5D Library › Choice Architecture, Cumulative Prospect Theory and Computational Behavioural Economics of Pricing Interfaces · 中文
Chapter 3 Cumulative Prospect Theory: Value Functions, Probability Weights and Rank Dependence
From expected utility to prospect theory
Expected utility theory assumes that people evaluate a risky choice by summing “probability × utility”, with utility defined over final wealth. Kahneman and Tversky’s 1979 prospect theory made three changes: value is defined over changes relative to a reference point rather than over final wealth; losses weigh more than gains of the same size (loss aversion); and probabilities enter the evaluation not linearly but through a decision-weight function (Kahneman and Tversky 1979). The 1979 version weighted each outcome’s probability separately and needed an “editing” phase to remove obviously dominated options. Tversky and Kahneman’s 1992 cumulative prospect theory replaced separate weights with cumulative (rank-dependent) weights, extending the theory to any number of outcomes and to uncertainty (unknown probabilities) (Tversky and Kahneman 1992).
The author’s framework says “implement the cumulative prospect theory value function v(x) of Tversky & Kahneman (1992) (α = β = 0.88, λ = 2.25) and the Prelec (1998) probability weighting ”. This chapter checks where each piece comes from, points out that the 1992 paper used a different weighting function, and explains what “cumulative” means mathematically.
The value function
Evidence
The 1992 paper uses power functions for value: for gains () and for losses (). The participants were 25 graduate students from Berkeley and Stanford (12 men, 13 women), each paid $25 regardless of their choices. Fitted to each person’s certainty equivalents, the median exponent of the value function was 0.88 for both gains and losses, and the median loss-aversion coefficient was 2.25 (Tversky and Kahneman 1992).
The framework’s three parameters match the paper, with three remarks. First, they are medians of 25 people, not universal constants; Chapter 4 discusses later meta-analyses. Second, because , for every , so “losses weigh 2.25 times as much as gains” is independent of the amount under these parameters; if , the ratio changes with the amount and alone is no longer an interpretable “multiple” (the program in Chapter 4 shows this). Third, the participants’ choices had no real monetary consequences; that was common practice at the time, but it is one of the qualifications later research discusses.
Two probability-weighting functions
Evidence
The 1992 paper’s weighting function is with median estimates for gains and for losses, the gain curve being slightly more curved (Tversky and Kahneman 1992). Prelec (1998) derived a different family from axioms such as “compound invariance”; its one-parameter form is , inverse-S-shaped for , with both fixed point and inflection point at ; he also gives a two-parameter form whose fixed point is not tied to . The paper’s figures and worked example use , an illustrative value; the paper compares empirical crossing points (about 0.30 to 0.39) with (Prelec 1998). Gonzalez and Wu (1999) estimated weighting functions non-parametrically and discussed the psychological meaning of “curvature” and “elevation” parameters (Gonzalez and Wu 1999).
So the framework’s “TK1992 + Prelec” combines two works: the 1992 value function and the 1998 weighting function. The combination is common in applications and a reasonable modelling choice, but it should be labelled as a combination, and the Prelec parameter should be estimated separately or marked as illustrative. The program below compares the two weighting functions and finds by least squares the Prelec parameter closest to the 1992 gain curve.
