By
Bob Williams
Conventional
IQ tests have been in use for close to a century. After so much time, standard IQ tests have been refined and now
stand on well-understood statistics, but there is continuing interest in
alternative ways of measuring intelligence.
Concerns over potential test bias led to the development of culture-fair
and culture-free tests, which are based on the use of figures, pictures, and
other culturally inert test items. IQ
tests have in common that they measure the cognitive process downstream. If one were to use a computer analogy, IQ
tests are much like the benchmark tests used to compare computational
power. A more basic measurement of
computer processing might consist of measurements of the time required to
perform individual operations, the size of information units, and the speed
with which information can be moved from one processing stage to another.
Among the
limitations of IQ tests is the fact that they are comparative and do not have a
true ratio scale. An IQ of 100
corresponds to a mean for a large population sample, but does not indicate an
intelligence that is twice that of IQ 50 or 2/3 of IQ 150. It would be useful to measure intelligence
both from a more fundamental perspective and in a manner that could lead to a
true ratio scale; there is some hope that chronometric measurements will
eventually provide such a scale, at least over the center portion of the
intelligence distribution curve.
Chronometric
measurements go back as far as Galton around 1908. The concept of linking intelligence to reaction time failed to
gain traction in part because measurements did not separate reaction time from
motor time,[1]
and in part because researchers did not use an adequate number of trials per
test. IQ tests were being developed at the same time and were so successful
that they became the focus of psychometric attention for decades. By the 1960s a number of researchers began
to investigate mental chronometrics, but had relatively little impact on the
focus of psychometric research.[2] During this time, “choice reaction time” was
suggested by Roth and was then investigated by Jensen.
Arthur
Jensen was among the first to begin time related measurements as a means of
studying Hick's Law.[3] Hick's Law is based on the psychometric
definition of a "bit."[4] A bit is the logarithm to the base 2 of the
number of choice alternatives. One bit
of information reduces the total amount of uncertainty of choice by one-half.
Hick observed that reaction time (RT) increases linearly as a function of the
number of bits contained in the stimulus.
The implication of Hick’s Law is that as the brain responds to
increasingly complex external stimuli, it calls upon additional sequential
processing steps (sometimes called modules), each of which adds to the
response time of the brain. When RT is
plotted against bits for homogeneous groups, the slopes of the more intelligent
groups are always less than for the less intelligent groups. Thus, there is less variance in RT as a
function of intelligence among more intelligent people.
Since
Jensen’s experiments, research in various aspects of chronometrics has steadily
increased. The number of papers that
address various aspects of brain speed, intake speed, and information
processing speed has become so large that just locating and listing them would
be a huge task. For that reason, this
discussion will consist of a brief review of some of the more salient aspects
of chronometric research, especially as it may apply to very high intelligence,
with some space devoted to the application of speed measurements in the upper
range.[5]
There are
several approaches to chronometric measurements. The two that are most directly related to the large-scale
research that has taken place in the past three decades are Reaction Time
and Inspection Time. Other means
of time related testing include timed tasks, eye blink response,[6]
and electroencephalography.[7]
RT
measurements are typically based on a simple laboratory apparatus, which
consists of a home button and multiple response buttons.[8] The device used by Jensen consists of a home
button and 8 response buttons, arranged at equal distanced from the home button
in an arc of 180 degrees.[9] Each response button has a light directly
above it (or, in later devices incorporated lighted buttons). The person being tested holds the home
button down and then must press the button closest to the light that is
illuminated as the “external stimulus.”
The measurement device records the time the stimulus lights goes on; the
time the home button is released; and the time the response button is pressed:
T1 ………. T2
………. T3
The time
interval from T1 to T2 is the reaction time, RT. The time from T2 to T3 is the movement time (MT).[10] This simple test is an example of an
Elementary Cognitive Task (ECT) and can be completed by any adult (even with
IQs as low as 15 to 20), usually in less than 1 second. Individuals with IQs below 40 require some
acclimation to the test. Jensen’s tests
covered individuals from IQ 15 to 150.
There is essentially no correlation between MT and intelligence.
RT studies
have used a variety of ECT formats. For
example, a very simple test apparatus involves a response pad that has only the
home button and two response buttons.
The response buttons may be labeled as yes-no, true-false,
same-different, etc. One frequently
discussed ECT is known as “odd-man-out.”
