Notes on Progress: Breakfast with g
IQ, lived experience, and my boyfriend’s underpants
In this issue of Notes on Progress, Rosalind Arden (who previously wrote for us about intelligence in dogs) argues that general intelligence is not a true underlying ability, but merely a reliable finding that different sorts of intellectual abilities correlate. If you enjoy it, please share it on social media or by forwarding it to anyone you think might find it interesting.
I cannot imagine, now, what possessed me to show my boyfriend’s underpants to my new boss at our first meeting. I should like to say it was years ago when I was young and foolish. But actually, it was just after 11 o’clock this morning (not this, this morning, but this morning when I wrote that sentence). There was a handsome Go board in his office and an orange plush zoomorphic thing on a shelf; they relaxed me. Perhaps too much. Also, he had shown me the base layers of his skillful lily painting and I wanted him to see the flower I’d embroidered. On the front—technically called the crotch I suppose, but front sounds more wholesome—there had been a rip along a seam which I’d repaired with red embroidery silk before getting into the swing of things and going for the full celandine. It would never have convinced Elizabeth Blackadder, botanically, but it was yellow with green leaves which features had exhausted my sewing skills.
Reflecting afterwards on the train down to visit soon to be bewildered (appalled, possibly frightened) boyfriend, I thought about the quality of emotion generated by meeting one’s new boss. It isn’t quite nervousness, but it carries a sort of heightened energy that increases the likelihood of exchanging confidences, even unexpected ones. And the new boss’s twitter account was way funner than expected for a philosopher and, lastly, he was blindingly good-looking which I think made me a bit panicked yet also intrigued because he was helpful and seemed smart. I mean, Go, right? Who doesn’t love Go? I’d gone to see him about various things, all of which he'd been kind and supportive about. The only matter I had forgotten to raise was the issue of reporting on direct expenses.
Why do institutions set up systems which take a week to learn, are replete with arbitrary codes, are encountered so infrequently that one never remembers how to work them from last time? You have Professors of Recursive Clauses giving up class preparation time to spend a week squinting at bus tickets and rain-spattered bookstore receipts, quibbling with the photocopier, struggling to parse error messages that sound like missing shoehorns (Boot device BSOD awol or something). Surely someone who has developed expertise at these sadistic systems could be paid richly to do the job more competently causing fewer shoes to be hurled at innocent monitors. Didn’t Adam Smith have some idea about the manufacturing of nails, and how not to create processes that make you want to drive them into your motherboard?
While in grumbling mode: what’s with the proliferation of the phrase ‘lived experience’? It seems like a dog whistle for something: mostly ‘mine trumps yours’. Or ‘never mind all that counting stuff’. I drifted into a free exhibition on Understanding Cancer, at the glorious huge new Francis Crick Building in London. Some dearly beloveds have cancer, so I thought I’d supplement the bit of reading I’ve done with a visit. It was a great disappointment. Several videos all played at the same time, which was a lot like an exercise designed to help people understand the experience of having schizophrenia – plural voices make it hard to concentrate on any single thing. Video must be about the slowest way to get information out of one head into another.
The exhibition was information poor; several mentions of the ‘lived experience’ of cancer. Why not just ‘the experience’? what is ‘lived’ whistling up? Also, the Royal Marsden, Britain’s superb cancer hospital and the lucky recipient of my coffee money recently, supplies plenty of stories about people’s experience – and guess what, they are all individual. Alex’s experience doesn’t tell me much about what Bertie’s going through with cancer. Which is why I’d hoped for more stuff with counting, measurement, facts, mechanisms, what do we know, what are the key questions and where are we at with any of them. What treatments work for what kinds of cancers?
What was he thinking? Charles Spearman calling it g for general intelligence. I mentioned in my earlier diary that Spearman had assigned the term g or the g factor for the parameter that captures the propensity for school grades to overlap (students who did well in Classics also tended to do well in Mathematics, Geography and so on). Spearman’s g stood for general. This nomenclature has led us astray and the error shows up starkly when we think about non-human animals.
Let us think about non-human animals, then, because a colleague and I are on opposite sides of this general versus not general argument and we have a fried breakfast with all the extras riding on the outcome. I only hope we both still have teeth by the time we learn who’s right. Although I suppose you could gum your way through a fried egg and toast. But first let’s re-cap on g.
It's worth saying, ahead of the g recap, that much of psychology is currently convulsing like an electrocuted walrus because many long-held facts, based on famous studies, have turned out not to replicate. To be, in fact, false; to not be facts at all. Science is ineluctably provisional; it offers approximations towards the truth which lead us, like the (possibly red) embroidery thread Ariadne gave Theseus, out of the dark maze of ignorance and into the sanitising sunlight. Learning that one was wrong at the margin about this or that is situation normal in science, but it’s a bit upsetting when the table gets tipped up and the tureens, dishes, cruets, salvers, plates, cutlery, and glassware containing our collective wisdom slide off in a cacophonous crashing heap. What a lot of clearing up to do. High school students and undergraduates whose textbooks are expensive, yet misleading in important places, must feel bilked.
