Monday, April 23, 2007

Mission Impossible – IQ as a meaningful measure

A generally accepted definition of intellligence is “the global capacity to act purposefully, to think rationaly, and to deal effectively with the environment.” However, if we want to objectively evaluate the intelligence of a human it becomes difficult because there is no operational definition for intellligence. Psychologists solve that problem today by using the score achieved on one of two tests – the Stanford Binet or the Wechsler, as the operational definition. However, insistence upon using such a figure presents both psychometric and sociological problems.

I will tackle the psychometric problem first. If we look at the problem of measuring the capability or output of even simply objects the task is daunting. The biggest difficulty is coming up with only one measurement that is a reasonable and fair representation. As a result, attempting to measure the capability or intelligence of the human brain is probably an impossible task. Let me attempt to explain this rather emphatic statement by looking at a few nonhuman examples.

A light bulb is a very simple device. Using wattage, or the power it consumes, is a very good representation of its light output. However, if the object gets more complex we run into problems. If we look at the receiver in today’s home theater there are several measurements that are important to its performance – power, digital decoding and processing (dolby, THX), and connectivity (devices, optical, digital). None is sufficient to evaluate its performance. Finally, if we look at a computer, which is closer to the task of human measurement, we are faced with even a bigger problem. For like humans it is multilayered with both hardware and software. Potential hardware measurements are: cpu speed, bus speed, memory size, and connectivity capabilities, while software measurements are operating system, and appication programs. None gives us a true and accurate picture of the capability of the computer, evaluation requires at least these measurements and maybe more.

With these examples in mind we can look at the challenge of measuring human output or intelligence. Like the computer we are multilayered. We have a genetic component, our brain, and a cultural component which in itself is multilayered. The cultural component consists of language and our experiences. To date we have not even been able to evaluate the proportion each contributes to a human let alone a measure of their of their capability. Thus, given our complexity a single number representing our mental capacity isn’t currently possible.

Even the consideration of future capabilities present problems. What if we had a probe that could be placed within the brain to measure IQ. Where would be put it? The problem becomes similar to that of measuring the computer. Our physical thought process is so complex that measurement of its capabiltiy becomes impossible.

The sociological problems have to do with the label that such a measurement produces. An IQ score that is below or substantially below average could have a detrimental effect upon an individual and to the extent that IQ shows a cultural bias, could have an impact upon large groups of people. Childen who are slower at mental maturity are at a disadvantage. In fact, the State of California has prohibited the use of IQ tests in the placement of certain students.

If a single figure is problematic that could mean that we have more than one “intelligence” and multiple measurements are necessary. Gardner has suggested that there might be multiple intelligences. This seems like a much better theory, however, the exact number may be difficult to determine.

Notwithstanding this there are good reasons to have some measure of a persons specific capabilities. Such tests are referred to as attitude tests. These tests can successfully predict a person’s ability to acquire a set of specific motor or intellectual skills, though future education. They perform a useful function in both business and academia. Their validity can be checked with statistical analysis thus reducing or eliminating any sociological problems.