Calculation
# Value function and probability weighting of cumulative prospect theory.
# Tversky & Kahneman (1992): v(x) = x^a (x >= 0), -lam (-x)^b (x < 0), medians
# a = b = 0.88, lam = 2.25; weighting w(p) = p^g / (p^g + (1 - p)^g)^(1/g),
# g = 0.61 (gains), 0.69 (losses).
# Prelec (1998), a DIFFERENT form: w(p) = exp(-(-ln p)^c); c = 0.65 is the value he uses
# for illustration, not an estimate. We also fit c to TK's gain curve by least squares.
import math
A, B, LAM, GG, GL = 0.88, 0.88, 2.25, 0.61, 0.69
def v(x):
return x ** A if x >= 0 else -LAM * (-x) ** B
def w_tk(p, g):
if p <= 0 or p >= 1:
return float(p >= 1)
return p ** g / (p ** g + (1 - p) ** g) ** (1 / g)
def w_pr(p, c):
return 0.0 if p <= 0 else 1.0 if p >= 1 else math.exp(-(-math.log(p)) ** c)
def fixed_point(f):
lo, hi = 0.01, 0.99 # f(p) - p changes sign once on (0, 1)
for _ in range(60):
mid = (lo + hi) / 2
lo, hi = (mid, hi) if f(mid) > mid else (lo, mid)
return (lo + hi) / 2
grid = [i / 1000 for i in range(1, 1000)]
cfit = min((0.30 + i / 1000 for i in range(701)),
key=lambda c: sum((w_pr(p, c) - w_tk(p, GG)) ** 2 for p in grid))
print(f"v(100) = {v(100):.1f}, v(-100) = {v(-100):.1f};"
f" |v(-x)| / v(x) = {LAM} for every x"
" because a = b")
print(f"Prelec exponent fitted to TK gains (g = 0.61): c = {cfit:.3f}")
print(" p TK gain TK loss Prelec .65 Prelec fit")
for p in (0.001, 0.01, 0.05, 0.1, 0.25, 0.5, 0.75, 0.9, 0.95, 0.99):
print(f" {p:5.3f} {w_tk(p, GG):7.3f} {w_tk(p, GL):7.3f} {w_pr(p, 0.65):10.3f}"
f" {w_pr(p, cfit):10.3f}")
for lab, f in (("TK gain", lambda p: w_tk(p, GG)), ("TK loss", lambda p: w_tk(p, GL)),
("Prelec .65", lambda p: w_pr(p, 0.65))):
print(f" fixed point w(p) = p, {lab:10s}: p = {fixed_point(f):.3f}")
print(f" 1/e = {1 / math.e:.3f}; largest |TK gain - Prelec fit| ="
f" {max(abs(w_tk(p, GG) - w_pr(p, cfit)) for p in grid):.3f}")
print("fourfold pattern: certainty equivalent CE of (x, p; 0) under the TK medians")
print(" p EV gain CE gain EV loss CE loss")
for p in (0.01, 0.05, 0.5, 0.95, 0.99):
ce = []
for x, g in ((100, GG), (-100, GL)):
val = w_tk(p, g) * v(x)
ce.append(val ** (1 / A) if val >= 0 else -((-val / LAM) ** (1 / B)))
print(f" {p:4.2f} {100 * p:8.1f} {ce[0]:8.1f} {-100 * p:8.1f} {ce[1]:8.1f}")
# v(100) = 57.5, v(-100) = -129.5; |v(-x)| / v(x) = 2.25 for every x because a = b
# Prelec exponent fitted to TK gains (g = 0.61): c = 0.501
# p TK gain TK loss Prelec .65 Prelec fit
# 0.001 0.014 0.008 0.030 0.072
# 0.010 0.055 0.040 0.067 0.117
# 0.050 0.132 0.111 0.130 0.177
# 0.100 0.186 0.170 0.179 0.219
# 0.250 0.291 0.294 0.290 0.308
# 0.500 0.421 0.454 0.455 0.435
# 0.750 0.568 0.626 0.641 0.585
# 0.900 0.712 0.775 0.793 0.723
# 0.950 0.793 0.850 0.865 0.798
# 0.990 0.912 0.945 0.951 0.905
# fixed point w(p) = p, TK gain : p = 0.339
# fixed point w(p) = p, TK loss : p = 0.378
# fixed point w(p) = p, Prelec .65: p = 0.368
# 1/e = 0.368; largest |TK gain - Prelec fit| = 0.062
# fourfold pattern: certainty equivalent CE of (x, p; 0) under the TK medians
# p EV gain CE gain EV loss CE loss
# 0.01 1.0 3.7 -1.0 -2.6
# 0.05 5.0 10.0 -5.0 -8.3
# 0.50 50.0 37.4 -50.0 -40.8
# 0.95 95.0 76.9 -95.0 -83.1
# 0.99 99.0 90.0 -99.0 -93.8The results make several points. First, and , a ratio of exactly 2.25. Second, the Prelec parameter closest to the 1992 gain curve is 0.501, not 0.65; the largest gap between the two curves is 0.062, at of about 0.005 to 0.01 (: 0.055 against 0.117). Third, the fixed point of the 1992 function is 0.339 for gains and 0.378 for losses; the one-parameter Prelec function always has 0.368, that is . Fourth, the last table reproduces the “fourfold pattern”: for small-probability gains the certainty equivalent exceeds the expected value (: 3.7 against 1.0, risk seeking), for large-probability gains it falls short (: 76.9 against 95.0, risk aversion); losses go the other way. This is the “fourfold pattern of risk attitudes” summarised in the 1992 paper (Tversky and Kahneman 1992). Figure 3 shows the value function and three weighting curves.