The format is to use the 8-button console, presenting 3 lights on; the
correct response is to press the button which is illuminated and which is the
most isolated (two on-lights will be closer together than the third). It turns out that each of the ECT formats
has an IQ correlation of from about -.2 to -.4. [This compares to an average correlation between individual IQ
test items of about 0.1 and of 0.2 to 0.3 to the total test score.[11] Each also has a correlation to an ability
that is specific to the task. The RTs
from a battery of ECTs may be combined to produce a net correlation of about
.70 (sign intentionally converted to positive, since it is the test battery
correlation).[12]
·
As
task complexity is increased, RT increases linearly as a function of bits.
·
Just
a curiosity, but it happens that in a yes-no binary response test, the RT to select
“no” is longer than for “yes.”
·
Not
surprisingly, RT increases as a function of age.
·
RT is
both involuntary and unconscious.
·
People
typically respond to 0 to 3 bits of information in the range of 300 to 400
msec, which is faster than the 500 msec believed to be the threshold for
conscious control. In other words,
responses begin before people have conscious control over their movement.
·
The
factor structure of RTs looks much like the familiar structure associated with
IQ tests. RT tasks correlate to form
process factors. These processes then
correlate to two groups: information-processing (IP) and a common non g
factor. All of the g correlation
lies within the IP factor. As the non g
factor increases, it does so at the expense of the correlation to g,
thereby placing a ceiling on the RT correlation to g.
Because of
the last item (two loadings), it is attractive to find a chronometric mechanism
that does not load on the motor movement factor (uncorrelated to g). A test has been devised that measures the
minimum time required for the brain to perceive an external stimulus (auditory
or visual). This measurement is called
inspection time (IT). The standard
format for the visual test is to display a figure which looks like the Greek
letter phi. One of the two vertical
legs of the figure is shorter than the other by approximately 50% (sometimes
less). The figure is hidden by a mask,
which is essentially a thick display of the phi figure, with equal length
legs. The subject is shown the masked
figure, then the mask is briefly removed, revealing the test figure; he
responds to the display by pushing a left or right button to indicate which leg
of the figure was longer. The display
sequence is MASK --- FIGURE ---MASK.
Since there is no time pressure on the subject to press his response
button, the motor loading of RT testing is not present.
A quick
scan of chronometric research papers over the past decade shows that it is IT
that has been the focus of attention.
Brand, for example, discusses IT at length, with little comment
concerning RT, other than to note that it was Jensen who was the primary RT
investigator in the 1970s.
The device
used to display the standard test figure is known as a tachistoscope
(T-scope). These instruments[13]
are capable of displaying the response figure for a very short time. Because the general appearance of operation
of the T-scope can be replicated by a computer display, there has been a lot of
interest in developing desktop computer systems that can perform IT
testing. Unfortunately, there are
problems associated with computer screens.
IT tests
using computer screens have sometimes caused errors due to a visual cue that is
seen by some subjects; they see a “jump” at the bottom of shorter vertical line
when the mask is reapplied. Not all
people can perceive this cue and those who can do not seem to be differentiated
by intelligence. Some researchers have
devised a form of mask that uses randomly selected patterns and visual noise to
counter the use of the visual cue and have reported increased negative IT to IQ
correlations when the masking adjustments were used.[14]
As
recently as last year, there was a study reported in which LEDs were used to
form the traditional IT figure and mask.[15] The arrangement consisted of 88 LEDs. The interesting finding was that when a
computer display IT format was used, it did not compare directly to the LED
measurements. The researchers believe
that there was a high degree of specificity associated with each measurement
device and that the LEDs were giving movement cues. Problems of this sort illustrate the difficulties of comparing data
taken by different procedures.
Another
aspect of computer screen display, as compared to the T-scope, is that a
computer screen paints the display one line at a time, causing a variable that
relates to the refresh rate and causing the display times to be adjustable to
values that correspond to the limit imposed by the speed of the display screen.[16]
There is a
similar auditory test which consists of a mask sound, followed by a brief tone
of higher or lower pitch, followed by the mask. The subject presses a button to indicate whether the tone changed
to a higher or lower pitch. The
auditory IT is slightly shorter than the visual IT (similar results are seen
when sound is used for RT measurements).
One limitation to auditory IT, is that some people are unable to distinguish
between the tones, independently of their intelligence.
