3. Causal Inference How do we test an hypothesis?

Transcription

1 Methods (cont d)

2 3. Causal Inference How do we test an hypothesis? 1. Choose the appropriate measurement 2. Gather data using some method 3. Use data yielded by the method to draw a conclusion What kind of data will allow you to draw a conclusion about causation?

3 3. Causal Inference Hypothesis: Tobacco smoke contains substances that are toxic to human tissue when deposited by contact.

4 Expected Correlations If smoking causes cancer, then you might find The longer a person has smoked, the greater the risk of cancer 2. The more cigarettes a person smokes over a given time period, the greater the risk of cancer 3. The greater time since quitting, the lower the risk of cancer. 4. Smokers cancers tend to occur in the lungs, and to be of a particular type 5. Smokers have elevated rates of other respiratory diseases 6. People who smoke cigars or pipes, the smoke usually not inhaled, have abnormally high rates of lip cancer 7. Smokers of filter-tipped cigarettes have somewhat lower cancer rates than do other cigarette smokers. 8. Nonsmokers who live with smokers have somewhat higher cancer rates

5 Expected Correlations Correlation coefficient (r) quantifies the strength of the relation between two variables. Takizawa, 2000 Caramori et al.,, 2009

6 Two Problems with Correlations 1) Ambiguous Direction of Causation 2) Potential Third Variable The longer a person has smoked, the greater the risk of cancer Cancer Risk Nervous Tension Smoking Nervous Tension Smoking High Rates of Disease

7 Causal Evidence Non-smokers Experimental designs random assignment experimental control Smoking Non-smoking High Rates of Disease Low Rates of Disease

8 Non-Experimental Designs No control, no pretest No control, pre-post Control, no random assignment

9 Experimental Designs Random Assignment ONLY experimental designs will allow you to draw a conclusion about causation

10 3. Analyzing the Data (Statistical Methods) Descriptive Statistics: What the data looks like Describing the Central Tendency & Variability Mean: average score Standard deviation: average difference between each score and the mean

11 Low Median Low Median High Median High Median # of Students Receiving Score # of Students Receiving Score Score Score Low SD Number of Students with Score High SD Number of Students with Score

12 3. Analyzing the Data (Statistical Methods) Descriptive Statistics: What the data looks like Describing a Correlation Correlation coefficients range from -1 to +1.

13 3. Analyzing the Data (Statistical Methods) Inferential Statistics: Is there a (non-chance) difference? One-group tests against chance ANOVA tests of group differences

14 Genetic Foundations of Mind & Behavior Honors Psychology

15 Two Main Questions The Classic Nature-Nurture Question: Are there biological reasons for why people are so different from one another? (Behavioral Genetics) The New Nature-Nurture Question: How is the nervous system built? (Developmental Neuroscience)

16 Two Main Questions The Classic Nature-Nurture Question: Are there biological reasons for why people are so different from one another? (Behavioral Genetics) NOT What matters more--the genes or the environment?

17 What makes a rectangle big? height or length? 6 9

18 What makes a rectangle big? Area = height X length! 6 9

19 What makes us X? nature or nurture? genes or environment?

20 Genetic and Environmental Forces Hereditary and environmental interactions are best illustrated by this model:

21 Genetic and Environmental Forces Hereditary and environmental interactions are best illustrated by this model:

22 A gene is a segment of DNA that affects development by specifying a protein template by regulating other genes Children get all of their genes from their parents.

