You are on page 1of 5

CHAPTER 4

Contemporary Models of Intelligence

Janet E. Davidson and Iris A. Kemp

Few constructs are as mysterious and contro- and social (Eysenck, 1988; Flynn, 2007).
versial as human intelligence. One mystery Each level has its own organizing concepts,
is why, even though the concept has existed hypotheses, research methodologies, and
for centuries, there is still little consensus conclusions that can limit comparison and
on exactly what it means for someone to consensus. For example, the physiological
be intelligent or for one person to be more approach typically employs advanced tech-
intelligent than another. Oddly enough, the nology to examine indices of intelligence
heterogeneity among views of intelligence in the brain, whereas the social (or soci-
seems to have increased over time rather etal usefulness) approach uses performance
than decreased (Stanovich, 2009). This lack on “real-world” tasks to study intellectual
of agreement fuels unresolved controversies, skills in context. Fortunately, there has been
such as whether intelligence is comprised of some recent cross-fertilization between lev-
one main component or many, and it results els, which bodes well for future agreement
in claims that intelligence is too imprecise a on what it means to be intelligent (Flynn,
term to be useful (Jensen, 1998). A related 2007).
mystery is why the field has generated rela- Why are the mysteries surrounding intel-
tively few new models of intelligence in the ligence important ones to solve? Even
past 20 years. Is this scarcity due to a per- though the construct is difficult to define,
ceived futility? Will it eventually result in assess, and explain, the goal is a worthy one.
the field’s demise? Or has scientific progress If humans continue to live among each other
been sufficient enough to make the pursuit and differ in their abilities to learn and adapt,
of new directions unnecessary? the concept of intelligence is going to endure
The existence of the first mystery socially and scientifically. Fully and indis-
is understandable and perhaps inevitable, putably understanding this elusive construct
given that intelligence is currently defined, means that cultures can fairly identify and
assessed, and studied on at least three dif- cultivate it (Nisbett, 2009). Scientific knowl-
ferent levels: psychometric, physiological, edge about the workings of the human mind

58
64 JANET E. DAVIDSON AND IRIS A. KEMP

shortcomings. The first has to do with our does not account for all individual differ-
criterion that models be based on relevant ences in intellectual performance and little
assumptions. It is not clear that psychomet- consensus has been reached on g’s meaning,
ric theories meet this requirement; they rest it seems unlikely that correlations between
on the supposition that analyses of scores, g and scores on mental ability tests will ever
from tests taken once, reveal the true struc- capture the full story of intelligence. This
ture of intelligence. Test taking occupies a point brings us to the next section on the
relatively small part of most people’s lives physiological approach to intelligence.
and it does not necessarily reflect their intel-
ligent behavior in daily problem-solving sit-
uations. Even though the scores are moder- The Physiological Level and its Models
ately predictive of school achievement and
work success (Flynn, 2007), they fall short Everyone we have met believes that the
of capturing many aspects of what is con- brain plays a central role in intelligence, and
sidered intelligence. For example, as Horn no one we have met knows exactly what
and Blankson (2005) note, standard tests of this role entails. Fortunately, this lack of
Gc do not measure the depth of knowl- knowledge is likely to change because of
edge and reasoning required for expertise the physiological level’s focus on the rela-
in a domain. Mental ability tests will prob- tionship between brain activity and mental
ably always exist and we are not advocat- ability. The primary goal of this level is to
ing their demise. However, it might be too determine the neural basis of intelligence.
much to assume that they can tell us all that Recent theories, hypotheses, and empirical
we would like to know about the structure results related to this goal will be reviewed
of intelligence. in this section.
The second shortcoming has to do with
the criterion that models should contain
Brain Efficiency and the Parieto-Frontal
only relevant and comprehensible compo-
Integration Theory (P-FIT )
nents. Unfortunately, g and its role in intel-
ligence are not well understood. For exam- The parieto-frontal integration theory iden-
ple, the Gf-Gc theory does not propose g tifies a network of discrete brain regions
as a latent source of individual differences related to individual differences in general
in intelligence, while Carroll’s three-stratum intelligence and reasoning (Jung & Haier,
theory does. Partly because of these hierar- 2007). As the theory’s name implies, these
chical models, g remains a controversial and areas are primarily located in the parietal and
pervasive issue for contemporary theories of frontal lobes, and one of their main functions
intelligence. is to integrate information among various
At one time we thought that the mean- parts of the brain. Many of the P-FIT regions
ing of g needed to be resolved before intel- are related to basic cognitive processes, such
ligence could fully be understood (David- as attention and working memory. In other
son & Downing, 2000). Perhaps it is time words, the attributes of general intelligence
to consider that this might never occur. Is g are not associated with one central part of
a useful construct if there is never consen- the brain but with a network of structures
sus on what it represents? Earlier versions and functions distributed throughout the
of the CHC model omitted g because of its cortex. According to Jung and Haier’s the-
irrelevance to the development, interpreta- ory, highly intelligent people have cortical
tion, selection, and revision of intelligence networks that operate more accurately and
tests. In contrast, patterns of broad and nar- quickly than those of less intelligent individ-
row abilities are relevant (McGrew & Flana- uals.
gan, 1998) and some of these abilities explain The argument for brain efficiency is not
school achievement beyond the effects of g new. Studies using positron emission tomog-
(McGrew, 2009). Given that a single factor raphy (PET) found that individuals who
CONTEMPORARY MODELS OF INTELLIGENCE 65