Until
recently, IT was believed to have the advantage over RT in that the largest IT
factor loading is on g. While
this may still be true, with respect to “largest,” it is now known that IT
loads on “a perceptual speed factor that figures in some tasks that are
typically loaded on a spatial ability factor or the nonverbal 'performance'
factor of the Wechsler IQ battery.”[17]
IT
correlations to IQ have been reported over the range of -.55 to -.76. Most of these have suffered from restriction
of range, leading Brand to estimate that the most likely true correlation, for
the entire population, would be -.75.[18] This number is approximately the average for
correlations from one standard IQ test to another.[19]
Chronometric
research has established its importance beyond any doubt. Whether RT or IT is used as a measurement,
the results are clearly due to basic functions of the brain and those reflect
the variance in human intelligence. As
Jensen put it, “… it is evident that RT and IT are probably closer to the
interface of brain and behavior than any other purely behavioral g
loaded measure in experimental psychology’s armamentarium.” There is, however, a question as to what
factors are responsible for discrepancies between various experiments. The huge amount of interest in this area of
research has resulted in diverse studies that cannot be directly compared
because the researchers were investigating different phenomena, using different
testing procedures, and using different
instruments. In the examples briefly
discussed above, it is apparent that measurements done with a T-scope, computer
monitor, and LED array are not necessarily comparable, nor can the results of
different computer based set-ups be directly compared. RT and IT measurements have components that
load on factors other than g and have turned out to be more complex than
initially suspected.[20]
The expert
in the study of the study of the psychology of Aging is Timothy Salthouse. He has published several papers in Intelligence
that deal with age-related variance.
The most general finding is the obvious one that people slow down as
they age, thereby making direct comparisons of various studies impractical,
when the test subjects are not selected from the same age group. The decline in RT/IT measurements has been
independently verified to be substantially caused by a slowing of cognitive
processing (as opposed to sensory factors).[21] “Salthouse predicted that the correlation
between age and IQ should virtually disappear if mental speed was partialled
out. … the correlation between age and RAPM was - .28. After partialling out
latency, the correlation between age and RAPM was reduced to a not statistically
significant -. 10. Again, this generally supports Salthouse’s contention that a
decrease in mental speed is responsible for all age-related declines in
fluid intelligence.”[22]
[underscore added]
It is
commonly believed that timed IQ tests favor the more intelligent test
takers. The fact is that the best way
to measure intelligence (defining that as g) is by the use of power
tests, which are not timed. Jensen
notes that as time limits on standard IQ tests are decreased, the g
loading on those tests decreases slightly at first, then rapidly, as the time
allowed is decreased. The reason for
this is that when there is little time to respond to a test item, correct
answers are more likely to be those that are loaded on specificity (s)
instead of g.[23] When a test is being used to measure g,
the loading on s is equivalent to an error, because it does not
contribute to the g measurement.
The untimed RAPM[24]
is considered to be one of the most heavily g loaded IQ tests and has the
largest correlation with RT.[25] So, we have opposite time parameters with
respect to test time limits and RT as measured by ECTs.
The large
amount of research in chronometrics has demonstrated that the relationship of
speed to intelligence is quite complex and difficult to model. A thorough examination of the literature
would produce a list of hypotheses and hints as to what is and is not
happening, but there are two items that are of particular interest. The first is Jensen’s initial explanation of
the relationship between working memory efficiency and RT. He suggested that because working memory is
highly volatile and has a small capacity, information held in working memory
must either be used quickly, refreshed (amazingly similar to cathode ray tube
displays or even some forms of RAM) or lost.
Individuals with fast information intake can presumably make use of the
information in working memory more efficiently. This notion is consistent with both the observed lower glucose
uptake (correlates -.7 to -.8 with RAPM) associated with more intelligent
brains; the +0.41 to +0.46 correlation[26]
between neural conduction velocity and IQ; and with the simple negative
correlation of mean RT to IQ. Jensen’s
observation is supported by observations that the correlation between RT and g
is more highly correlated when working memory is forced to its limits than
otherwise.
Ed Miller
noted that not only does RT correlate to IQ but the standard deviation (skewness) of RT correlates to IQ. Smarter people have less skewness of their
RT measurement sets than do low IQ people.
And there is the even more interesting finding that skewness correlates
to g independently of median RT.