What “cumulative” means
The framework’s wording invites the reading that cumulative prospect theory computes for each outcome’s probability and then . That is the form of the original 1979 theory (without its editing phase), not of the 1992 theory.
Evidence
In cumulative prospect theory, gains are ranked from best to worst, and the -th best gain receives the weight of the probability of “at least this good” minus the weight of the probability of “strictly better”: losses are ranked from worst to best with analogous differences of (Tversky and Kahneman 1992). A weight therefore depends on the outcome’s rank (rank dependence), not only on its own probability.
The distinction is not a technicality. Separable weighting has two known defects: the weights generally do not sum to one; worse, it violates first-order stochastic dominance — an option that is no worse in any state can be valued lower. The 1979 theory excluded such cases by hand in the editing phase; cumulative weights avoid them automatically.
Calculation
# Cumulative (rank-dependent) decision weights versus weighting each probability on its
# own. CPT (Tversky & Kahneman 1992): rank gains from best to worst; the weight of the
# i-th best gain is w+(P(X >= x_i)) - w+(P(X > x_i)); losses likewise from the worst with
# w-. Separable weights pi_i = w(p_i), as in the 1979 theory without its editing phase,
# do not sum to one and can violate first-order stochastic dominance.
import math
A, LAM, GG, GL = 0.88, 2.25, 0.61, 0.69
def w(p, g):
if p <= 0 or p >= 1:
return float(p >= 1)
return p ** g / (p ** g + (1 - p) ** g) ** (1 / g)
def v(x):
return x ** A if x >= 0 else -LAM * (-x) ** A
def cpt(prospect): # prospect: list of (outcome, probability)
gains = sorted([o for o in prospect if o[0] > 0], key=lambda o: -o[0])
losses = sorted([o for o in prospect if o[0] < 0], key=lambda o: o[0])
total, weights = 0.0, []
for group, g in ((gains, GG), (losses, GL)):
cum = 0.0
for x, p in group:
pi = w(cum + p, g) - w(cum, g); cum += p
total += pi * v(x); weights.append((x, pi))
return total, weights
def separable(prospect):
return sum(w(p, GG if x > 0 else GL) * v(x) for x, p in prospect if x != 0)
def ce(val):
return val ** (1 / A) if val >= 0 else -((-val / LAM) ** (1 / A))
L1 = [(100, 0.5), (0, 0.5)]
L2 = [(100, 0.25), (99, 0.25), (0, 0.5)] # dominated by L1
print("L1 = (100, .5; 0, .5) L2 = (100, .25; 99, .25; 0, .5): L1 dominates L2")
for lab, f in (("separable w(p_i)", separable),
("cumulative (CPT)", lambda q: cpt(q)[0])):
a, b = f(L1), f(L2)
print(f" {lab:17s}: CE(L1) = {ce(a):6.2f}, CE(L2) = {ce(b):6.2f} ->"
f" {'prefers L1' if a > b else 'prefers dominated L2'}")
print("splitting 100 into ever more near-equal pieces (separable weights):")
for k in (1, 2, 5, 10, 50):
q = [(100 - 0.01 * i, 0.5 / k) for i in range(k)] + [(0, 0.5)]
print(f" k = {k:2d}: separable CE {ce(separable(q)):7.2f};"
f" CPT CE {ce(cpt(q)[0]):6.2f}; sum of separable weights"
f" {sum(w(p, GG) for x, p in q if x > 0):.3f}")
M = [(200, 0.1), (50, 0.3), (-20, 0.4), (-150, 0.2)]
val, ws = cpt(M)
print("mixed prospect (200, .1; 50, .3; -20, .4; -150, .2), CPT decision weights:")
print(" " + "; ".join(f"x = {x}: {pi:.3f}" for x, pi in ws))
print(f" gain weights sum {sum(pi for x, pi in ws if x > 0):.3f}, loss weights sum"
f" {sum(pi for x, pi in ws if x < 0):.3f}; value {val:.2f}, CE {ce(val):.2f},"
f" EV {sum(x * p for x, p in M):.1f}")
