At Our Wits' End, page 19
In 2012, Woodley of Menie discovered a published survey of historical reaction time data.[6] This demonstrated something very striking. There had apparently been a big slowing of sRTs from the time of Sir Francis Galton in the late 19th century until the late 20th century. It should be stressed that the instruments used to measure reactions times in the 1880s are accepted as having been perfectly adequate for the job. They had adequate temporal resolution. These data carried the strong implication that there had been a rapid and substantial decline in g over the past hundred years. This initial finding was improved with the addition of extra data and a more sophisticated analysis, and was published in the leading journal Intelligence.[7] It was then replicated and confirmed.[8] This replication study furthermore found, using sRTs, that the decline in g had been around 1 IQ point per decade between 1885 and the year 2004. That is about 10 points, in a century—and probably more over the past two hundred years. Dutton and Charlton have explained that, to put this in perspective, 15 points would be approximately the difference in average IQ between a low level security guard (85) and a police constable (100), or between a high school science teacher (115) and a biology professor at an elite university (130).[9] In other words, in terms of intelligence, the average Englishman from about 1850 would be in roughly the top 15% of the population in the year 2000—and the difference would be even larger if we extrapolated back further towards about 1800 when the Industrial Revolution began to initiate massive demographic changes in the British population.
Woodley of Menie’s numbers, note Dutton and Charlton, are not intended to be precise but they are certainly a rough guide to what is happening. Dutton and Charlton put it very trenchantly in summarising what these numbers mean. They mean that in everyday terms:
‘the academics of the year 2000 were the school teachers of 1900, the school teachers of the year 2000 would have been the factory workers (the average people) of 1900, the office workers and policemen of the year 2000 were the farm labourers of 1900, those who were around 10 to 15 IQ points below average at that time. The low-level security guards and shop assistants of the year 2000 were probably in the workhouse, on the streets, or dead in 1900. The substantial long-term unemployed or unemployable, the dependent ‘underclass’ of the year 2000, simply didn’t exist in 1900. And even this estimate is ignoring the expansion of education since 1900, which expanded the middle class occupations and would, in itself, reduce the average intelligence of academics and teachers and even shop assistants in 2000 compared to what they would have been in 1900.’[10]
What this means, in terms of everyday life, should be quite clear by now from our table setting out the correlates of intelligence. And new studies are replicating the finding. A study in Sweden, for example, based on a sample of more than 7,000 people found that simple audio reaction times had slowed by between 3 and 16 milliseconds between 1959 and 1985.[11]
Colour Discrimination
But reaction times are not the only objective measure that correlates with general intelligence which is showing a decline. Colour discrimination—the ability to distinguish between ever more subtle shades of colour—is declining as well. Now, we might think, ‘What on earth has colour discrimination got to do with g?’ But, in fact, it has a great deal to do with it. As we have seen, Spearman showed that pitch discrimination strongly correlated with other examples of academic performance and, by implication, with general intelligence. Spearman also found that teacher ratings of pupil ability correlated with the ability of these pupils to judge subtle differences in lightness and weight. As such, Spearman posited a ‘general factor of discriminative ability’,[12] which he expected to correlate very strongly with general intelligence. Sensory discrimination correlates with g because the more acute your sensory discrimination is, the better able you are to notice subtle differences between physical quantities which can help you to more efficiently solve problems. This is why more recent studies have found the correlation between general intelligence and general discriminative ability to be as high as 0.92 in some cases.[13] It follows that the more intelligent people are, the better they will be able to discriminate among increasingly subtle differences in colour.
