Daniel Voyer, Ph.D.,

Daniel Voyer Ph.D.

Perceptual Asymmetries

What Do Your Fingers Say About Your Prenatal Life?

Can you tell how much androgens you saw as a foetus from your fingers?

Posted Nov 07, 2014

When I started this blog, I thought that this would be a good place to air some of my pet peeves. This post is about one of them. For a number of years, I have seen papers using the ratio of the second digit to the fourth digit (2D:4D ratio) in humans as a measure of prenatal exposure to testosterone (T), with the idea that a smaller ratio means more prenatal T. Of course, that ratio tends to be smaller in males.

Many authors have taken this association for granted since the notion was presented by Manning et al. (2003), to the point that some seem to view this measure as a direct reflection of prenatal  T, when it is, at best, an indirect measure. In any case, since then, researchers have shown a correlation between the 2D: 4D ratio and a plethora of measures and traits ranging from mental rotation skills to empathizing/systemizing (see my post on “Sexing the autistic brain”). Such results have then been interpreted as reflecting a prenatal T connection. For example, one such conclusion might be that homosexuals tend to have a higher 2D: 4D ratio than heterosexual individuals, therefore, low prenatal T contributes to homosexuality (e.g., Putz, Gaulin, Sporter, & McBurney, 2004).

Since I first encountered it, I have had issues with this kind of reasoning. From my perspective, the reasoning seems to be as follows: one finds a correlation between prenatal T (let’s call it variable A to simplify discussion) and finds that it correlates with the 2D:4D ratio (variable B). Thus, we have a correlation between A and B. Then, we find a correlation between variable B and something else (variable C), and we make the inference that this reflects a correlation between A and C, even though we never measured A. This is problematic for a number of reasons. In particular, this claim ignores the possible influence of extraneous variables in this relation and the correlation could be spurious (i.e., due to another variable). For example, the 2D:4D ratio is also related to sex. However, there is much more that defines “sex” than hormones. In fact, sex differences in 2D: 4D ratio might just be arising from sex differences in overall hand size.  I do not believe that this has ever been verified, although it seems that men do tend to have bigger hands than women (Peters, Mackenzie, & Bryden, 2002). In addition, there are many other in-utero influences that can affect various physical and psychological traits later in life. Prenatal diet, amount of exercise, drug consumption, and stress level experienced by the mother are only a few examples.

A better way to illustrate the conceptual issue at work here is to consider the fact that some researchers have reported a link between handedness and birth stress (higher stress is related to left handedness). From this finding, would one ever claim that a study investigating the relation between handedness and mental rotation uses handedness as a proxy for birth stress? Of course not!  Therefore, in a similar vein, it is no more possible to state that we can use the 2D:4D ratio as a proxy for T levels, based on a correlation between prenatal T levels and the 2D:4D ratio.  This is compounded by the fact that the base correlation between androgen and the digit ratio is often small. For example, Manning et al. (2003) reported a correlation of .29 between CAG number (essentially reflecting androgen responsivity) and the ratio, and only for the right hand (the correlation was only .007 for the left hand!). The right hand correlation would only account for 8.8% of variance, leaving much room for other variables, as mentioned earlier.

I have used this kind of arguments for years without getting anywhere with authors or editors. It had been driving me nuts! Then, I came across the meta-analysis published by Voracek (2014). When I found this paper, it made my day! This author conducted a comprehensive meta-analysis of studies examining correlations between variants of the androgen receptor gene and the 2D: 4D ratio.  Considering these gene variants along with the right hand ratio, left hand ratio, and left-right difference, he found what he called a null association, based on an overall correlation of .045 for the right hand, .034 for the left hand, and .019 for the difference.

Essentially, there is no such connection as the one hypothesized between prenatal T and the 2D: 4D ratio. Table 1 of the Voracek article is quite revealing as it shows that only two of the studies produced a significant correlation, with the Manning et al. study mentioned earlier producing the largest correlation ever observed! Essentially, we can conclude that the 2D: 4D ratio might reflect something, but that thing is definitely not the prenatal T levels. This casts doubt on all that research that used the ratio as a surrogate measure for prenatal T and the observed correlation between the 2D: 4D ratio and whatever else they could think of. Voracek has much to say (including noteworthy criticisms of a similar meta-analysis published by Hönekopp, 2013), and I definitely recommend it as an important paper to read. The most puzzling conclusion that I draw from the Voracek paper is that the original Manning et al. findings were not replicable, yet everyone who has cited their research to support the use of the 2D: 4D ratio as a proxy for prenatal T interpreted these results as a solidly established finding. In fact, Voracek also reports evidence of a citation bias in favor of the studies showing large effects compared to non-significant ones.  I guess that this is a warning to all of us that we should always remain critical, ensure replication of research findings, and examine the literature thoroughly before we take a finding for granted.

All in all, it seems that I was correct in my view that reported associations of cognitive or psychological variables with the 2D: 4D ratio are due to a third variable, not to prenatal T.  It is so nice to be right once in a while! Thank you Martin Voracek!

In conclusion, going back to the question we started with: What do your fingers say about your prenatal life? The answer is: Nothing as far as T levels are concerned!


Hönekopp, J. (2013). No evidence that 2D:4D is related to the number of CAG repeats in the androgen receptor gene. Frontiers in Endocrinology, 4, 185.

Manning, J. T., Bundred, P. E., Newton, D. J., & Flanagan, B. F. (2003). The second to fourth digit ratio and variation in the androgen receptor gene. Evolution and Human Behavior, 24, 399–405.

Peters, M., Mackenzie, K., & Bryden, P. (2002). Finger length and distal finger extent patterns in humans. American Journal of Physical Anthropology, 117, 209-217.

Putz, D. A., Gaulin, S. J. C., Sporter, R. J., & McBurney, D. H. (2004). Sex hormones and finger length: What does 2D:4D indicate? Evolution and Human Behavior, 25, 182–199.

Voracek, M. (2014). No effects of androgen receptor gene CAG and GGC repeat polymorphisms on digit ratio (2D:4D): a comprehensive meta-analysis and critical evaluation of research. Evolution and Human Behavior, 35, 430–437