This observation, among a rather large number of others, supports
Miller’s neural noise model, which accounts for the skewness as well as a host
of other observed intelligence parameters.[27] Miller suggests that the degree of myelin is
largely responsible for the variance seen in RT and intelligence. If Miller is correct, there is potentially a
duel role for myelin in that the level of myelin correlates with higher neural
conduction velocity (consistent with Jensen’s RT observation) and is the major
factor in determining the degree of neural cross-talk – presumed to trigger
neural oscillations which disrupt information transfer in the brain.[28]
One
observation that Miller contends is consistent with the belief that myelination
is a large factor in IQ variance is that myelin builds up from childhood to
early adulthood – a period over which mental ability also increases. But some children display very advanced
mental abilities, that closely match those of college students. Miller suggests that such precociousness is
the result of myelin buildup at an earlier than normal age. His position is consistent with chronometric
findings that show that very bright children have faster RTs than their age
peers and that match those of college students.[29] The established relationship between
myelination and NCV adds additional credibility to the link.
ECTs have
been shown to be highly heritable, with varying correlations, depending on what
was studied. As ECT task complexity
increases, higher heritability is found.
Jensen reported two twins studies of the heritability of RT. The findings were 0.84 and 1.0. The clear significance of this strong
heritability in RT is that there is a corresponding and directly related
heritability in intelligence.
A number
of other areas of discussion can be tied directly to chronometrics. For example, RT measures correlate to
intelligence without regard for race (confirming various other race related
findings that have been based on conventional tests).
One of the
most interesting findings of RT was reported in January of 2004. The study[30]
consisted to two sets of children from the same primary school. Both sets were tested by the same researcher
(Wilson), using the same technique in 1981 and in 2001. IQ scores were based on the Peabody Picture
Vocabulary Test (PPVT) and IT measurements were done with identical
T-scopes. So, the research data was
taken from two very well controlled (similar, even to the point of verifying
normal vision) groups, separated by 20 years.
As one
would predict from the massive awareness of the Flynn Effect, the test scores
on the identical IQ test should have moved up.
They did. The 20 year gain was 5
points on the PPVT. If the Flynn Effect
caused real gains in g, one would expect a measurable decrease in IT,
but this was not observed. The median
and skew of the 1981 data matched the 2001 data. The two findings support the prior findings of J. P. Rushton,
using the method of correlated vectors, that the Flynn Effect is hollow with
respect to g and is entirely due to gains in specificity.[31] Nettelbeck and Wilson were very hesitant to
overstate the importance of their finding.
It is the only such study on record and is of limited scope. Even so, the result is quite interesting and
would have been predicted by those who believe Rushton’s conclusion is correct.
In the
same January issue of Intelligence, there is a paper that attempts to relate
personality and information processing speed.[32] The data from this study confirm the robust
correlation between IT and intelligence (Raven’s score). The finding was that personality and arousal
effects were apparent with respect to Raven’s scores, but had no effect in
IT. The authors concluded that
personality and processing speed (associated with g) exert independent
effects in IQ scores. The importance of
this is primarily in adding some information to the still murky models of
intelligence. Secondarily, it confirms
that people can be distracted when taking IQ tests (no surprise to anyone).
In the
foregoing discussions were selected to bring out some of the problems related
to IQ testing with chronometric techniques. Prometheus considered the use of
Thinkfast[33] and has
discussions of its attractions on the Web.
Whether Thinkfast is a useful wide-range IQ measurement tool or not is
unclear. Any chronometric measurement
must be carefully age corrected. If it
is an IT measurement, using a computer screen display, the test must be
verified against a T-scope.
There is a
natural limit to the negative correlations to IQ found in various chronometric
tests – nerve conduction is a chemical process, which is considerably slower
than an electrical transmission. As
intelligence goes up and brain speed increases brain speed has to approach the
physical limit of information transfer through sensory inputs and the
brain. There is also the matter of
loading on a speed perception factor, which is independent of g (as
previously discussed).[34] Both of these phenomena limit the maximum
resolution of high end intelligence by chronometric techniques.
There are
multiple references to IT measurements being nonlinear and possibly of little
use at the upper end. In particular,
Brand wrote that the full population r of -0.75 (his number is presently
disputed by some researchers[35])
drops to around -0.30 above IQ 115.[36] He also states that Nicholson found an IT r
of –0.64, but when divided into top and bottom halves of the distribution
curve, her data showed –0.80 and 0.0 respectively.
Brand’s comments,
however, are not completely consistent with other references. Jensen apparently found IQ to speed
correlations for individuals well above the median IQ (correlations present at
the Mensa level). As with much of
chronometric research, contradictory results are most likely due to differences
in test procedures or in the groups studied. It is probably too early to draw
firm conclusions concerning the applicability of chronometric measures for high
range testing, and possibly for even moderately high range testing.