# L1 = (100, .5; 0, .5) L2 = (100, .25; 99, .25; 0, .5): L1 dominates L2
# separable w(p_i) : CE(L1) = 37.38, CE(L2) = 53.73 -> prefers dominated L2
# cumulative (CPT) : CE(L1) = 37.38, CE(L2) = 37.26 -> prefers L1
# splitting 100 into ever more near-equal pieces (separable weights):
# k = 1: separable CE 37.38; CPT CE 37.38; sum of separable weights 0.421
# k = 2: separable CE 54.00; CPT CE 37.38; sum of separable weights 0.581
# k = 5: separable CE 92.24; CPT CE 37.37; sum of separable weights 0.932
# k = 10: separable CE 136.59; CPT CE 37.37; sum of separable weights 1.316
# k = 50: separable CE 316.63; CPT CE 37.32; sum of separable weights 2.763
# mixed prospect (200, .1; 50, .3; -20, .4; -150, .2), CPT decision weights:
# x = 200: 0.186; x = 50: 0.184; x = -150: 0.257; x = -20: 0.261
# gain weights sum 0.370, loss weights sum 0.518; value -30.27, CE -19.18, EV -3.0Lottery L1 pays 100 with probability 1/2; L2 pays 100 with probability 1/4 and 99 with probability 1/4, and 0 otherwise; L1 first-order stochastically dominates L2. Separable weighting gives L2 a certainty equivalent of 53.73, above L1’s 37.38, preferring the dominated lottery; cumulative weights give 37.38 against 37.26 and correctly prefer L1. Splitting 100 into 50 nearly equal small outcomes raises the sum of separable weights to 2.763 and the certainty equivalent to 316.63 — more than the largest possible payoff; cumulative weights stay at about 37.3. The last mixed prospect shows cumulative weights at work: the gain and loss weights sum to 0.370 and 0.518, not to the corresponding probabilities (0.4 and 0.6); this is the effect of probability weighting, not an error.
Note
Be careful with implications for page design. Some infer that “splitting one offer into several small ones is more attractive”. Under separable weights the inference is mathematically valid, but separable weighting is exactly the form that violates dominance and was abandoned in the 1992 theory; in cumulative prospect theory, splitting outcomes of the same probability does not change the evaluation. Whether splitting works is an empirical question about mental accounting and presentation (Thaler 1985 discussed “segregating gains and integrating losses” (Thaler 1985)); it cannot be derived from cumulative prospect theory.
Small-probability losses and add-on protection at checkout
One cell of the fourfold pattern bears directly on checkout pages: small-probability losses. The table in the earlier program shows that at the 1992 median parameters the certainty equivalent of “a 1 % chance of losing 100” is −2.6, against an expected value of −1.0; the model’s decision-maker would pay up to about 2.6 to remove a risk whose expected loss is 1.0. This is the pricing room for add-ons such as extended warranties and shipping insurance: , a small-probability loss inflated about fourfold.
The inference has three qualifications. First, it uses the median parameters of 25 people, and individuals differ widely (Chapter 4). Second, cumulative prospect theory deals with known probabilities; a shopper on a page usually does not know the probability that the product breaks and faces a probability she must estimate herself, which is the problem of judgement and calibration treated in Book 26, different from “weighting a known probability”. Third, “the model’s decision-maker would pay more” does not mean “pre-ticking the add-on is legitimate”: pre-ticking exploits defaults and procrastination (Chapter 6), not probability weighting, and the Chapter 8 checklist flags it separately.
Chapter summary
The value-function parameters of 1992 cumulative prospect theory, and , match the author’s framework, but they are medians from 25 graduate students not paid according to their choices. The 1992 weighting function is (, ); is Prelec’s different 1998 function, 0.65 is an illustrative value, and the parameter closest to the 1992 gain curve is 0.501. Cumulative prospect theory weights cumulative probabilities and takes differences; separable weighting prefers a dominated lottery and lifts the certainty equivalent to 316.63 when one outcome is split 50 ways, cumulative weights do not.
References
Fan, Y. (2026). Choice Architecture, Cumulative Prospect Theory and Computational Behavioural Economics of Pricing Interfaces: From Loss-Aversion Value Functions and Asymmetric Dominance to Hyperbolic Discounting, Read Beside the Qingzhong Chapters of the Guanzi [选择架构、累积前景理论与销售界面价格锚定的计算行为学] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23836260 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).