obtained high IQ scores had brains that P-FIT accounts for a range of empir-
expended less energy, and consequently ical findings on individual differences in
consumed less glucose, than the brains of intelligence and reasoning (e.g., Colom
individuals with lower IQ scores (Haier et al., 2009; Jung & Haier, 2007; Schmithorst,
et al., 1988). Similarly, research employ- 2009), However, the theory is not without
ing electroencephalography (EEG) mapping its critics. For example, several researchers
methods discovered that highly intelligent (Blair, 2007; Lee, Choi, & Gray, 2007; Ror-
participants exhibited more focused corti- ing, Nandagopal, & Ericsson, 2007) claim
cal activation, and less overall brain activa- that the P-FIT network focuses primarily
tion, than did their lower ability counter- on fluid intelligence and working memory
parts (Neubauer & Fink, 2005). P-FIT builds rather than on the broader construct of intel-
on this earlier work and extends the neural ligence.
efficiency hypothesis by specifying where in Developmental perspectives. It is not
the cortex this neural efficiency occurs. yet clear how P-FIT addresses systematic
More specifically, P-FIT is based on con- changes in intelligence across the life span.
verging evidence from 37 cognitive neu- In their comparison of P-FIT with a model
roimaging studies that varied in their oper- of cognitive development, Demetriou and
ational definitions of intelligence and their Mouyi (2007) found areas of agreement and a
methods of assessing it (Jung & Haier, 2007). few shortcomings. As Jung and Haier (2007)
Despite procedural differences, there was note, more empirical work and revision of
reassuring consistency across studies in the P-FIT need to occur to account for develop-
brain regions associated with individual dif- ment.
ferences in performance on general intel- Applications. After extensive testing and
ligence and reasoning tasks. The underly- modification, P-FIT will most likely have
ing theoretical assumptions tying the data practical implications for societal issues.
together are that (a) regions within the According to Jung and Haier (2007), for
occipital and temporal lobes help humans example, the model might eventually be
begin processing relevant visual and audi- useful in developing treatments for men-
tory information from their environments; tal retardation and other neurological con-
(b) the results from this early sensory pro- ditions.
cessing are sent to areas in the parietal cor-
tex for more in-depth processing; (c) the
The Neural Plasticity Model of Intelligence
parietal cortex then interacts with regions
in the frontal cortex that perform hypothe- The ability to adapt to a wide range of cir-
sis testing on solutions to a known problem; cumstances is central to many definitions of
(d) after an optimal solution is reached, the intelligence (Binet & Simon, 1916; Neisser
anterior cingulate constrains response selec- et al., 1996; Sternberg, 1985). Dennis Gar-
tion and inhibits competing responses; and lick’s (2002, 2003) neural plasticity model
(e) the underlying white matter facilitates of intelligence imports adaptability to the
efficient transmission of data from the poste- physiological level. According to this model,
rior to frontal regions of the brain. According intelligent individuals have brains that pro-
to Jung and Haier (2007), regions of the brain ductively change in response to their sur-
that are not part of the P-FIT network con- rounding environments.
tribute minimally to individual differences A great deal of empirical research has
in intelligence; their role is to ensure the shown that neural plasticity allows synap-
reliability of basic brain functions common tic connections between neurons to develop,
to all humans. In contrast, regions within change, and reorganize in response to
the P-FIT network set no limits on potential environmental stimulation (Hebb, 1949;
variations between individuals and can dif- Rosenzweig, 2003). For example, enlarged
fer in terms of their blood flow, volume, and hippocampi were commonly found in
chemical composition. London taxi drivers, who heavily relied on
66 JANET E. DAVIDSON AND IRIS A. KEMP

