5D Library › Dynamical Systems of Neurotransmitters and Endocrinology · 中文
Chapter 3 The HPG Axis, Testosterone and the Challenge Hypothesis
Starting with birds
The hypothalamus releases gonadotropin-releasing hormone (GnRH), the pituitary releases luteinising hormone (LH), and LH makes the testes (in women mainly the ovaries and adrenal glands) synthesise testosterone. Testosterone in turn inhibits GnRH and LH: another negative-feedback loop, the hypothalamic–pituitary–gonadal (HPG) axis.
Evidence
In 1990 Wingfield, Hegner, Dufty and Ball proposed the “challenge hypothesis”: in monogamous birds in which males help to raise the young, testosterone during the breeding season usually stays at a fairly low baseline sufficient for reproduction; it rises to higher levels only during social challenges such as territorial intrusions or competition for mates, and falls back afterwards. Sustained high testosterone is costly and interferes with parental care (Wingfield et al. 1990).
The point of the hypothesis is not “the more testosterone the better” but rather the opposite: testosterone is a signal that is turned up and down with the situation, and high levels appear only when needed. Archer applied it to humans and found the data partly consistent: testosterone changes around competitions, and the correlation between baseline testosterone and aggression is weak (Archer 2006).
How large are the human effects?
For testosterone, the gap between popular claims and measured effect sizes is wide.
Evidence
Eisenegger and colleagues gave women testosterone or placebo, double-blind, before an ultimatum bargaining game. Those given testosterone made fairer offers, not more selfish ones; but those who believed they had received testosterone, whatever they had actually received, made less fair offers. The authors’ interpretation was that testosterone may promote status-seeking, which in this setting is served by fair offers, and that the folk belief that testosterone makes people selfish and aggressive itself changes behaviour (Eisenegger et al. 2010). In a later review, Eisenegger, Haushofer and Fehr proposed that testosterone’s effects are better understood as promoting the search for status than as simply promoting aggression (Eisenegger et al. 2011).
Evidence
For the “winner–loser effect”, in which winners’ testosterone rises and losers’ falls, an updated meta-analysis by Geniole and colleagues found that winners do have slightly higher testosterone after a contest than losers, but the effect is small, especially weak in laboratory contests, larger in real sporting competitions, and highly variable between studies (Geniole et al. 2017).
Evidence
The popular “power pose” claim holds that two minutes in an expansive posture raise testosterone and lower cortisol. Ranehill and colleagues repeated the experiment in a larger sample of 200 people and found no effect of posture on testosterone or cortisol (Ranehill et al. 2015).
To see what a “small effect” means, a standardised effect size can be turned into two more concrete numbers.
Calculation
The meaning of an effect size. With a standardised difference between two groups, the chance that a random member of the higher group really is higher than a random member of the lower group is ; the overlap of the two distributions is .
# What small hormone-behaviour effects mean in practice.
import math
from statistics import NormalDist
N = NormalDist()
def from_d(d):
"""Standardised mean difference -> chance that a random member of group 1 scores higher than a
random member of group 2 (common-language effect size), and the overlap of the two distributions."""
cl = N.cdf(d / 2 ** 0.5)
overlap = 2 * N.cdf(-abs(d) / 2)
return cl, overlap
print("d P(random member of higher group is higher) overlap of distributions")
for d in (0.1, 0.2, 0.5, 0.8):
cl, ov = from_d(d)
print(f"{d:4.1f} {100 * cl:5.1f} % {100 * ov:5.1f} %")
r = -0.061 # testosterone x cortisol interaction, Dekkers et al. 2019 meta-analysis
print(f"\nr = {r}: explains {100 * r * r:.2f} % of the variance in status-relevant behaviour")
# People needed to detect such an effect once, two-sided alpha 0.05, 80 % power (Fisher z approximation)
z = 0.5 * math.log((1 + abs(r)) / (1 - abs(r)))
n = ((N.inv_cdf(0.975) + N.inv_cdf(0.80)) / z) ** 2 + 3
print(f"participants needed for 80 % power to detect r = {abs(r)}: {math.ceil(n)}")
# d P(random member of higher group is higher) overlap of distributions
# 0.1 52.8 % 96.0 %
# 0.2 55.6 % 92.0 %
# 0.5 63.8 % 80.3 %
# 0.8 71.4 % 68.9 %
#
# r = -0.061: explains 0.37 % of the variance in status-relevant behaviour
# participants needed for 80 % power to detect r = 0.061: 2108Cortisol inhibits testosterone
The HPA and HPG axes are not independent.
Evidence
Viau reviewed the two-way links between the axes: in animals, stress and glucocorticoids can inhibit the reproductive axis at the level of the hypothalamus, the pituitary and the testis, while sex steroids also regulate the response of the HPA axis to stress (Viau 2002).
From this, Mehta and Josephs proposed the “dual-hormone hypothesis”: testosterone is linked to dominance and status behaviour only in people with low cortisol; when cortisol is high, the link disappears or reverses (Mehta and Josephs 2010).
