BIOLOGICAL CORRELATES OF IQ SCORES DO NOT NECESSARILY MEAN THAT G EXISTS
Book Review of Jensen on Intelligence-g-Factor
Nicholas R Burns
Institute of Advanced Studies
Australian National University
Canberra ACT
0200 Australia
Burns@rsbs.anu.edu.au
http://cvs.anu.edu.au/nick/home.htm
ABSTRACT: Jensen (1998) argues that biological correlates of IQ
scores establish the substantiveness of general intelligence (g).
This review critically examines three pieces of evidence Jensen
adduces to support this claim: event-related potentials, brain
nerve conduction velocity, and inspection time. It is argued that
Jensen's conclusions are premature and therefore unwarranted.
1. The force of many of Jensen's (1998) arguments on the real-world
significance of psychometric g derives from his claim that g is a
biological phenomenon. The reason that biological or information
processing correlates of IQ scores are important for Jensen's position
is that it is taken to be obvious that IQ differences are caused by
differences discovered in putatively more fundamental measures or
processes. In not one of the areas that I will discuss is the
conclusion on direction of causation obvious or warranted. The three
issues that I specifically comment on are correlations of cognitive
abilities with: cortical event-related potentials (ERPs; what Jensen
refers to as average evoked potential, AEP); brain nerve conduction
velocity (Jensen's, 1999, Precis points to this work as absolutely
demanding replication); and inspection time (IT). These will be dealt
with in turn.
2. Jensen (1998, pp.152-157) discusses four measures derived from ERPs
that have shown significant correlations with IQ. However, the
situation with each of these measures is that the relationships
reported in the literature do not provide strong evidence for the
status of g as a biological phenomenon. The first of these measures,
number of zero-crossings, has been convincingly dismissed by Verleger
(1999). The second, latency of ERP response, no doubt correlates with
various measures of cognitive ability at about 0.3. The interpretation
of this outcome is not clear-cut: such results merely support the view
that differences in physiological speed (to the extent that
physiological speed is manifested as ERP latencies) form a component of
cognitive abilities; our level of understanding of the bases of
individual differences in cognitive abilities has not been greatly
advanced. The third, amplitude of the ERP waveform, is discussed
selectively by Jensen. A more comprehensive treatment (Burns,
Nettelbeck & Cooper, 1997) identified 28 studies as having calculated
some comparison of ERP amplitude measure and intellectual ability; 17
studies reported a positive although not necessarily large or
statistically significant relationship and the remaining 11 studies
reported a negative although not necessarily large or statistically
significant relationship. While it could be argued that methodological
differences between studies account for the inconsistency in these
correlations between ERP amplitude measures and IQ, it is unlikely that
these inconsistent findings are a statistical artefact. The possible
confounding factors that can affect scalp recorded amplitude measures
(skull thickness and inter-electrode impedences to name but two) would
add to error variance and increase the possibility of Type II error.
While it is difficult to see how this would reverse any effect that
might exist, the inconsistency in the relationships is not surprising.
The fourth ERP measure Jensen (1998) mentions, waveform complexity
(i.e., the string measure), can be dealt with by noting that both a
proponent for g (Robinson, 1997) and the current author (Burns et al.,
1997) independently concluded on the basis of their empirical studies
that this measure was useless for understanding the mechanisms that
link ERPs and IQ scores. To summarise: such ERP correlates of IQ that
prove to be replicable, likely only latency measures, do not speak to
the fundamentality of g.
3. Turning now to relationships of cognitive abilities with measures of
brain nerve conduction velocity (i.e., the speed of axonal conduction
of action potentials). This discussion centres on Reed and Jensen's
(1992) use of ERPs to derive estimates of nerve conduction velocity
within the brain. The rationale for their experiment rested on the
assumption that, when studying the response of the visual system,
almost all of the latency between the retina and the cortex is axonal
conduction time. The visual ERP procedure adopted by Reed and Jensen
was chequerboard pattern reversal stimulation; this VERP consists of
two deflections, N70 and P100, with latencies of about 70 and 100 ms,
respectively. Two visual pathway NCV estimates were calculated by
dividing the subjects' head length by the relevant latency. While these
estimates were of subcortical NCV, it was argued that they should
correlate highly with cortical NCV. Reed and Jensen offered no argument
as to why this relationship should hold. It could be argued that NCV
would be a general property of the brain but much of the activity flow
in the cortex involves radial flow through columns of neurons (i.e.,
local circuits) rather than through longer distance cortico-cortical
connections. Therefore, the notionally measurable subcortical NCV may
not relate meaningfully to transmission in the cortex.
