General and Specific Cognitive Abilities

Cognitive Abilities

• Specific cognitive abilities– E.g., verbal ability, spatial ability, memory,

speed of processing

• General cognitive ability (g)– Often used to be called “intelligence”

Hierarchical Models

• Very prevalent in cognitive sciences

• Work on the premise of interconnected levels

• Different “units” in each level– Specific units might interconnect within and

between levels

Hierarchy of Cognitive AbilityGeneral cognitive ability (g)

Specific cognitive ability

Individual tests

Interactions

• Hundreds of individual psychological tests used in assessment

• Moderate correlation between performance on different specific cognitive abilities– E.g., do well on spatial, probably do well on

memory

Correlations

• Not empirical; correlation is not causation

• Correlations can not tell why/how one factor relates to another, just the degree to which they do (or do not)

Genetic Regulation

• Not really much question that there is heritability involved in cognitive ability

• Specific gene and environmental control, however, is still pretty much unknown in humans

• Better understanding in nonhumans– Empirical testing can be conducted

Intelligence Testing

• Various intelligence (IQ) tests– Alfred Binet– Identify students needing special help– First test, 1905– Revised to Binet-Simon (1908, 1911), then

Stanford-Binet (1916)

Studies in Human Intelligence

• Early adoption studies (Burks 1928; Leahy 1935)– IQ correlates higher in nonadoptive families than in

adoptive families

• Adopted away children’s IQ correlates with their biological parents (Skodak & Skeels 1949)– This is increasingly true as child ages

• 1960 Louisville Twin Study, longitudinal study of environment and genetic effects begun

Heritability and Intelligence Correlations

• First degree relatives, ~0.45

• Adopted away children and biological parents, ~0.25

• Sibs adopted apart, ~0.25

• MZ, ~0.85

• DZ, ~0.6

• MZ raised apart, ~0.75

Couple Complications to This…

• Assortative mating

• Nonadditive genetic variance

Assortative Mating

• Non-random mating; when mates have similar features/characteristics

• Important for our discussion• Affects estimates of heritability• In first-degree relatives can inflate heritability

– E.g., sibs are more similar in trait because parents are similar for same trait

• In twins, though, can underestimate heritability– Raises DZ correlations because they’re 1st degree

relatives, so lessens difference b/t MZ and DZ twins

Nonadditive Genetic Variance• Additive genetic effects

– Alleles at locus and across loci “add up”

• Nonadditive effects– Effects of alleles different in presence of other alleles

• Dominance– Alleles at same locus interact – E.g., heterozygous phenotype different from homozygous dominant

phenotype

• Epistasis– Alleles at different loci interact to affect behaviour; phenotype of

different genes suppressed or expressed

• Emergenesis– Epistatic effects producing extraordinary effects; won’t be heritable

due to interactive nature

General Intelligence

• Charles Spearman– Schoolchildren’s grades across unrelated subjects

positively correlated– Proposed “general” intelligence– Initial interpretation that variation in intelligence due to:– Factor specific to an individual mental task– A general factor, g, that governs performance on all

cognitive tasks

• Ignored group factors, however… need factor analysis to identify this

g-Factor

• Is g real?

• What is the actual interaction between specific and general cognition?

• Correlations

g-Loading

• Tests of cognitive ability derive most of their validity from the extent to which they measure g

• g-loaded if quantifiable measure(s) of a task correlate highly with g

• Primary goal of IQ tests is to create reliable and valid tests; thus, the tests tend to be intentionally g-loaded

Non-specificity

• However, g not specific to any particular domain of knowledge or mental skill

• Also, seems independent of cultural content

• Support idea that g is real and not simply an artifact of particular opportunities to learn specific “skill sets”

Biological Correlates

• Brain size correlate with g, ~0.4

• Various brain wave activity and g, 0.5-0.7

• Speed of nerve conduction with g, ~0.4

• Even elementary cognitive tasks (ECTs) correlate with g (tasks like identify the colour of a light, number of figures on a page, etc.)

g

• g widely accepted• Seems to have moderate to high heritability• That said, less clear what g really is• Single general process?