Evidence
Dekkers and colleagues meta-analysed the dual-hormone hypothesis and found that the testosterone × cortisol interaction on status-relevant behaviour, although statistically significant, was very small (), with signs of publication bias, analytic flexibility and low power in single studies. They concluded that the hypothesis has only “marginal” support and called for a large pre-registered study (Dekkers et al. 2019).
The program above has already shown that explains about 0.4 % of the variance, and that detecting it once with 80 % power takes about 2,100 people. Most single studies are far smaller.
Challenge and threat appraisals
The same stressful situation can be appraised as a challenge or as a threat.
Evidence
Blascovich and Mendes’s biopsychosocial model holds that when people judge their resources sufficient for the demands of a task they are in a “challenge” state: cardiac output rises and peripheral vascular resistance falls; when they judge the demands to exceed their resources they are in a “threat” state: cardiac output changes less and vascular resistance rises (Blascovich and Mendes 2000). Jamieson and colleagues found that guiding people, before an exam or a social stress test, to reinterpret their bodily arousal as helpful to performance moved their cardiovascular responses towards the challenge pattern (Jamieson et al. 2012).
Calculation
A toy model coupling the two axes: during a 60-minute contest, cortisol rises with the appraised demand; testosterone production is boosted in a “winnable challenge” and inhibited by high cortisol. All quantities are in arbitrary units (1 at rest), and the parameters are purely illustrative.
# Toy model of two coupled axes during a 60-minute contest (arbitrary units, rest = 1).
# Cortisol C rises with appraised demand; testosterone T production is boosted by a contest the person
# appraises as a winnable challenge and is inhibited by high cortisol (HPA -> HPG cross-talk, Viau 2002).
# Parameters are illustrative; effect sizes in real people are much smaller and more variable (see text).
def contest(c_drive, t_boost, minutes=240, dt=0.5):
C = T = 1.0
hC, hT, KI = 1 / 60, 1 / 45, 2.0 # clearance per minute; inhibition constant
out, t = {}, 0.0
while t <= minutes:
on = t < 60
dC = hC * ((1 + (c_drive if on else 0)) - C)
# production: boosted during a challenge, inhibited by cortisol (equals 1 at C = 1 without a boost)
prod = (1 + (t_boost if on else 0)) * (1 + (1 / KI) ** 2) / (1 + (C / KI) ** 2)
dT = hT * (prod - T)
C += dt * dC; T += dt * dT
for m in (30, 60, 120, 240):
if abs(t - m) < dt / 2:
out[m] = (C, T)
t += dt
return out
print(f"{'appraisal':32s} min C T T/C")
for name, cd, tb in (("challenge (demands <= resources)", 0.8, 0.4),
("threat (demands > resources)", 2.5, 0.0),
("no contest", 0.0, 0.0)):
o = contest(cd, tb)
for m in (60, 120, 240):
C, T = o[m]
print(f"{name if m == 60 else '':32s} {m:4d}: {C:5.2f} {T:5.2f} {T / C:5.2f}")
# appraisal min C T T/C
# challenge (demands <= resources) 60: 1.50 1.15 0.76
# 120: 1.18 0.95 0.81
# 240: 1.02 0.98 0.95
# threat (demands > resources) 60: 2.57 0.71 0.28
# 120: 1.58 0.68 0.43
# 240: 1.08 0.92 0.85
# no contest 60: 1.00 1.00 1.00
# 120: 1.00 1.00 1.00
# 240: 1.00 1.00 1.00In the model, a challenge appraisal brings a small rise in testosterone and a moderate rise in cortisol, with recovery within two hours; a threat appraisal brings a large rise in cortisol, suppresses testosterone, and leaves it low for some time after the contest. This is a simplified model of the cross-inhibition described by Viau and the two appraisals described by Blascovich, not a prediction of hormone levels in people: the human studies cited above show real effects that are much smaller and far more variable than the model’s curves.
Note
This book does not discuss or recommend any supplementation with testosterone or other hormones, off-prescription use, or “natural boosting” methods. Testosterone therapy is a medical treatment with indications and risks, and only a doctor can decide on it.
Hypothesis · not established
Not established. People preparing an important public presentation who receive five minutes of “arousal reappraisal” instruction (interpreting a racing heart as the body preparing to perform) show a smaller fall in the ratio of salivary testosterone to cortisol from before to after the talk than people given neutral instruction of the same length. If, in a pre-registered randomised trial, the 95 % confidence interval for the difference between groups in the change of the ratio lies entirely within a pre-set equivalence margin of ±0.1 (log units), the hypothesis is refuted.
Summary
Testosterone is a signal turned up and down by the social situation, not a “confidence hormone”. In human studies its effects on behaviour are small and depend on context, and popular accounts exaggerate them. There is physiological cross-inhibition between cortisol and testosterone, but the human evidence for the dual-hormone hypothesis is currently weak. What a person can influence more than the hormone numbers themselves is how the situation is appraised.
References
Fan, Y. (2026). Dynamical Systems of Neurotransmitters and Endocrinology: Dopamine Reward Prediction Error, Serotonergic Homeostasis, HPA-Axis Allostatic Load and the Challenge Hypothesis, Read Beside East Asian Classics of Mind Regulation [神经递质与内分泌的动力系统] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23738782 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).