4. Visual ERP latency, I argue, cannot be considered to be mainly nerve
conduction time (as required by Reed and Jensen's, 1992, rationale).
The implication of this argument is that it is not nerve conduction
velocity that is being estimated by their procedure. Rather, it is a
measure of VERP latency corrected for head size, this interpretation
being consistent with the fact that evaluations of clinical data on
pattern reversal ERPs recommend the adoption of different normal ranges
of latencies for males and females because of gender differences in
head size. As such, the findings of Reed and Jensen should be
interpreted in light of Paragraph 2, above. Furthermore, there are
other reasons why the use of visual ERP latencies to estimate visual
pathway NCV is not straightforward (which is not to say that it is
impossible).
5. The first reason is based on a consideration of some functional
properties of the visual pathways. The estimation of visual pathway NCV
from visual ERP latency requires the assumption that the pathway from
retina to visual cortex is a unidirectional one. That is, the effect of
a stimulus impinging on the retina is taken to be the propagation of
nerve impulses along the visual pathway to the visual cortex from
whence they are passed on for higher processing. However, there is
evidence that visual processing is not so linear. For example, there
are more projections back from the visual cortex to the thalamus than
there are forward from the thalamus to the visual cortex and there is
evidence that these projections exert feedback control from the visual
cortex on activity in the lateral geniculate bodies. Such a
consideration should not affect the very first response of the cortex
to the afferent input but it may affect the later responses (i.e.,
P100). Furthermore, knowledge of the interconnectedness of areas within
the visual cortex and of extrastriate input to the visual cortex
modifies the simple notion of information passing from the visual
cortex to higher processing stations. It is plausible that these
projections act as modulators of activity within the visual cortex and
visual pathway and that this modulation reflects the effects of
activity within the visual pathway itself. These types of
considerations mean that there is no simple beginning or end point in
brain processing. In other words, the assumption that ERP latency
solely reflects NCV may be an oversimplification because the delay from
retina to visual cortex may not be a constant which is solely dependent
on axonal conduction of action potentials.
6. The second reason is methodological. The chequerboard pattern
reversal ERP and the cortical origins and characteristics of the
deflections recorded are not as simple as the description given by Reed
and Jensen (1992) implied. Factors that influence pattern reversal ERP
latency are well known and include the luminance of the stimulus,
cheque size, and sex and age of subject. Moreover, the shape of the
primary visual cortex is known to differ from individual to individual
and this can lead to problems in obtaining standardised
single-electrode recordings. These factors may not have directly
affected Reed and Jensen's results but if they had used differently
configured stimuli then their estimates of NCV would have been
different. They also treated retinal transduction time as a constant;
the validity of this is an open question.
7. The third reason is anatomical. The visual pathway consists of the
retinal receptors and ganglionic cells, the optic nerves (which unite
and cross at the optic chiasm then proceed as the optic tracts), the
lateral geniculate bodies of the thalamus (there are also projections
to the superior colliculus), and the optic radiations (which terminate
in the visual cortex). At even the most basic level of analysis, this
complexity renders simple interpretation of visual ERP latency as NCV
problematic. This is because structurally, the optic nerve, the visual
tract and the optic radiation vary in terms of the diameter of the
axons involved. This difference means that axonal conduction velocity
will vary along the pathway. Conduction velocity will be fastest in the
optic nerve and tract which are of larger diameter and will be slowest
in the optic radiation which is of smaller diameter. Whether the ratios
of these diameters is constant across individuals is not known.
