关于「权重从哪里来:条件化学习、解释与仪式的可加性」,《开放标签与价格标签》给出的定量结论是什么?
结论:p41 说明,先验的权重可以由经验建立:四次真实缓解之后,明说是安慰剂的药丸仍带来 11.82 分的感受缓解,六次之后才降到 9.97。洛赫尔与卡姆-汉森的研究说明,解释与标签移动的是期望的均值。p42 说明,仪式与关系在概率单位上各高约 0.432 与 0.456,近似可加,而开放标签相对双盲的损失只有仪式台阶的 −0.09。下一章把同一套机制翻过来:期望伤害时会怎样。
数据口径:书中参数为示意;全部计算由书中程序生成;本书不是医疗建议,不宣称任何产品或做法对身体或健康有任何作用,书中流程与工具为非个人化经营建议。 书中以标注框区分已发表证据、可运行的 Python 计算(21 段程序)、原典与作者假说。
对照数据表
下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 11.82, 9.97
| session | phase | prior mu | prior sd | w | perceived relief |
|---|---|---|---|---|---|
| 1 | active | 0.00 | 20.10 | 0.198 | 15.00 |
| 2 | active | 12.95 | 7.70 | 0.628 | 15.00 |
| 3 | active | 13.93 | 5.90 | 0.742 | 15.00 |
| 4 | active | 14.31 | 5.15 | 0.790 | 15.00 |
| 5 | OLP | 14.51 | 4.77 | 0.815 | 11.82 |
| 6 | OLP | 13.81 | 4.56 | 0.828 | 11.43 |
| 7 | OLP | 13.22 | 4.44 | 0.835 | 11.05 |
| 8 | OLP | 12.71 | 4.37 | 0.840 | 10.68 |
| 9 | OLP | 12.25 | 4.32 | 0.843 | 10.32 |
| 10 | OLP | 11.81 | 4.30 | 0.844 | 9.97 |
| final expectation 11.40, sd 3.79 |
查看完整验算程序
# Chapter 4. p41: How experience sharpens a prior, and how it fades.
# A one-dimensional Kalman filter learns "how much relief this ritual
# brings". Sessions 1-4: an active intervention gives 15 points of
# relief; sessions 5-10: an open-label placebo brings no active relief,
# so what is felt is only the placebo pull w * mu (precision weight of
# p31), and the filter learns from that felt relief.
# All values illustrative: start mu 0, sd 20; observation sd 8; drift
# (forgetting) variance 4 per session; signal sd for w is 10.
# 条件化与消退:经验让先验更精确,落空的期望让它慢慢退回。
import math
mu, var = 0.0, 20.0 ** 2
OBS_VAR, DRIFT, SD_L = 8.0 ** 2, 4.0, 10.0
print("session phase prior mu prior sd w perceived relief")
for t in range(1, 11):
var += DRIFT
w = (1 / var) / (1 / var + 1 / SD_L ** 2)
active = 15.0 if t <= 4 else 0.0
pull = w * mu
perceived = active if t <= 4 else pull
print("{:7d} {:<7} {:8.2f} {:8.2f} {:.3f} {:8.2f}".format(
t, "active" if t <= 4 else "OLP", mu, math.sqrt(var), w,
perceived))
obs = perceived
gain = var / (var + OBS_VAR)
mu += gain * (obs - mu)
var *= (1 - gain)
print("final expectation {:.2f}, sd {:.2f}".format(mu, math.sqrt(var)))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《开放标签安慰剂预测编码、价格标签期望增益与杯弓蛇影反安慰剂母典》第 4 章「权重从哪里来:条件化学习、解释与仪式的可加性」(版本 1.0.0). 5D Library. DOI 待发布
DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-201 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026201,
author = {Fan, Yafu},
title = {Open-Label Placebos and Predictive Coding, Price-Tag Expectancy Gains and the Nocebo of the Bow in the Cup},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}