My study
of chronometrics has convinced me that mental intake speed, or perceptual
response, or whatever we wish to call it is a reflection of an important component
of intelligence. It demonstrates that
there are very low level factors which account for a significant portion of
psychometric g. It does not,
however, measure the full range of variables that contribute to intelligence,
because the high level components of the thought process are “out of the
picture.” It remains for future studies
to resolve the various loose ends and to assemble them into a full and
verifiable model of the mental process.
Models already exist, but must be refined as additional studies clarify
the mysteries of the brain.
[1] When total response time is measured, the presence of movement time (discussed later in this article) obscures the relatively strong correlation between reaction time and IQ.
[2] Jensen, A.R. (1980). Bias in mental testing. New York: Free Press. (a detailed history of chronometrics history)
[3] For a detailed discussion of Hick's Law, see Jensen, The Suppressed Relationship Between IQ and the Reaction Time Slope Parameter of the Hick Function, Intelligence 26 (1) 43-52
[4] Defined throughout the literature. For example Jensen, A.R. (1980). Bias in mental testing. New York: Free Press., P. 692.
[5] This discussion is drawn from many sources, which have been identified in the endnotes. A disproportionate amount of the material, however, comes from Jensen’s The g Factor. Information which is not othrwise noted can be found in that reference.
[6] Mary Smith of The University of Western Australia has reported a technique called Modified Blink Reflex, which behaves similarly to RT and IT tests and can be used for all ages, even infants. I made several attempts to contact her, but learned that she is no longer at the university and her present address was not known by the university staff.
[7] There is a large body of information pertaining to the correlation of average evoked potential features to IQ. All of these techniques (amplitude ratios, string length, and zero crossing points) also deal with low level factors which account for the variance in intelligence. Some of them are actually chronometric measures, but their discussion would require much more space than is appropriate for this article.
[8] Multiple response buttons are used for reaction time testing known as choice reaction time. There is an IQ correlation to even the most elementary RT testing, known as simple reaction time. Simple reaction time is measured by having the test subject release a button when he is presented with an external stimulus. Discrimination reaction time is a variant of simple reaction time which also uses one button, but requires the test subject to release the button only when the stimulus matches a predefined condition.
[9] Other researchers have used the same measurement devices. The topic of RT testing is discussed at length in numerous places. Bias in Mental Testing is now nearly a quarter of a century old, but has a good discussion. For a more recent reference, see The g Factor.
[10] Some psychometricians define RT as the total time from the start of the test to the end and divide that time into two components, designated “decision time”(DT) and MT. In this case, RT = DT + MT. Jensen designates RT as the first component and does not bother to discuss the sum.
[11] Jensen, Arthur R. (2000) Cognitive Components as Chronometric Probes to Brain Processes, Psycoloquy: 11,#11
[12] Somewhat higher and lower measurements have been reported. Jensen mentions one at high as .745 (Bias in Mental Testing, P. 229)
[13] T-scopes were produced as early as 1903, using gravity operated shutters.
[14] Brand references experimental work by Evans & Nettelbeck and Bates & Eysenck (all from 1993); see The g Factor: General Intelligence and Its Implications.
[15] Burns, R. R., and Nettelbeck, T. Inspection time in the structure of cognitive abilities: Where does IT fit? Intelligence 31, #4 (2003) 237-255
[16] For example, a refresh rate of 120 Hz offers presentation durations as multiples of 8.3 msec.
[17] Jensen, Arthur R. (2000) Processing Speed, Inspection Time, and Nerve Conduction Velocity , Psycoloquy: 11,#19
[18] Brand, C. (1996). The g Factor: General Intelligence and Its Implications. Chichester, England: Wiley. Note that Brand made this estimate before IT was found to have a spatial ability factor loading. Jensen (Psycoloquy 11 #19) estimated the IT to g correlation at -.50.
[19] It should be noted that Burns and Nettelbeck have disputed some of Brand’s claims, but their differences lie with the measurement of fluid intelligence as a group factor and whether IT correlations are as strongly correlated to g as claimed by Brand.
[20] See Burns and Nettelbeck, Intelligence 31, #4.
[21] DOUGLAS A. BORS and BERT FORRIN, Age, Speed of Information Processing, Recall, and Fluid Intelligence, Intelligence 20, 229-248 (1995).
