Skeletons and Unconscious Bias

As scientists we like to believe that we seek and interpret evidence impartially. That has been the accepted position for generations. The reality is of course that we are sometimes influenced, unconsciously or otherwise, by received opinion, ‘experts’ or other cultural factors. In a few cases, which these days are frequently well-documented after the fact, substantial cherry-picking of data and evidence may occur (read Ben Goldacre if you doubt that statement). These thoughts are prompted by reading the book by Londa Schiebinger The Mind has No Sex? (which she kindly gave me after we met at the meeting I described previously in Austria), and specifically the chapter on the changing views of male and female bodies, during the 18th century as anatomy developed. This chapter appears as a stand-alone paper here, in at least very similar form to the book (I haven’t made a detailed comparison).

Before the early 18th century, anatomists had never thought specifically about differences between male and women’s bony structures, all the emphasis on differences had been associated simply with the sexual organs. However, as cultural attitudes moved towards stressing differences between men and women, rather than similarities, further attention became focussed on the whole body and not just the genitalia . The womb was no longer seen as an imperfect penis; on the contrary it was regarded as a perfect vessel to fulfil woman’s manifest destiny of Mother. But how, or indeed whether the skeleton differed between the sexes had not been explored. The first serious studies and drawings were produced under the direction of Marie Thiroux d’Arconville (interestingly herself one of the rare women who were active in scientific studies at this time in France, having studied at the Jardin du Roi) and published in her French translation of Monro’s Anatomy. This drawing of the female figure is shown below, in a later version appearing in Barclay’s The Anatomy of the Human Body (the appearance of the ostrich in this figure will be explained later), along with its male companion. One has to assume that Thiroux d’Arconville took her anatomy seriously. Nevertheless there are two extremely unrealistic features of this drawing: the head is far too small in relation to the body, and the rib-cage is unnaturally narrow thereby accentuating the pelvis – again stressing the motherhood motif. It is possible that this second ‘inaccuracy’ could actually merely have meant that the poor cadaver she chose to draw had spent a life bound in corsets, which were already known to distort the body.

The next female skeleton to be drawn was by Samuel Thomas von Soemmering. He prided himself on the accuracy of his drawing, but nevertheless one has to ask how ‘typical’ his woman was. She was, apparently, carefully selected but nevertheless Soemmering chose to draw the skull of someone else atop the body and for posture and proportions he made comparison with classical beauties such as the Venus de Medici and the Venus of Dresden, both facts which might be thought to affect the accuracy of his rendition. However, he made a better job of both head size (a woman’s head is larger in proportion to the body than a man’s) and the width of the rib-cage than d’Arconville. As a consequence he came in for subsequent criticism because, it was proposed that women’s rib cages are necessarily smaller than men’s because their ‘restricted life style requires that they breathe less vigorously’(according to the Scottish anatomist John Barclay). Recognition of the relatively larger size of the head in women also posed problems for subsequent generations who wanted to believe size of head/brain equated to intelligence (the ‘explanation’ that was advanced in order to maintain the position of man’s superiority was that women’s large skulls indicated incomplete growth, reasoning that seems distinctly flawed on many counts).

Female (left) and male skeletons taken from Barclay's The Anatomy of the Bones of the Human Body, 1829

Now back to that ostrich in the figure. Cultural subliminal messages seemed to be part of the artistic endeavour at this time; this was certainly something that was true of all the portraiture of the period where man’s accoutrements in the paintings were deliberately chosen as status symbols to convey messages of power and wealth. So it is not surprising that subsequent reproductions of D’Arconville’s female skeleton in the book by Barclay had, for comparison, an animal known for its long elegant neck and the large size of its pelvis. These were attributes deliberately intended to be stressed as relevant to feminine beauty and good child-bearing abilities. The equivalent male figure in this same book by Barclay was based on a drawing by Albinus and the comparator animal was selected to be a horse, remarkable for its strength and agility – though interestingly not intelligence.

So, what can we learn from this tale? Firstly, that despite attempts at objectivity, scientists may make choices in the way they portray their ‘data’, something that is as likely to be true now as then in different ways. For instance, as a microscopist I am always faced with the challenge of choosing a ‘typical’ micrograph to include in a paper. Which one is chosen might be influenced by many things including peripheral things such as how central the feature of interest is in the field of view, whether the contrast was optimised in one image more than another, and whether that annoying piece of dust on the lens left by someone else’s student got in the way. None of that has anything to do with whether the image is in truth typical. One will probably also be influenced by what one was looking for, what the hypothesis under consideration predicts to occur, so that it is too easy to fail to see other things. (As an aside, I have never forgotten the comment made to me after a seminar I gave, by a biologist working on amyloid fibrils who, having heard me talk about spherulites as suprafibrillar aggregates remarked ‘oh maybe that’s the crud at the bottom of the test-tube!’. In other words, they had been looking for fibrils, had been using transmission electron microscopy to study the fibrils, and the ‘crud’ was too large to be investigated in the TEM and so had been ignored.) However, finding a ‘typical’ skeleton is always going to be a challenge and, given the labour intensive work of drawing (and measuring) during the 18th century, it is hardly surprising drawn examples were so limited and hence potentially so atypical.

Secondly, back then as now we see that stereotyping occurred. If it was believed that women had small heads and rib-cages it would be perfectly possible to demonstrate that legitimately by finding and accurately drawing a particular figure. Whether D’Arconville deliberately chose a skeleton whose bones had been compressed by years of corset-wearing we will never know, but clearly intentionally or not, she perpetuated a myth about the shape of women for many years. Cultural norms may well have affected what she was looking for and what she felt ‘should’ be represented.

Thirdly, then as unfortunately too often now, people were ‘male by default’ so that thinking about the female skeleton at all came comparatively late in the days of anatomy. This was a point Londa discussed at length at the Vienna meeting at which I met her, but there are many familiar examples of it around us. Drug trials carried out, certainly in the recent past, on male participants only is a well-known example which may have disastrous consequences when women subsequently take the drugs; or psychological assessments done on males which are then extrapolated to females who are found ‘wanting’ because they do not fit the same mould, a topic discussed with great clarity in the classic book by Carol Gilligan In a Different Voice. We need to be on our guard for such assumptions, which of course underlie unconscious bias in all its manifestations.

So this tale may reinforce the message found in the common paraphrase of Hegel, ‘history teaches us that history teaches us nothing’. We are still capable of making the same mistakes in our approaches to science and the wider cultural embedding of science as this accounts identifies from the 18th century.