Cremieux asks in his tweet below (with lots of great references) whether testosterone "radically increases risk-taking, aggravation, the propensity to commit violent acts, and so on." I'm glad he's explaining that testosterone's effects are not so clear cut! But I wish he'd given more consideration to the question of whether male-typical levels of testosterone increase the propensity for such behaviors in males. The answer to that question is "yes."
Cremieux is right that there is no clear evidence that male testosterone levels (within the normal T range) predict levels of sexual or aggressive behavior. It's great to get more eyes on that, and to bust the many myths. But the explanations for differences in behavior within a sex are quite different from those concerning differences between the sexes.
Cremieux primarily relies on research that addresses questions like: does a man's T level predict the intensity of his sex drive or the frequency of his violent behavior? The answer, generally, is no. (And he's right that a bigger body, more muscles or deeper voice, which are potentially associated with higher T at critical developmental stages, do impact behavior indirectly, but not in a clear way.)
But he then uses this āwithin sexā research to answer questions about the source of the gap between the sexes ā in aggression, in sex drive, whatever. But answers to questions about how T predicts behavior in one sex do not address questions about the role of testosterone in explaining differences in behavior between the sexes. Such questionsāe.g. why are males more violent than femalesā require different kinds of evidence, answering questions like what T does across development, why it does it, what happens when you manipulate it in other animals, and what do ānatural experimentsā in humans tell us? It's about the evolution of sexually differentiated reproductive strategies and the hormonal mechanisms that coordinate body and behavior in each sex.
An evolutionary perspective would have improved Cremieux's perspective. He tweeted that "the [sex] gap in violent behavior is far too large to be explained by testosterone levels." But the size of the sex difference is part of the point: 90-95% of all murders are committed by males, pretty much everywhere in the world. (The size of the gap is driven by differences in male behavior, which tends to respond to cultural incentives.) This pattern is not caused by small differences in T level; it is about large and impactful differences between the sexes in testosterone over a lifetime, that exist for a reason.
On average, a human male is exposed to T levels that are about 10-20 times higher than in females. Across development, male brains and bodies go through "critical periods" in early life (in the womb and shortly after birth) and adolescence, during which the brain and body are particularly sensitive to T's actions, which have permanent effects. So even if T is low in adulthood, it has already permanently affected reproductively relevant behaviors.
Evidence suggests that "sex-typical behavior" (with wide variation!) in male animals (including humans) is elicited once T passes something like a relatively low but male-typical threshold T level, at different phases of life. Above that, what seems to matter more is age, health, history, the ecological environment, and social cues and norms. In humans, these can include things like religion; socioeconomic status; availability of fertile females; disease; threats to resources, mates or children; community values; standards of masculinity; and the law.
Higher lifetime T in males (and not necessarily current T) is why many animals show similar sex differences in aggression, including higher rates of rough play. This is true even in humans, in which boys play rougher than girls around the world, even though little boys don't have higher T than girls. The male preference for rough play has been clearly shown to be driven by higher T, specifically in the womb and/or shortly after birth in male animals. Such play is part of a male reproductive strategy, motivating young males to practice status-competition and cooperation skills they will need in adulthood. In girls, those with unusually high T levels in the womb also tend to prefer rougher play.
Testosterone has these effects in males because of evolutionary pressures: males produce sperm rather than eggs, and thus have evolved reproductive strategies that differ from those of females. On average, males have a higher potential rate of reproduction and invest less in parenting, and generally (not always) compete more intensely for mates. Testosterone is the biological mechanism that coordinates the production of sperm with the other physical and behavioral attributes that help males get sperm to egg, given the species and environment.
Again, environment matters tremendously, but it is not the source of the sex differences in basic reproductive behaviors.
You can read more about testosterone and behavior in my book T: The Story of Testosterone, the Hormone that Dominates and Divides us.