[22] DOUGLAS A. BORS and BERT FORRIN. Age, Speed of Information Processing,Recall, and Fluid Intelligence, Intelligence 20, 229-248 (1995)
[23] 1 = g2 + s2 + e2, where e is random error.
[24] Raven’s Advanced Progressive Matrices
[25] Jensen, The g Factor, P. 234.
[26] There have been five studies cited by Jensen and Miller which show strong correlations between NCV and IQ, but four of these (which used peripheral NCV measurements) have been inconsistent. One study (Barrett, Daum, and Eysenck) showed no correlation to velocity but did find a -.44 correlation with variability from trial to trial.
[27] Edward M. Miller, Intelligence and Brain Myelination: A Hypothesis, Personality and Individual Differences, Vol 17, (December 1994) No. 6, 803-833. ----- Also discussed in Jensen, The g Factor, under Information Processing and g.
[28] Miller’s 1994 paper goes into great detail, which is beyond the scope of this discussion of chronometrics, as to why myelination is likely to relate directly to processing delays and ultimately to increased numbers of delays and finally to a cascading of errors, which effectively prevent the brain from further progress in its process.
[29] Jensen, Arthur R. (2000) IQ Tests, Psychometric and Chronometric G, and Achievement, Psycoloquy: 11,#14
"We have found, for example, that children of ages 12 to 13 who are enrolled in selective universities and are succeeding remarkably well in their course work, perform on the reaction time (RT) tests (which have no academic content) on a par with their university classmates, who average seven years older. But these gifted students have an average RT that is markedly faster, on average, than that of their age-mates, who are in the 7th or 8th grade. These academically gifted children also perform on a par with their university classmates on the SAT, vocabulary, and general knowledge, indicating that, because of their speed in information processing, they have acquired in 13 years a level of knowledge and skills which is commensurate with that acquired in 20 years by ordinary university students (whose IQs average about 120). What the chronometric tests tell us that ordinary IQ tests do not is that these gifted children differ from the average not only in the kinds of knowledge and skills typically associated with an advantaged environment, but also in some basic cognitive processes that favorably affect their powers of learning, retention, and comprehension of complex academic subject matter. Such chronometric tests would be especially useful with individuals for whom conventional IQ tests may be of doubtful validity because of their atypical background.
Hence "mental age" differences are reflected in content-free RT tests and in knowledge-based achievement tests. This applies to all kinds of knowledge, not just scholastic or academic knowledge. Of course, opportunity and interests govern the types of knowledge a person acquires, but g level is the stronger correlate of the amount of knowledge acquired."
[30] Nettelbeck, T. and Wilson C. The Flynn Effect: Smarter not faster, Intelligence 32, #14 (2004) 85-93
[31] Rushton, J. P. (1999) Secular gains in IQ not related to the g factor and inbreeding depression unlike Black-White differences: A reply to Flynn. Personality and Individual Differences, 26: 381-389.
[32] Bates, T. C. and Rock, A. Personality and information processing speed: Independent influences on intelligent performance, Intelligence 32 (2004) 33-46.
[33] Thinkfast is discussed in detail in the Prometheus Membership Committee Report (1998-1999) and its references. I have read all of the articles in the reference section and have found them to be interesting. Good points were made. There is one point, however, which I think must be kept in mind with respect to discussions of RT and IT related measurements: These measurements happen in a time-frame that is from 100 to 400 milliseconds faster than the minimum time for a person to be consciously aware of an external stimulus. Chronometric measurements are, therefore, not directly comparable to anything involving conscious thought (such as an IQ test, a conversation, or the solving of a math problem). That is not to say that the short time-frame measurements which are associated with most chronometric studies do not affect higher level brain processes most likely they do, via the mechanisms of enhancing the efficiency of working memory, reducing brain effort (as in energy expenditure), and in preventing transmission errors of the type which most likely are caused by neural noise.
[34] Although the speed of perception factor is independent of g, it is not necessarily independent of group factors (such as spatial ability) which typically appear in the factor analysis of IQ tests.
[35] Jensen, Arthur R.
(2000)
Processing Speed, Inspection Time, and Nerve Conduction Velocity , Psycoloquy:
11,#19 Intelligence
G Factor (32)
[36] Brand, C. (1996). The g Factor: General Intelligence and Its Implications. Chichester, England: Wiley