A 2015 study found four standardisation studies conducted between the 1980s and 2000s employing the Farnsworth-Munsell 100-Hue Colour Perception Test, which was developed in 1943. This test was first found to correlate with IQ in the 1960s. The participants have to physically arrange a series of 85 caps, each of a very subtly different hue, along a spectrum defined by two clear end caps such as blue and green or pink and purple. Participants are awarded an error penalty for each cap that is in the wrong order on the spectrum, allowing a quantification of their colour discrimination abilities. It was found that, across the four studies, colour discrimination ability had significantly declined, at the equivalent rate of 3.15 IQ points per decade—even after controlling for the IQ of the countries from which the participants were drawn (Belgium, Finland, UK, and US) and also participant age. Clearly, this is what we would expect if our hypothesis is correct.[14]
Use of High-Difficulty Words
The more intelligent people are, the larger are their vocabularies and the more likely they are to use difficult-to-learn words. Indeed, this seems to be fairly clear to people. We know, implicitly, that intelligent people are more prone to using ‘big words’ or at least unusual words. This fits in with the association between g and colour discrimination. The more intelligent people are, the more able they are to perceive subtle differences; differences which require slightly different words in order to accurately encapsulate them. This use of the technically correct word by the highly intelligent can be distinguished from intellectual poseurs who try to showcase their supposed intelligence by using high-order words quite unnecessarily in order to intellectually intimidate people and to attempt to appear cleverer than they are. But, that aside, part of the linguistic intelligence dimension of the IQ test is vocabulary and the ability to understand subtle differences in the meaning of words. Scores on vocabulary tests are very highly g-loaded and highly heritable.[15]
To test what was happening with vocabulary, a study in 2015[16] examined historical changes in the frequency with which words from the highly g-loaded WORDSUM test were employed across 5.9 million texts published between 1850 and 2005. They also examined the association between WORDSUM scores and completed fertility; how many children you have had by middle age when, typically, you don’t have any more. They found that words with higher difficulties (those that are harder to learn and use correctly) and also stronger negative correlations between pass rates and completed fertility declined in usage over time. By contrast, less difficult words and less strongly selected words increased in use over time—an effect that was predicted by rising literacy. This finding would be consistent with the Flynn Effect stemming, in part, from the vocabulary enriching effects of increases in education level, which would be part of Flynn’s ‘scientific spectacles’ model. These findings persisted when explicitly controlled for word age, and other confounding factors.
More recently, another study has taken this further in a way that clearly illustrates the accuracy of our model. Google’s Ngram Viewer (a truly massive text archive of scanned books, newspapers, scientific journals, and other printed materials) includes texts that go all the way back to the 16th century. So, drawing upon this, Woodley of Menie and colleagues analysed changes in the use of the four very high difficulty WORDSUM words over time between the 16th century and the modern day. They found that the use of these words increases from the 16th century up until the early 19th century and then goes into decline, as can be observed in Figure 7. This is precisely what our model of the rise and fall of Western intelligence would predict, if indeed the usage patterns of these words among those who contribute to literature really does reflect their underlying level of g.[17]
Figure 7. The usage frequency trend for a common factor of the four hardest WORDSUM words between 1600 and 2005, fitted to a third-order polynomial curve.[18]
This system also allows us to estimate roughly how intelligent we are compared to people in the past. Based on the usage frequencies of these WORDSUM words, we currently have about the same level of g as people in the mid-18th century; a generation or two before the Industrial Revolution. It must be remembered that WORDSUM is likely to be subject to the Flynn Effect (recall that the usage of easy words is actually increasing), which may make our vocabulary level artificially high in relation to our underlying general intelligence. So, putting these influences aside, it is likely that we have regressed considerably further, genetically, than the mid-18th century.
Backward Digit Span
Another good proxy for g is working memory, or the capacity to manipulate information committed to memory for the purposes of solving problems. More intelligent people tend to be better at this. This makes sense because if you have a good working memory the amount of information that you can handle will be greater, allowing for more complex problems to be solved. This ability is reflected in measures such as ‘digit span’, where the subject is presented with a list of digits (called bits) and must immediately repeat them back from memory. If they can do this successfully then they move onto a longer list. The number of bits that the subject can recall successfully is their ‘digit span’. They can be asked to recall the digits in the order in which they were given (‘forward’), which gives a measure of short-term memory, or they can be asked to recall them in the opposite order. The latter is known as their ‘backward digit span’ and is a measure of working memory. Clearly, remembering numbers in reverse order is likely to be more cognitively demanding and is therefore a much better measure of g.
In a re-analysis of previously published data, covering the period 1923 to 2008, it was found that forward digit span (short-term memory) had slightly improved over this period. However, backward digit span (working memory) had declined—equating to an IQ loss of 0.16 points per decade. In other words, we have gotten better at the less g-loaded memory task and worse at the more g-loaded task over a period of 85 years. This is clearly more evidence for the co-occurrence model.[19]
Replication of this finding came from a very large meta-analysis of short-term and working memory performance from several countries across several decades, conducted by a team from King’s College London. In this study, it was found that two separate measures of working memory (backwards digits and Corsi Blocks) both showed evidence of having declined, even after controlling for the sample’s national origin, its age, and differences in the way in which the measurement had been obtained. The short-term memory tests (forwards digits and the short-term memory variants of the Corsi Blocks test) both showed the opposite trend—a Flynn Effect, exactly as predicted by the co-occurrence model.[20]
Improvements in the environment are therefore raising certain weakly g-loaded abilities, such as short-term memory, but, at the same time, general intelligence (as reflected in working memory) has been decreasing throughout the 20th century.