8. The final issue I wish to comment on is the correlations of
cognitive abilities measures with inspection time (IT). The IT task was
designed to measure the periodicity of an hypothesised sampling
mechanism of the brain. Latterly, it has been claimed to measure the
speed with which the brain transduces simple stimuli (speed of
apperception). Jensen's (1998) treatment of this area trivialises the
problems of interpretation that exist. Nothing is known of what IT is
measuring and little is known of what established relationships mean.
Briefly, the pattern of relationships has been that IT correlates most
strongly with performance IQ rather than with verbal IQ from the
Wechsler scales, and with tests of g such as Ravens matrices. One
interpretation of this pattern of relationships has been that
performance IQ and matrices tests measure fluid ability and therefore
IT provides an index of the biological substrate of g. An alternative
interpretation is that IT shares variance only with measures of general
processing speed and general visualisation ability. Moreover, there are
suggestions that IT indexes even more specific abilities such as the
ability to detect motion in noise. It seems reasonable to suppose that
IT is but one of a series of similar tasks (i.e., tasks incorporating a
backward masking procedure) which together would define a lower-order
factor within a hierarchy of abilities that define general speed of
processing. Jensen appears unaware of the complexities of the topic;
the interested reader is referred to White (1993, 1996), Burns,
Nettelbeck & White (1998) and Burns, Nettelbeck & Cooper (1999).
Mackintosh (1998) also discusses some of these issues; his discussion
accords with my contention that if we are to use psychometric test
scores as criterion variables, which is what we do when we correlate IT
with IQ, then we should be using a multi-dimensional model of
intelligence. Such a contention does not fit well with Jensen's
unidimensional description of human abilities.
9. It is extremely doubtful that the non-specialist would be able to
glean from reading Jensen (1998) that much of the evidence he adduces
is highly controversial. While the whole appears comprehensive and
consistent, many of his conclusions are unwarranted. More particularly,
since much of his argument on heritability and genetic determination of
g rests to a degree on evidence the like of which has been dissected
here, strong conclusions on anything, let alone the basis of group
differences in IQ scores, are premature.
REFERENCES
Burns, N.R., Nettelbeck, T & Cooper, C.J. (1997). The string measure of
the ERP: What does it measure? International Journal of
Psychophysiology, 27, 43-53.
Burns, N.R., Nettelbeck, T. & Cooper, C.J. (1999). Inspection time
correlates with general speed of processing but not with fluid
ability. Intelligence, 27, 37-44.
Burns, N.R., Nettelbeck, T. & White, M. (1998). Testing the
interpretation of inspection time as a measure of speed of sensory
processing. Personality and Individual Differences, 24, 25-39.
Jensen, A. (1999) Precis of: "The G Factor: The Science of Mental
Ability" PSYCOLOQUY 10(23).
ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/
psyc.99.10.023.intelligence-g-factor.1.jensen
http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?10.023
Jensen, A. (1998) The G Factor: The Science of Mental Ability. Praeger
Mackintosh, N.J. (1998). IQ and human intelligence. Oxford: Oxford
University Press.
Reed, T.E. & Jensen, A.R. (1992). Conduction-velocity in a brain nerve
pathway of normal adults correlates with intelligence level.
Intelligence, 16, 259-272.
Robinson, D.L. (1997). A test of the Hendrickson postulate that reduced
EEG response variance causes increased AEP contour length: Implications
for the neural transmission errors theory of intelligence. Personality
and Individual Differences, 22, 173-182.
Verleger, R. (1999). The g factor and event-related EEG potentials.
PSYCOLOQUY 10(39).
ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/
psycoloquy.99.10.039.intelligence-g-factor.2.verleger
http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?10.039
White, M. (1993). The inspection time rationale fails to demonstrate
that inspection time is a measure of the speed of post-sensory
processing. Personality and Individual Differences, 15, 185-198.
White, M. (1996). Interpreting inspection time as a measure of the
speed of sensory processing. Personality and Individual Differences,
20, 351-363.
This archive was generated by hypermail 2b29 : Tue Jan 11 2000 - 14:04:10 PST