this area of their brains to navigate the city intelligence tests (Newman & Just, 2005;
(Maguire et al., 2000). In short, the environ- Stankov, 2005).
ment shapes specialized neural connections Recently, Eduardo Mercado III (2008,
that are required for different cognitive abil- 2009) refined the neural plasticity model of
ities (Garlick, 2002, 2003). intelligence by focusing on cortical mod-
On the surface, the plasticity and special- ules. In short, these modules are specific,
ization of neural connections in response to vertical columns of interconnected neurons
environmental stimuli implies that a highly located in different areas of the cerebral cor-
genetic general factor of intelligence (g) does tex. According to Mercado, the capacity to
not underlie all mental activities. Instead, learn (i.e., cognitive plasticity) is directly
individual differences in intelligence would related to the availability, reconfigurability,
be due to individual differences in envi- and customizability of the cortical modules.
ronments and to the specialized synap- In other words, the neural modules and their
tic connections these environments create. flexibility provide the structural basis for
However, through the use of computer acquiring knowledge and improving skills.
simulations and neurophysiological data, Individual differences in intelligence are a
Garlick (2002) demonstrates that some product of the number and diversity of avail-
human brains may be more “plastic” than able cortical modules.
others and, therefore, better able to adapt Developmental perspective. According to
to a range of circumstances. According to Garlick (2002), intellectual development and
Garlick, this capacity for neural adaptation is its time frame are due to a “long-term pro-
dependent, in large part, on a variety of neu- cess whereby the brain gradually alters its
ral substrates encoded in the genes. More- connections to allow for the processing of
over, the brain’s overall ability for neural more complex environmental stimuli” (p.
plasticity would be reflected in a general fac- 120). In addition, he emphasizes critical peri-
tor of intelligence. ods for neural plasticity in different regions
Garlick’s model also explains individ- of the brain. These periods influence the
ual differences in neural efficiency. Indi- development of intelligence. Fortunately,
viduals who have neural networks that are some plasticity has been found to occur
shaped and organized to fit a variety of task throughout the life span (Kaas, 1991).
demands are better able to process informa- Applications. Models of neural plasticity
tion quickly and accurately. In addition, only highlight the importance of being exposed
task-appropriate regions of their brains are to stimulating environments. According to
activated, which limits the amount of glu- Mercado (2009), research on the relation-
cose needing to be metabolized. ship between cognitive and neural plasticity
Two recent theoretical views are related has relevant implications for education and
in many respects to Garlick’s model of other societal practices.
neural plasticity. The first explains fluid
intelligence as the product of a flexible,
Critique of the Physiological Level and
adaptive neural system. More specifically,
Its Model
Newman and Just (2005) propose that intel-
ligent individuals have dynamic neural net- The physiological level and its models are
works that alter their composition in order appealing for a variety of reasons. From
to accommodate task demands, and corti- a scientific standpoint, this approach pro-
cal regions that work in synchrony to per- vides a potentially uncomplicated, parsimo-
form a specific function. In support of this nious view of intelligence as a biological
theory, results from neuroimaging studies phenomenon. Furthermore, recent advances
have found that neural synchrony becomes in neuroimaging techniques make it possi-
more precise when tasks become more diffi- ble to examine the brain regions associated
cult. In addition, this synchrony is positively with intelligence, reducing the need to make
related to task performance and scores on inferences about the brain from behavioral
CONTEMPORARY MODELS OF INTELLIGENCE 67