23 Each parent contributes 50% of the child s genes, which has copies residing in the bodies of relatives.

24 Because chromosomes come in pairs (one from each parent), so do genes Some pairs of genes are identical (homozygous) and some are not (heterozygous)

25 The difference between homozygous and heterozygous gene pairs (alleles) has implications for how the child s genotype affects his/her phenotype

26 2 GENOTYPE/Child PHENOTYPE/Child

27 2 GENOTYPE/Child PHENOTYPE/Child Two ways children s genes affect their behavior One gene can control behavior (Mendelian inheritance) Many genes can control behavior (Polygenic inheritance)

28 2 GENOTYPE/Child PHENOTYPE/Child A third of genes have two or more different forms, known as alleles. Some physical traits, such as a negative Rhfactor in blood type, require matching recessive alleles, one from each parent, for expression. Others, such as a positive Rh-factor in blood type, require only the inheritance of one dominant allele, which will override a recessive allele from the other parent. When traits are controlled by a single allele, a Mendelian distribution will be observed

29 Mendelian distribution: Mendel s Peas

30 Mendelian distribution: Eumelanism

31 Mendelian distribution: Sickle-cell anemia

32 Mendelian distribution: Wizarding

33 Mendelian Psychological Traits Like the skin of peas, hair color, and wizarding, some psychological traits also appear to be controlled by a single gene Scott & Fuller (1965) found that all bansenjis were afraid of a novel person, whereas the cocker spaniels were seldom afraid. Was there a genetic link?

34 Mendelian Distribution: Fear Responses First, they crossbred pure cockers and pure basenjis. Some hybrids were raised by cockers; some by basenjis. All hybrids were fearful, like purebred basenjis They hypothesized that fearfulness was controlled by a singe dominant gene

35 Mendelian Distribution: Fear Responses To test their hypothesis, they crossbred the hybrids. They found the same pattern Mendel found with the peas. FF Ff

36 Mendelian Distribution: Fear Responses Finally, they backcrossed the original hybrids with purebred cockers. Half were afraid (like the basenji), and half were not (like the cockers), just as one would predict if a single fear-controlling gene were dominant.

37 Mendelian Distribution: PKU & SLI PKU Phenylketonuria: total language loss and mental retardation (prevented by monitoring the child s diet) SLI Specific Language Impairment has been linked to the FOXp2 gene

38 Polygenic Distribution Most traits and behaviors of psychological interest involve contributions by several genes, such as infant temperament shyness aggression risk-taking behavior empathy TV viewing

39 Polygenic Distribution When many individuals are tested for a polygenic characteristics, the results follow a normal distribution. What is this kind of figure called?

40 Population Genetics Population genetics attempts to find a role for genes by looking at differences between people and linking it to (genetic) family history ask how much variation in the behavior of a group is a function of genetic differences ( heritability, h 2 ) versus environmental differences If h2 = 1, all of the differences stem from genetic differences If h2 = 0, none of the differences stem from genetic differences If h2 =.5, half of the differences stem from genetic differences

41 Variation in Area World A World B Variation due to lengths Variation due to heights How much of the variation in area is a function of height versus width?

42 Variation in Behavior World A World B Variation due to lengths Variation due to heights How much of the variation in behavior is a function of genetic differences versus environmental differences?

43 Heritability Heritability In experimental organisms, such as fruit flies, it is possible to control the environment and thereby distinguish effects of genetic similarity vs. effects of environmental similarity.

44 Experiment to Establish Heritability Breeder s Equation: R / S = h 2

45 Tryon s Study of Maze Learning Tryon (1942) This chart depicts the progress Tryon made in selectively breeding rats for their ability to get through mazes after only a few errors. The critical step Tryon made was to cross-foster the rats (bright rats raised by dull rats, dull rats raised by bright ones). Regardless, offspring scores resembled those of their parents.

