Friday, September 9, 2011

Flynn Effect 101

A new study on the Flynn effect (Wai and Putallaz 2011) is providing evidence that the Flynn effect has similar impact across the entire range of human intellectual ability (from low IQ folks to high IQ folks) and the Flynn effect continues to be prominent in the U.S. school-aged population through the present day. No matter when, where or how we look at the Flynn effect, it continues to sit there stubbornly as a cold hard ubiquitous fact.

An unexplained cold hard ubiquitous fact.

Let me take this occasion to describe once more the concept of intelligence from a new perspective, a perspective that apparently has never been considered before but a perspective that has the merit of making some sense out of the Flynn effect.

I think most scientists struggle to explain both intelligence and the Flynn effect because they have prematurely convinced themselves these are brain-based (neural) phenomena. At best, this is only partially true, and it’s not the most important part. The brain plays only a secondary role in human intelligence (an essential role, but secondary nonetheless). The primary source, the primary location, of human intelligence is the human environment.

So as not to confuse concepts, let me dub this primary source of intelligence as environmental intelligence. As such, environmental intelligence can be defined as the amount of non-biological pattern, structure and form that has been tangibly introduced into a given physical environment. In the current human environment, covering nearly the entirety of the Earth’s surface, that would encompass a very broad list of items—the structure in our languages, the way our streets are laid out in grids, the funny-looking but rule-based markings in arithmetic books, the symmetry and repetition of ceiling fans, and so on.

The Earth’s surface once had none of these items that form the corpus of environmental intelligence, but is now literally covered in them, with more plasterings arriving every day. And that observation is the simple key to understanding the Flynn effect.

If you accept that the amount of environmental intelligence is the measure of intelligence, then the Flynn effect becomes, if you’ll pardon the pun, a no-brainer. No one would question that the amount of environmental intelligence has been steadily increasing over the years—the physical world today contains far more non-biological pattern, structure and form than it did in previous times. And this steady increase would date all the way back to man’s great leap forward, meaning that the Flynn effect, at least as measured by environmental intelligence, is not just a twentieth-century phenomenon, and there is no reason to expect it will end anytime soon.

In contrast to environmental intelligence, the other concept that plays a role (the secondary role) in this new perspective on intelligence is a concept I will dub individual intelligence. Individual intelligence is much closer to what people traditionally think of when they consider intelligence, and like environmental intelligence, individual intelligence can be provided with a definition. Here is what I propose:

Individual intelligence is the ability to absorb environmental intelligence and respond constructively to it.

The merit in this definition is that it corresponds almost precisely to what we measure with intelligence tests. If you look at the content of any intelligence test, you’ll see that it consists almost entirely of the types of items that fall under our given definition of environmental intelligence—that is, we pose problems regarding such things as the structure of our languages, the geometry of our surroundings, the workings of our arithmetic, the recognition of symmetry and repetition, and so on; we ask questions based upon the non-biological pattern, structure and form that generally exists within our surrounding environment. (Note that we do not measure an individual’s ability to mate, gather food or run the 100-meter dash. When it comes to measuring intelligence, our biological capabilities are mostly irrelevant.) For a person to do well on an intelligence test requires first that he understand and recognize the context of the problems (that is, he has to absorb something about their environmental intelligence) and next that he respond accurately to them (which is to say, constructively, not destructively or randomly). Intelligence tests are of course not perfect—they serve only as a proxy for individual intelligence, but by and large they are considered to be a fairly decent proxy.

As the definition says, individual intelligence is an ability (as opposed to environmental intelligence, which is an artifact). This ability varies from person to person, and therefore it’s not unreasonable to assume an assortment of genetic, neurological and developmental factors could be driving this variableness. Indeed, if individual intelligence were the only aspect of intelligence to consider, then perhaps I might understand science’s preoccupation with the human brain. But as it turns out, individual intelligence is not the only aspect to consider, and when it comes to the Flynn effect, individual intelligence is not even the consequential aspect.

Armed with these definitions of environmental intelligence and individual intelligence, let’s examine what happens when humans take intelligence tests across a range of time. Let’s begin with a hypothetical moment in time we’ll number as 1. At time 1, we can consider three individuals who are taking the standard battery of intelligence tests for that moment, and we’ll name these three individuals A1, B1, and C1.

Keep in mind that these intelligence tests are serving as a proxy for measuring individual intelligence, which by definition means they are measuring the amount of environmental intelligence each test-taker is able to absorb and to respond to correctly. As such, we can state the test-takers’ results as a percentage of environmental intelligence that each has accurately handled (as contained in and measured by the test):

  • · A1: 80% of environmental intelligence accurately handled.
  • · B1: 70% of environmental intelligence accurately handled.
  • · C1: 60% of environmental intelligence accurately handled.

These results can then be compared to those of other test-takers and normed (as is done routinely in the non-hypothetical world), and based upon such statistics, we might come to the conclusion that A1 is displaying high intelligence (high individual intelligence), B1 is displaying medium intelligence, and C1 is displaying low intelligence.

Note that everything quantified and assessed so far is only a relative measure of intelligence (no absolute measure of anyone’s intelligence has yet been stated). Even so, nothing is preventing scientists from taking the details of these results and performing all sorts of psychometric analyses, including such things as g-factor analysis, intellectual disability analysis, and so on (and of course in the non-hypothetical world, that is precisely what gets done). No one needs an absolute measure of intelligence to perform all the currently popular forms of psychometric intelligence analysis; everything is based solely upon relative measures of intelligence. That is the nature of individual intelligence—it is premised strictly upon relative abilities.

