the earlier and more often the contact came. What would show it wrong: in a pre-registered prospective study recording the timing and frequency of contact after break-ups, the frequency of early unwanted contact is unrelated to how later approaches are received, once the closeness of the relationship and the reason for the rupture are held fixed.
Note
What this chapter is not. The model describes how pressure lowers the chance of a welcome; it is not a timetable for persuading someone. A person who has asked not to be contacted has given an answer, and repeated contact after that can be harassment.
Summary
Pressure while the other person is defensive adds resistance that outlasts the anger; in the model it can take the chance of a later welcome from about four in five to nearly nothing. One calm message, sent after the anger has faded but before life has moved on, is the most the model can recommend, and the answer to it belongs to the other person.
References
Fan, Y. (2026). Relationship Rupture Survival Analysis, No-Contact Memory Decay and Reconciliation Game Windows: Negative-Affect Dissipation Rates, Dual-Process Nostalgia Peaks and Signaling Equilibria [关系断裂生存分析、断联期记忆衰减与复联博弈窗口] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090241 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).