– E.g., information processing speed, executive function

• Interaction/intersection of specific cognitive functions?

• Frequently, g used synonymously with “intelligence”

QTL and g

• Highly likely that many separate components contribute– Polygenic– Environment

• Effect at what level?– Elementary properties, specific cognitive

ability, general cognitive ability

Top down

• Genes act directly on g– E.g., perhaps through neural activity speed, etc.

General cognitive ability

Specific cognitive abilities

Elementary processes

Genes

Bottom Up

• Genes affect each basic element of information processing– Highly reductionistic model

General cognitive ability

Specific cognitive abilities

Elementary processes Genes

Multi-level Interaction

• Unique genetic effects at each level, but also genetic effects in common across levels

General cognitive ability

Specific cognitive abilities

Elementary processes Genes

Genes

Genes

Evidence

• Some support for top down

• Modularized view of brain function would fit well with bottom up

• However, multivariate genetic analysis supports multi-level interaction– Keep in mind, this model incorporates elements

from both top down and bottom up

Non-human Animal Models

• Can look for g-like abilities in non-humans

• Look for specific cognitive abilities that are directly comparable across species (e.g., spatial ability)

Maze Dull/Maze Bright• Tolman and Tyron• Selectively bred rats for ability in maze

learning• Maze bright rats showed few errors, maze

dull rats many errors after few generations

Generations1 22

Mea

n E

rror

s 20

0

Maze dull

Maze bright

Heritability for Learning

• Inbred strains of mice

Bovet et al. (1969)

Heritability in Learning

• In and of itself, not that novel, unexpected, surprising• But, environmental effects can come in…

Bovet et al. (1969)

Genotype-Environment Interaction

• Cooper & Zubek (1958)• Enriched, restricted,

standard lab conditions• Enriched improves MD,

not MB• Restricted detrimental to

MB, not MD

Popularity of Mice

• Mouse genome

• Can test for specific gene effects– E.g., transgenic critters

• Very useful for genotype-environment interactions with respect to cognitive abilities

• Obviously, more difficult in humans, but starting to get there

Caspi et al. (2007)

• Children’s intellectual development

• Interaction of genetic and environmental experience

• Breastfeeding

• IQ scores

Breastfeeding

• Long-chain polyunsaturated fatty acids (LC-PUFAs)– Present in human milk, absent in cow’s milk– Specifically, DHA (docosahexaenoic acid) and ARA

(arachidonic acid)– Required for efficient neurotransmission, neurite

outgrowth, dendritic arborization, and neuron regeneration post cell injury

• DHA and ARA accumulate in human brain in early postnatal months– Higher concentrations in breastfed than formula fed infants

Effect on IQ

• Breastfed children have higher IQs than non-breastfed children– Effect persists into adulthood

• Not due to SES or other culture-specific factors– Important to control for, as in Western

countries, higher SES is related to higher IQ, and higher SES women are more likely to breastfeed

Non-human Animal Models

• Animals deprived in n-3 fatty acids show neuronal deficits in memory, sensory, and visual abilities

• DHA supplementation in rodents and nonhuman primates increases DHA concentrations; enhances performance on learning, memory, and problem solving tasks

FADS2

• Chromosome 11 candidate gene• Role in modification of dietary fatty acids• Encodes delta-6 desaturase, the rate limiting step

on the metabolic pathway for ARA and DHA production

• Hypothesis: cognitive advantage of breastfeeding related to genetic differences in LC-PUFA metabolism, specifically at FADS2

Markers and Subjects

• Used two SNPs– Genetic polymorphisms rs174575 and rs1535

– Strong linkage disequilibrium through promoter and intragenic region of FADS2 (and also FADS1, another gene involved in fatty acid metabolism)

• Over 1000 New Zealand children born 1972-73, IQ measures at age 7, 9, 11, 13

• Over 2200 children from British twins born 1994-95; IQ measured at age 5

IQ Outcomes and Genotype110

105

100

95

90

Mea

n IQ

CC CG GG CC CG GG Genotypes

New Zealand Cohort

BritishCohort

Not breastfed Breastfed

Overall, breastfed children had IQ scores 5.6 and 6.3 points higher than non-breastfed children in New Zealand and British cohorts, respectively. About 90% either CC or CG.