46 Estimating H 2 For estimating H2 in humans, researchers cannot perform experiments in selective breeding. Two common alternatives: (1)To estimate environmentality the proportion of variance due to environmental variation, compare identical twins raised apart (some environmental similarity) to identical twins raised together (more environmental similarity) (2)To estimate heritability the proportion of variance due to genetic variation, compare identical twins raised together (some environmental similarity) to fraternal twins raised together (same amount of environmental similarity)

47

48 Family Studies of IQ Relationship Correlation MZ, reared together 0.86 DZ, reared together 0.6 MZ, reared apart 0.72 unrelated, reared together 0 h 2 = (MZ - DZ) x 2 = ( ) x 2 =.52 e 2 = (MZrt - MZra) = =.14

49 Meta-analysis of the heritability of human traits based on fifty years of twin studies rdz Frequency h2 h2m h2f c2 c2ss ,10, Christiaan A de Leeuw1,3, Patrick F Sullivan4 6, Tinca J C Polderman1,10, Beben 0.5 Benyamin ,9,11 Arjen van Bochoven, Peter M 0.5 Visscher2,8, & Danielle 0.5 Posthuma 1.0 Correlation (r) rmz Age (years) d h2ss c2m c2f 65+ Despite a century of research on complex traits in humans, the relative importance and specific nature of the influences of rmz rmzm rmzf rdz rdzss rdzm rdzf r genes and environment on human traits remain controversial. We report a meta-analysis of DOS twin correlations and reported variance components for 1 7,804 traits from 2,748 publications including 1 4,558,903 partly dependent twin pairs, virtually all published twin studies of complex traits. Estimates of heritability cluster strongly within functional domains, and across all traits the reported heritability is 49%. For a majority (69%) of traits, the observed twin correlations are consistent with a simple and parsimonious model where twin resemblance is solely due to additive genetic variation. The data are inconsistent with substantial influences from shared environment or non-additive genetic variation. This study provides the most comprehensive analysis of the causes of individual differences in human traits thus far and will guide future gene-mapping efforts. All the results can be visualized using the MaTCH webtool. 7,804 traits from 2,748 publications including 14,558,903 twin pairs Insight into the nature of observed variation in human traits is impor- on classical twin studies, as the twin design has been used widely h2 and h2ssevolutionary h2m biology. h2f c2ss F tant in medicine, psychology, social sciences toc disentangle thecmrelativeccontributions of genes and environment, 0.8 It has gained new relevance with both the ability to map genes for across a variety of human traits. The classical twin design is based 0.6 human traits and the availability of large, collaborative data sets to do on contrasting the trait resemblance of monozygotic and dizygotic so on an extensive and comprehensive scale. Individual differences twin pairs. Monozygotic twins are genetically identical, and dizygotic 0.4 in human traits have been studied for more than a century, yet the twins are genetically full siblings. We show that, for a majority of traits 0.2 causes of variation in human traits remain uncertain and controver- (69%), the observed statistics are consistent with a simple and parsisial. Specifically, the partitioning of observed variability into underly- monious model where the observed variation is solely due to additive 0 ing genetic and environmental sources and the relative importance of genetic variation. The data are inconsistent with a substantial influence additive and non-additive genetic variation are continually debated1 5. from shared environment or non-additive genetic variation. We also Recent results from large-scale genome-wide association studies show that estimates of heritability cluster strongly within functional (GWAS) show that many genetic variants contribute to the variation domains, and across all traits the reported heritability is 49%. Our in complex traits and that effect sizes are typically small6,7. However, results are based on a meta-analysis of twin correlations and reported the sum of the variance explained by the detected variants is much variance components for 17,804 traits from 2,748 publications includsmaller than the reported heritability of the trait4,6 10. This missing ing 14,558,903 partly dependent twin pairs, virtually all twin studies of Figure 2 Twin correlations and for all human traitsthat studied. (a) Distribution of rmzpublished and rdz between estimates across the This traitsstudy investigated heritability hasheritabilities led some investigators to conclude non-additive complex traits 1958 and provides in 2,748 twin studiesvariation published between 1958 and r estimates are based on 9,568 traits and 2,563,628 partly dependent twin pairs; rdz 4,11 must be important. Although themzpresence of gene-gene the most comprehensive analysis of the causes of individual differences 5,12 17, little twin estimates are based on 5,220 2,606,252 partly dependent pairs (Table 1).traits (b) Relationship rmzgene-mapping and rdz, using all 5,185 traits interaction hastraits been and demonstrated empirically is known in human thus far and willbetween guide future efforts. All 18 for which both were reported. (c) Random-effects meta-analytic estimates of twin correlations (top) and reported variance components (bottom) across about its relative contribution to observed variation. results can be visualized with the accompanying MaTCH webtool. all traits separately for fourstudy, age cohorts. Error bars,first, standard errors. (d) Random-effects meta-analytic estimates of twin correlations (top) and reported In this our aim is twofold. we analyze empirical estimates of the relative genes functional and environment for forresults variance components (bottom) across contributions all traits, andofwithin domains which data on all correlations and variance components were virtually all human traits investigated in the past 50 years. Second, we The distribution of studied traits is nonrandom available. Error bars, standard errors. Social interactions Nutritional Social values Gastrointestinal Environment Psychiatric Endocrine Cardiovascular Activities Cognitive Neurological Respiratory Dermatological Ophthalmological Metabolic Ear, nose, throat Reproduction Skeletal All traits 1.0