But there are pockets of psychology that have replicated well, partly perhaps because they always depended on large samples. The g factor is, as the tune has it, Still Standing. It is hard to find any task containing a modicum of mental effort, that is uncorrelated with valid and reliable measures (such as the Weschler Adult Intelligence Scale or WAIS) of cognitive ability. It’s has been an entertaining parlor game for psychologists to demonstrate that X or Y mental task is uncorrelated with cognitive ability, but so far, it’s nul points for the naysayers.
Life is an IQ test. Everything we do that involves thinking is saturated to a greater or lesser extent with intelligence. Imagination is saturated with it, although we are a long way from being able to measure imagination, perhaps mercifully. Take an average day: get out of bed, find bathroom, clean teeth, find kitchen, make breakfast, leave house, remember clothes, return, put them on, go to stable, ride to work, dismount, clean manure off boots, feed horse, practice juggling. You know, you live this. But think about it. Every step of every day involves your brain. Daily life is a series of activities, some are taxing, others ‘automatic’, but it’s all output from your brain. What g is, formally, is a quantified expression of the fact that brain-juice soaks life.
If you create a correlation matrix from scores derived from ten cognitive tasks administered to a thousand people, you’ll find that each cell off the diagonal will show a positive value. Every score correlates positively with every other score. The values will vary, indicating stronger or weaker relationships, but there will be an all-positive manifold. Higher scores go with higher scores throughout the mental magisterium.
Statistical data-reduction techniques like factor analysis, can be used to ‘boil-off’ cognitive scores to leave a single residue that captures much (usually around half) of the variance in scores. The ‘usually around half’ is called g. It’s an emergent property, it’s statistical. g is not an ability, not a place in the brain, not a physical thing. It is the most robust, best replicated, least likely to be false, finding in the whole of the human behavioral sciences, but its nature is not yet well understood. What causes g, and is g really general intelligence? Yes, shouts my colleague hoping for a free full English. No, I assert, pining for fried bread.
Here's my Gedankenexperiment: take 100 vertebrate species (for simplicity). Let’s assume that we can operationalize four distinct tests which tap the cognitive abilities of each of those 100 species (our dream tests are both quantitative and reliable). Let me emphasize that the four tests are unique to each species. Now we administer the four tests to 500 members of each of the 100 species (each sample will contain animals of a similar age and a 50/50 sex split). What will we find? My guess is that we will find that the four test scores form an all-positive correlation matrix for every single species. A g factor accounting for 30%-60% of the common variance will emerge from each species. Yet among those 100 species there will be a dimension that I’ll call ‘cognitive flexibility’. By which I mean, for example, that border collies are, mentally more agile and flexible than, say, the black-spotted cuscus.
The g factor is not general intelligence, it is the common variance in mental abilities within a species; this may be somewhat general as in Pica pica, Vulpes vulpes or Gorilla gorilla, or somewhat specific as in Sula sula. Humans far exceed other species in wit if not always in wisdom. But even our flexible intelligence is only somewhat general. Our cognitive abilities evolved to solve human-relevant problems. These occur in one, two or three dimensions. That’s why we need mathematics to work through higher-dimensional problems. Some people can make the imaginative leap to four dimensions, but nobody can imagine in twenty. We generalize within our local planetary setting. Our minds are not infinitely plastic.
My full-english breakfast assertion is that the dimension capturing differences in cognitive flexibility between species – the dimension referred to by journalists who have often asked me whether cats are smarter – is distinct from the g factor which I believe is an individual level, within-species parameter.
My colleague, who’s hoping I’ll fork out for the sausages, argues instead, that the g factor itself has been under natural selection. He asserts that there is a true sense in which it evolves across species increasing from, say, monkeys who have less g, towards us, who have more; thus g can be aptly called general intelligence. This interspecies g is called G by my frenemy, to distinguish it from g (try that on the Audio book). He argues that discovering inter-species G paves the way for a trans-species IQ-type test. Naturally if he were here on the page with me to present his view, he would offer a more comprehensive and convincing account of his pozish. That’s why he's currently locked in the understairs shoe cupboard. The dust bunnies won’t count themselves, you know. Does it matter a fig either way? As a general principle, knowing more rather than less is good. Since g is a key phenomenon, a potent predictor of important outcomes, getting a grip on it is good basic science.
In summary, with black pudding eggs bacon fried bread hanging on it. Is g general or is it simply common variance with a species (or even within a population of the same species)? It’s breakfast epiphanies folks. Waffles and berry Capote one might say. Votes on a postcard please.