What would it take to obtain an absolute measure of the intelligences of A1, B1, and C1? Note that even though the contents of the intelligence tests are based upon the artifacts composing environmental intelligence, and even though the results are stated as a percentage of environmental intelligence accurately mastered, and even though the comparison of these percentages allows scientists to obtain a wealth of information regarding individual intelligence, still, an assessment of absolute intelligence can only be obtained by going back and quantifying environmental intelligence. By definition, an individual’s absolute level of intelligence consists of two separate inputs: one, the individual’s ability to master environmental intelligence (measurable by intelligence tests), and two, the amount of environmental intelligence that exists to be mastered.

As a practical matter, quantifying environmental intelligence would not be an easy task, but as a theoretical matter, there should be no objection—environmental intelligence simply consists of tangible aspects from the physical world. In my hypothetical example, I’m going to assume someone has surmounted the practical difficulties at time 1 and has measured the level of environmental intelligence to be about 200 (200 ei, if you need a unit). Keep in mind that this quantity represents the amount of pattern, structure and form that can be found in the human environment at time 1.

Using this quantification of environmental intelligence at time 1, we can now state the intelligences of A1, B1, and C1 in absolute terms:

  • · A1: 160 ei (80% times 200 ei)
  • · B1: 140 ei (70% times 200 ei)
  • · C1: 120 ei (60% times 200 ei)

Interestingly enough, this effort actually produces nothing of psychometric significance for time 1. Everything scientists would need to study individual intelligence is already available in the relative scores—the addendum of absolute scores serves as little more than an adornment. Still, we will be glad to have gone the extra mile, for the possession of these absolute scores will soon prove to be invaluable, when we make the leap forward to time 2.

To set the stage for time 2, we’re going to assume we’ve crossed the temporal range of a handful of generations. The world has changed a bit. Transportation has improved—now more widespread and somewhat faster. Languages are a bit richer, mathematics has added a few discoveries and insightful techniques. The architectural arrangement of the everyday world has become just a little more complex and a little more structured—you know, the usual improvements. Those with the tools and the know-how have taken a measurement of all these external changes, and at time 2 they quantify the total amount of pattern, structure and form existing within the physical human world to be approximately 400 ei, twice what it was at time 1.

A1, B1 and C1 are no longer with us, but we have the next best thing, some of their descendents, namely A2, B2 and C2. Because it’s been only a handful of generations, A2, B2 and C2 are genetically and neurologically quite similar to their time 1 ancestors; in fact, we can assume that biologically they are nearly exact equivalents. Even better, A2, B2 and C2 have retained the exact same intelligence strengths relative to one another and relative to the general population (as measured by performance on intelligence tests). Here’s a chart that highlights this equivalence:

  • · A2: 80% of environmental intelligence accurately handled (high intelligence).
  • · B2: 70% of environmental intelligence accurately handled (medium intelligence).
  • · C2: 60% of environmental intelligence accurately handled (low intelligence).

When a psychometric analysis is made of these individuals and their population cohorts, the results come out essentially unchanged from time 1. Factor analysis and correlation statistics look quite similar, all relative measures have the same relativity, and everything is pointing to the unmistakable conclusion that as far as individual intelligence goes, time 2 is no different than time 1.

On the other hand, there is one remarkable thing to note about those intelligence tests. (Well, at least Professor Flynn would have found it remarkable—he might have been inclined to make my hypothetical world more aware of its existence.) The first set of intelligence tests that were given to A2, B2, and C2 were the exact same tests that had been given to their ancestors, A1, B1 and C1. And wouldn’t you know it, A2, B2 and C2 found those tests to be laughably easy! In fact, the tests were so laughably easy and everyone did so well on them, they were no longer useful for making relative comparisons. The test-givers had to throw them out and come up with new tests.

More difficult tests. Tests containing a larger amount of pattern, structure and form.

It doesn’t take much investigation to figure out why the time 1 tests are no longer useful in the time 2 world. Although A2, B2 and C2 have not changed one iota in their relative intelligence to one another and in their relative intelligence to the extant population, and although A2, B2 and C2 have remained exact equivalents to their ancestors in all neurological aspects, nonetheless their absolute intelligence has taken a quantum leap. And this is solely due to the fact that environmental intelligence has taken a quantum leap as well:
  • · A2: 320 ei (80% times 400 ei)
  • · B2: 280 ei (70% times 400 ei)
  • · C2: 240 ei (60% times 400 ei)

An intelligence test given at any particular moment in time must reflect (must consist of) the environmental intelligence that exists at that same moment. A2, B2 and C2 are prepared to face an intelligence test that reflects the time 2 environment because that is precisely the environment they have been absorbing. There will be individual differences of course, but the overall environmental influence remains primary and impacts each individual nearly the same, and since the time 2 structural environment is essentially an augmentation over the time 1 structural environment, the inevitable consequence is that anyone raised at time 2 will find himself making relative mincemeat out of a time 1 test.

And that, my friends, is the Flynn effect in a nutshell. Of course this description requires we give up our old notions regarding intelligence, requires first and foremost we quit insisting intelligence must be found inside our neurons. That scientific dogma is certainly quite popular, almost to the point of unanimity, but I’m convinced it will do nothing for helping explain the Flynn effect.

No, the primary source of human intelligence, the primary driver of the Flynn effect, is to be found in the increasing amount of environmental intelligence that can be readily witnessed all around us. Cognitive scientists are certainly free to keep searching for intelligence within the neuronal tangles of the human brain, but I’m afraid they will be looking long and hard without result. If scientists really want to see the material form of human intelligence, then I would suggest they do something quite different, something quite radical.

I would suggest they open their eyes.

(Wai and Putallaz 2011): Wai, Jonathan; Putallaz, Martha. 2011. “The Flynn effect puzzle: A 30-year examination from the right tail of the ability distribution provides some missing pieces.” Intelligence (in press).

No comments: