Mark Brosnan

Digit Ratio

The relative length of the second (index) finger when compared to the fourth (ring) finger is a proxy measure of exposure to prenatal testosterone and oestrogen.

Using the figure on the right, the length of finger 1 is divided by the length of finger 2 to give the digit ratio for the right hand.

Similarly, the length of digit 4 is divided by the length of digit 3 to give the digit ratio of the left hand.

Current projects:

Relationship to academic experience:

Digit ratio as an index of mathematical ability (SATs)

Digit ratio as a predictor of academic faculty membership

Digit ratio, mental rotation and systemising

Digit ratio, empathising and the ASQ

Digit ratio and the perception of pain

With Ed Keogh

Digit ratio and the awakening cortisol response

With Julie Turner-Cobb, David Jessop, University of Bristol

Indexing Science: Report on digit ratios of academic staff

1) Summary of findings

2) Introduction

3) Study 1

4) Study 2

5) Discussion:

6) References: Please note most references are available on line.

1) Summary of findings

Digit ratio is significantly different between:

Members of the Science Faculty and members of the HaSS/Management Faculty

Those with children and those without children

Those with children who have a psychological diagnosis in the family (typically Dyslexia) and those with children who do not have a psychological diagnosis in the family.

Digit ratio is NOT significantly different between:

Males and females

Lecturers, Senior Lecturers, Readers and Professors

Left and right handers

2) Introduction

To enable me to provide a summary before Christmas, the introduction is very brief. It will be written up in full and presented on Tuesday April 20th 5:15 in 3East 3.5 – all welcome (do check details on http://www.bath.ac.uk/psychology/about/seminarlist.cfm nearer the time). Meanwhile most references are available on line for the interested reader.

Finger length and testosterone:

The digits under investigation are the second digit (2D – next to the thumb) and the fourth digit (4D – next to the little finger). 2D divided by 4D (2D: 4D ratio) provides an index of exposure to testosterone while in the womb. The HOX gene family is required for the growth and patterning of digits and the differentiation of the genital bud. Hoxd and Hoxa genes are strongly expressed in the gonads and are also required for the growth and differentiation of digits. This sharing of causal factors in digit and gonad differentiation allows patterns of digit formation to be a marker for prenatal sex hormone concentration (Manning et al., 1998). Digit length is fixed in utero and relative digit lengths remains constant through development and is constant across ethnicities (Lippa, 2003). Prenatal gonadal hormones exert long-lasting organizational influences on brain and behaviour in humans (Collaer & Hines, 1995). For example, a longitudinal study of the testosterone levels of pregnant women found levels to be predictive of the gender-role behaviour of their female children at 3.5 years of age (Hines et al., 2002).

Finger length and sex differences:

Low values of 2D: 4D are associated with high concentrations of testosterone. The digit ratio is sexually dimorphic with males having a mean of 0.98, i.e. 4D longer than 2D and females having a mean of 1.00, i.e. equal length of 2D and 4D. This pattern is set before the age of 2 and remains constant over the life span (Manning et al., 1998). In males the 2D: 4D ratio is also related to sperm count, a lower 2D: 4D ratio (indicating higher exposure to testosterone relates to higher sperm count), though the impact of this upon fertility requires further research (Manning et al., 1998).

Finger length and ability:

Digit ratio is an index of exposure to prenatal testosterone. Prenatal testosterone slows the growth rate of the left side of the brain while enhancing growth of the right side (Geshwind and Galaburda, 1985). The right hemisphere is associated with better visual-spatial and mathematical abilities, as is the 2D: 4D ratio (Manning and Taylor, 2001). Handedness may also be affected (Manning et al., 2000). Thus, traditional sex differences in visual-spatial and mathematical abilities can be attributed to differences in exposure to prenatal testosterone, indexed by a sex dimorphic pattern in digit ratio.

Additionally, it would appear that the 2D: 4D ratio is a marker for within-sex variance in visual-spatial ability (Sanders et al., 2002).

These abilities underpin what Baron-Cohen (2003) terms ‘folk physics’ or more recently ‘systematising’, that is an intuitive everyday understanding of how things work: The understanding of the properties of physical objects, including their causal impact on other objects. Folk physics/ systematising is different from academic physics as it is found in children prior to formal education. Folk physics is independent from what Baron-Cohen terms ‘folk psychology’ or more recently ‘empathising’, that is an intuitive everyday understanding of how people work: The understanding of how actions are caused by mental/ intentional states of others. Both folk physics/ systematising and folk psychology/ empathising are naturally developing forms of knowledge not explicitly taught and acquired universally. Thus Baron-Cohen conceives of a continuum from folk psychology/ systematising through to folk physics/ empathising upon which females are more proficient in folk psychology/ empathising and males are more proficient in folk physics/ systematising. This underpins the male domination of academic disciplines, such as Mathematics, Computer Science and Physics that develop the principles underlying folk physics/ systematising (although focussing on such folk physics/ systematising skills could be as a result of a deficit in folk psychology/ empathising, Baron-Cohen et al. 1999b).