Spatial Perception
Two Austrian psychologists—Jakob Pietschnig (who we met earlier) and Georg Gittler—have worked together to examine changes in scores on a test of spatial perception: the three-dimensional cube test (3DC).[21] As we’ve seen, spatial intelligence is an excellent measure of general intelligence. The two Austrian researchers found 96 samples (amounting to 13,172 people) to whom the 3DC had been administered in German-speaking countries between 1977 and 2014. They ‘meta-analysed’ the results—this is a statistical procedure whereby you combine data from multiple studies—meaning that they could see what had happened to spatial perception ability in German-speaking countries over this period of almost 40 years.
Their results were entirely consistent with the other lines of evidence that we’ve examined and, in particular, with IQ scores. There was an initial increase in performance (a Flynn Effect) and a subsequent decrease in performance (a negative Flynn Effect) when controlling for age, sex, and sample type—whether examining general population, convenience samples (e.g. university students), or a mixture of the two. Thus, it would appear that spatial perception performance was pushed to its phenotypic limit by environmental factors after which the underlying decline in general intelligence began to reveal itself. Put simply, we are getting worse at understanding how three-dimensional objects work. This has obvious implications for road safety and, of course, safety in the skies.
Piagetian Developmental Staging
Jean Piaget (1886–1980) was a Swiss clinical psychologist who produced ground-breaking work on childhood development. He is best known for his theory of cognitive development, in which he charted the four stages of development that children go through up until around the age of 16. From birth to age 2 they are in the ‘sensorimotor stage’ in which they experience the world purely in terms of sensory stimuli. In stage two, from age 2 to 7, they are in the ‘pre-operational stage’. They have stable mental concepts, but difficulty with logic. In stage three (7–11), the ‘concrete operational stage’, they can think logically but are still limited in terms of what they can physically manipulate. Finally, in the ‘formal operational stage’ (11–16), they develop their abstract reasoning skills.
Piaget devised a series of methods for testing where children were developmentally, which, to a significant degree, can be regarded as measuring intelligence.[22] One such measure is ‘volume and heaviness’, which looks at the ability of children to correctly estimate these quantities. British psychologist Michael Shayer and others compared English 11-year-old school children in terms of their scores on this, looking at samples from 1975 and then from 2003. They found that children had become considerably worse at it: girls by roughly half a standard deviation and boys by an entire standard deviation. Whilst this would be in line with average general intelligence decreasing over this period, these declines are simply too massive to have been caused by this process alone.[23] A recent analysis noted that the collapse seems to have been concentrated specifically in the numbers of top scorers over this period.[24]
Genius Levels and Macro-innovations
As has been discussed already, per capita rates of genius and the macro-innovations for which they are responsible have been declining since the early to middle of the 19th century. Charles Murray (whom we encountered earlier) has shown that these major scientific breakthroughs reached a peak around 1825.[25] This, of course, is the height of the Industrial Revolution. Then they go into decline, as do the rates of macro-innovation. Based on Murray’s published data, we have calculated per capita levels of genius across time restricted to just the English-speaking countries. The trend can be seen in Figure 8.
Figure 8. Per capita geniuses in USA, UK, Canada, New Zealand, and Australia, 1600–2000 (using publically released data from Charles Murray), fitted to a third-order polynomial curve.
It should be stressed that there are likely to be other factors behind the decline in genius and macro-innovation apart from simply falling g. As was discussed previously, genius may be dependent in part upon rare genes with large effects on the phenotype that do not ‘breed true’—meaning that they do not transmit faithfully across the generations. It may also be dependent on interaction between genes that have a low probability of combining in a single individual (giving rise to a very unusual kind of personality). This means that as populations shrink, due to the demographic transition (i.e. the general tendency for Western populations to have fewer children since the Industrial Revolution), the size of the ‘pool’ of these rare gene combinations has decreased and so they are less likely to occur. Recall also that genius is likely a highly group-selected manifestation of intelligence—geniuses help their populations during times of war, and their innovations have furthermore fuelled the rise of empires. This means that populations that are in situations of war or which are building empires, also placing them in situations of conflict, will be under group selection pressure and the group with the optimum number of geniuses will be more likely to expand. This will increase the probability of rare genetic variants and combinations of variants emerging, leading to new geniuses.
Population shrinking in the West is in part, therefore, a manifestation of the collapse of group selection, and is characterised by decreasing inter-group conflict (i.e. reductions in the frequency, duration, and lethality of warfare). The environment is milder today climatologically (warmer temperature has historically promoted inter-group peace[26]), furthermore ecological stress stemming from disease is largely absent from modern life, with the advent of modern medicine. The result is that the populations of contemporary Western countries are pacified and are no longer inclined towards battling for living space. So, selection pressure for genius has decreased and, accordingly, the relevant genetic combinations are not being selected for and they are less likely to combine anyway because the (native) population is shrinking. So, we have a perfect storm—selection for low IQ comes at the expense of group-level fitness, which via population shrinking massively reduces the frequency of the rare elements that must be present in order for genius to manifest.