Genotype and IQ

• Dominant effect of C allele in response to breastfeeding

• New Zealand: breastfed children with C allele showed 6.4 IQ-point advantage (p<0.001) compared to non-breastfed children; GG homozygotes gained no advantage from breastfeeding

• British: breastfed children with C allele showed 7.0 IQ-point advantage (p<0.001); GG had no advantage from breastfeeding

• Averaging, this equates to a 6.8 IQ point advantage, or 0.48 standard deviation units in the general population

rs174575

• Genetic moderation of breastfeeding effects on IQ not likely directly due to rs174575– Actual molecular mechanism of influence by

rs174575 is currently unknown

• May be that rs174575 influences biosynthesis of LC-PUFAs from dietary precursors, possibly through increased transcriptional activity

Application

• Earlier studies looking at neurodevelopment of infants fed DHA-supplemented vs. unsupplemented formula– Results inconclusive– Current research may offer explanation; genetic

heterogeneity in fatty acid metabolism may dilute supplemental effects

Application

• FADS2 locus has not appeared on the first genome-wide scans for intelligence

• Such scans identify genes with associations with phenotypes regardless of participants’ environments; ineffective for detecting genes whose effects are conditional on environmental exposure

• In contemporary Western samples, significant portion of population is not breastfed; this would conceal link between FADS2 variation and IQ

Heritability and Maturation

• Early twin studies investigated development (e.g., Galton, 1876; Merriman, 1924)

• Heredity increasingly important as you develop

Developing Twins

Why• New genes come into effect• Positive feedback effect

– IQ increase when adopted by parents with high IQ

• Intellectual experience more self-directed as an adult• Shared environment effects decrease with age

Genetic Contributions to Developmental Change

• g is pretty stable, not perfectly so… if change happens, it has a genetic aspect

• Genetic effects seem to act at transitional ages– Infancy to early childhood (e.g., language acquisition)

– Early to middle childhood (e.g., theory of mind)

– Etc.

Infancy Early childhood Middle childhood

Gen. factors New gen. factors New gen. factors

Shared Env.

Environment & Specific Cognition

• Scarr & Weinberg (1978)• Adoption study• Little similarity for adoptive parents and adopted

children or between adopted siblings on specific subsets of intelligence test… except vocabulary

• Like g, specific cognitive abilities also little influenced by shared environment (i.e., heritability significant factor)

Academic Performance

• Achievement vs. ability– Semantics?

• Shared environment ~60%, heritability ~30% (for 6-12 year range)

• Heritability effect does increase, and environment effect decreases with age

Heritability and Subjects

Twin CorrelationsSubject MZ DZ History .80 .51Reading .72 .57Writing .76 .50Arithmetic .81 .48

From Grade 7 Report Card Grades

Twin CorrelationsSubject MZ DZ Social .69 .52Natural Sciences .64 .45English use .72 .52Mathematics .71 .51

From High School Achievement Tests

School Achievement = g?

• Multivariate genetic analysis shows a common genetic effect explains much of the correlation between scores in different domains (i.e., subjects)

• Is this g, or some other measure?– Some-to-much of this is g, but some is achievement

specific

• Implies that achievement scores (within normal range) that are not due to ability are largely due to environment

Overall

• Variance in thirds• One third of genetic variance of academic

performance is in common with general cognitive ability

• One third of genetic variance is general to academic performance, independent of general cognitive ability

• One third is specific to each domain• Means learning abilities are not exactly the same

thing genetically as general cognitive ability