Finger length and ability and sex differences: An interaction

There is evidence that the optimum level of testosterone for the performance of spatial tasks is in the lower male range (Kimura, 1999). There are cyclical changes in levels of testosterone. As the levels of testosterone decrease in males, performance upon visual-spatial measures increase whereas comparable cyclical declines in female testosterone result in decreased performance upon visual spatial measures (Moffat and Hampson, 1996; see also Sanders et al., 2002). Thus those with the greatest visualisation skills will be females with the lower male-typical digit ratio (0.98) and males with the higher female-typical digit ratio (1.0). Additionally, an extreme digit ratio (0.94) has been associated with Autism, suggesting that excessive prenatal testosterone may play a causal role in the development of autism (Manning et al., 2001). Using some of the visual-spatial measures described above, Dyslexia has been cast as the opposite end of the continuum to Autism (Brosnan et al., 2002). There is a strong heritability to dyslexia and we would predict the opposite finger pattern in the families of those with Dyslexia to the patterns found in families with Autism (i.e. 2D: 4D>= 1).

From, the above, we would predict that those from the Science Faculty with a better understanding of folk physics will have better visualisation skills, reflecting optimal right hemisphere ability resultant from exposure to testosterone typical of the lower normal male range, indexed by digit ratio –when compared to those from other faculties. For the purposes of comparison, the School of Management was combined with the Faculty of Humanities and Social Sciences (HaSS), which is consistent with previous research (Baron-Cohen et al., 2001). A second study will further examine, fertility, psychological conditions and handedness.

3) Study 1

107 respondents from all Departments and all academic grades responded to a request to complete an Internet-based questionnaire asking for basic demographic information and a self-assessment of 2D and 4D digit length of the right and left hand. There are around 500 academics working at the University making a response rate of 21.4%. 22.4% of respondents were female, which accurately reflects the proportion of female academics in the University (20%). Ages ranged from 23 to 62 with a mean of 44 years (standard deviation = 10 years, indicating most participants were between 34 and 54 years of age). A breakdown of respondents appears in Table 1 below:

Table 1: Breakdown of respondents by sex faculty and scale.

SEX	Number
Male	83
Female	24

FACULTY
Engineering	31
HASS/Man	33
Science	43

SCALE
Lecturer	46
Senior Lecturer	27
Reader	11
Professor	23

Digit ratio was calculated by dividing the length of the second digit (2D) by the length of the fourth digit (4D) for the left and right hand separately and then calculating the mean of these two ratios. The digit ratios for each of these subgroups (e.g. a male, engineer, lecturer) appear in Appendix A below.

Validation of self-report measure:

18 respondents forwarded a photocopy of their hands that was measured using callipers, digitally recoding distances to 2 decimal places. Each measurement was undertaken twice. The correlation between the self-measurement and the measurement of the photocopy was extremely significant indicating good inter-rater reliability (r=0.819, p<0.001).

Findings:

An ANOVA was conducted to statistically identify whether there were differences between different groups (sex, faculty or pay scale) that related to differences in digit ratio. The ANOVA is reported in Appendix 2. This shows that there was a significant difference between the faculties in digit ratio (F=3.519, p=0.034). A Tukey HSD post hoc analysis revealed that the significant difference was between the Science faculty and the HASS/Management faculty (p=0.02), with the Engineering faculty averaging between the other two faculties. This is illustrated in the Graph below (rounded to 2 decimal places).

To three decimal places, the mean ratio for the Science faculty was 0.996, HaSS/Management=0.977 and for Engineering=0.984.

Sex was not a significant factor (mean for males = 0.987, mean for females=0.984); Pay scale was not a significant factor (Means range from 0.986 to 0.989) and there were no significant interactions.

The primary hypothesis that there would be differences in digit ratio between members of different faculties has been supported. Interestingly, the Science faculty demonstrates a finger ratio similar to that of females within the normal population. The HaSS/Management faculty demonstrate a finger ratio similar to that of males within the normal population.

Obviously this is a specific sample but it is interesting that no sex differences emerged within this population as has been reported for less specific samples. Taking the female mean, this is much closer to the male norm reported for less specific samples, suggesting that digit ratio (and by implication exposure to prenatal testosterone) is not sex dismorphic within an academic population.

Post hoc analysis by Department:

The mean digit ratio is 0.986. An examination of those Departments with means above 0.995 and below 0.980 (to exclude those around the mean) reveals the following 2 groups of Departments:

Group 1: Above 0.995: (n=26, 81% males)

Chemical Engineering, Chemistry, Computer Science, Mathematical Science, Physics.

Group 2: Below 0.98: (n=38, 66% males)

Architecture, Economics, Education, Management, Psychology, Sports Science, SPS.

Obviously constructing headings for these two groups is going to be a crude approximation, not least because of the overlaps in skills between academics from different Departments. Given this proviso, there does seem to be a more ‘folk physics/ systematising’ orientation to the first grouping and a more ‘folk psychology/ empathising’ orientation to the latter grouping as described in the Introduction. There are not significantly more males in the former group (chi=1.12, p=0.29) as would be predicted by the Baron-Cohen hypothesis but this is a specific sample.