measures. From a practical standpoint, neu- Finally, the physiological models are not
rological measurements provide a glimmer yet fully explanatory. The mechanisms caus-
of hope for future “culture-fair” measure- ing neural efficiency and neural plasticity in
ment of intelligence. For example, physi- the brain still need to be established. Simi-
ological measures are less likely to penal- larly, the direction of causality is not known.
ize individuals for poor test-taking skills. For example, it is tempting to conclude that
Similarly, understanding the relationship brain efficiency is the underlying cause of
between the brain and intelligent behavior high intelligence. However, as implied by
could result in interventions and treatments the work on neural plasticity, some neuro-
that foster both brain development and cog- logical responses may be reacting to behav-
nitive abilities. ioral responses rather than causing them.
Unfortunately, fully understanding the Another possibility is that neurological func-
neural basis of intelligence will probably tions and cognitive performance are reflec-
not occur any time soon. Even though tions of some other aspect of physiological
the physiological models meet our cri- or psychological functioning that has yet to
teria of building on previous knowledge be discovered. Unfortunately, correlational
and generating new research that advances studies cannot explain causation. Different
the field, they are faced with some types of experiments will need to clarify the
difficult problems. One methodological relationship between the brain’s activity and
concern involves the inconsistency of neu- an individual’s intelligent behavior.
roimaging results across studies. For exam- In short, the physiological models have
ple, not all empirical results support the shortcomings but tremendous heuristic
neural efficiency hypothesis. Rypma and value. Current empirical support is primar-
Prabhakaran (2009) propose that replication ily positive and the physiological approach
failures are due to differences in cognitive will undoubtedly continue to generate a
tasks and analysis techniques. They propose great deal of intriguing research.
that neuroimaging studies need to separate
individual differences in processing speed
from individual variations in processing The Social Level and Its Models
capacity.
Another common problem for the phys- Our third approach focuses on the social
iological models is that the empirical sup- usefulness of intelligence and takes into
port tends to be based on the question- account individuals’ functional abilities and
able assumption that intelligence quotients skills that make significant contributions to
(IQ) and related tests are sufficient their societies (Flynn, 2007). Consequently,
standards of comparison for the physio- the resulting models view intelligence as a
logical measurements. As noted in this complex dynamic system involving interac-
chapter and elsewhere (Gardner, 1983; tions between mental processes, contextual
Kaufman, 2009; Sternberg, 1985), there is influences, and multiple abilities that may or
persuasive evidence that IQ is an incom- may not be recognized in an academic set-
plete measure of intelligence. Dempster ting. Although the following three models
(1991) and Kaufman (2009), for example, have been in existence for some time, their
note that the ability appropriately to resist recent applications, additions, and clarifica-
task-irrelevant information plays a crucial tions merit inclusion in this chapter.
role in intelligence that is frequently over-
looked on most standardized tests. Further-
The Triarchic Theory of Successful
more, extensive work still needs to be done
Intelligence and Beyond
cross-culturally to determine whether the
relationship between performance on the Robert Sternberg’s theories have an
neurological measures and the tasks of intel- admirable history of building upon them-
ligence is universal. selves. His componential theory (Sternberg,

You might also like