4) Study 2:

As participants submitted their Internet data they were asked if they would undertake an additional questionnaire. The questionnaire assessed academic history, whether the person was a biological parent and whether there were any psychologically diagnosed conditions within this immediate (genetically-related) family. Finally handedness was assessed. 49 participants (46% of the original sample) completed this questionnaire.

The examination of degrees indicated that the vast majority of academics were working within the faculty that they studied in. 84.4% of academics had a first degree in a discipline of their present faculty, 80% had a masters degree within a discipline of their present faculty and 92.5% had a PhD within a discipline of their faculty.

The questionnaire asked whether or not each participant was a biological parent. Age was a potential covariate here and was included as such: 17 participants reported that they were not a parent (finger ratio mean=0.995) and 32 reported that they believed themselves to be a biological parent (mean finger ratio = 0.983). This difference is statistically significant (F=4.968, p=0.031). Those with children have a significantly lower digit ratio to those without children (with age as a covariate). Existing theory predicts this to be the case for males. Running the analysis for males only, the result is very close to retaining significance (7 with no children, mean digit ratio=0.996; 27 with children, mean digit ratio = 0.982; F=3.889, p=0.058). There were fewer females but the relationship is nowhere near significance for females alone (F=0.341, p=0.570).

Of these 49, the next question asked if any (genetically-related) immediate family members had any psychological conditions. 6 replied yes (typically Dyslexia) with a mean digit ratio of 1.004, 43 replied no with a mean digit ratio of 0.985. Again, this difference is statistically significant (t=2.711, p=0.018: Equal variances not assumed) and retained significance when examining males alone (yes response, n=4, digit ratio = 0.999; no response n=30, digit ratio=0.983: t=2.24, p=0.039). The trend was the same for females but not significantly so, maybe as a result of the smaller numbers (yes response, n=2, digit ratio = 1.014; no response n=13, digit ratio=0.989: t=1.38, p=0.192).

Finally handedness was assessed using a 36-item questionnaire on which participants identified whether they always, usually or equally used their right or left hand. This measure of handedness was not a significant correlate of digit ratio(r=0.088, p=0.552).

5) Discussion:

Digit ratio is significantly different between members of the Science Faculty and members of the HaSS/Management Faculty

This is consistent with prediction based upon the assumptions:

a) Science requires enhanced visualisation skills over HaSS/Management

b) Prenatal testosterone plays a causal role in the development of these skills

c) The lower male-typical range of testosterone reflects the optimal level for the development of these skills

d) Prenatal testosterone is indexed by digit ratio

Accepting these assumptions suggests that our academic faculty destiny was determined (in part at least) in utero. Visual-spatial abilities were not assessed in this study.

Digit ratio is significantly different between those with children and those without children

Digit ratio has been linked with higher sperm counts but not fertility. The data does suggest that men with lower 2D:4D ratios (more prenatal testosterone) are more significantly likely to have children than those with higher 2D:4D ratios. The study failed to utilise sexuality as variable as current research is conflicting (Lippa, 2003; Robinson and Manning, 2000).

Digit ratio is significantly different between those with children who have a psychological diagnosis in the family (typically Dyslexia) and those with children who do not have a psychological diagnosis in the family.

Whilst this is consistent with Dyslexia reflecting the opposite pattern to autism (the only disorder in which finger length has been investigated), the finger ratio of those with familial associations with dyslexia (1.004) was not that dissimilar from those in the Science Faculty (0.996) associated with good visualisation skills. The digit ratios of those associated with dyslexia were not beyond the normal range for this sample. As noted above, however, the norm for the males from the Science faculty is similar to the female norm from a larger more diverse population. An interesting follow up would be to examine whether children with Dyslexia had fathers associated with Science/Mathematics (reflecting Baron-Cohen’s work on Autism).

Digit ratio is NOT significantly different between males and females.

It is interesting that the females in academia averaged a digit ratio (0.984) similar to the male norm reported elsewhere for larger, more diverse populations (0.98), suggesting perhaps that there is a link between prenatal testosterone and academic achievement in women. Possibly as a consequence of this, digit ratio is not sex dimorphic when examining an academic sample.

Academic level was also not significant (Lecturers, Senior Lecturers, Readers and Professors), suggesting that testosterone levels do not predict achievement within the chosen discipline (this was also the case when controlling for age). Also a failure to replicate differences between left and right handers may be reflective of the non-linear relationship between testosterone and literalised abilities such as visual-spatial.

The next step:

If you would be willing to complete a Systematiser/ Empthiser questionnaire, please click on previous web page (Bath Uni only).

A larger student-based study is under way examining visual-spatial abilities in students from the three faculties in addition to the variables described above. This research is supported by the British Academy. Thanks are due to Adrian Scott for his work on the project.

6) References

Currently being updated:

2 key papers:

The sytematising/ empathising Questionnaire (WORD file)