50 ENVIRONMENT/Child PHENOTYPE/Child 3

51 ENVIRONMENT/Child PHENOTYPE/Child 3 The same genotype can be expressed differently in different environments. The norm of reaction is the range of all possible phenotypes in relation to all possible environments.

52 Family Studies of IQ Relationship Correlation MZ, reared together 0.86 DZ, reared together 0.6 MZ, reared apart 0.72 unrelated, reared together 0 H 2 IS NOT A CONSTANT! h 2 = (MZ - DZ) x 2 = ( ) x 2 =.52 e 2 = (MZrt - MZra) = =.14

53 Norm of reaction: Example #1 For the sake of argument, assume that depression (i.e., chronic feelings of hopelessness) is entirely the result of genetics

54 Who is most likely to be depressed? A C Mr. A s identical twin has a history of depression Mr. C s fraternal twin has a history of depression B D Mr. B s identical twin has no history of depression Mr. D s fraternal twin has no history of depression

55 Norm of reaction: Example #1 A C D B Kendler et al. (1995) Am J Psychiatry, 152,

56 Norm of reaction: Example #1 Influence of Life Stress on Depression: Moderation by a Polymorphism in the 5-HTT Gene Avshalom Caspi, 1,2 Karen Sugden, 1 Terrie E. Moffitt, 1,2 * Alan Taylor, 1 Ian W. Craig, 1 HonaLee Harrington, 2 Joseph McClay, 1 Jonathan Mill, 1 Judy Martin, 3 Antony Braithwaite, 4 Richie Poulton 3 s = allele for short serotonin transporter l = allele for long serotonin transporter

57 Norm of reaction: Example #1 Influence of Life Stress on Depression: Moderation by a Polymorphism in the 5-HTT Gene Avshalom Caspi, 1,2 Karen Sugden, 1 Terrie E. Moffitt, 1,2 * Alan Taylor, 1 Ian W. Craig, 1 HonaLee Harrington, 2 Joseph McClay, 1 Jonathan Mill, 1 Judy Martin, 3 Antony Braithwaite, 4 Richie Poulton 3 s = allele for short serotonin transporter l = allele for long serotonin transporter

58 Norm of reaction: Example #1 Influence of Life Stress on Depression: Moderation by a Polymorphism in the 5-HTT Gene Avshalom Caspi, 1,2 Karen Sugden, 1 Terrie E. Moffitt, 1,2 * Alan Taylor, 1 Ian W. Craig, 1 HonaLee Harrington, 2 Joseph McClay, 1 Jonathan Mill, 1 Judy Martin, 3 Antony Braithwaite, 4 Richie Poulton 3 s = allele for short serotonin transporter l = allele for long serotonin transporter

59 Norm of reaction: Example #1 Influence of Life Stress on Depression: Moderation by a Polymorphism in the 5-HTT Gene Avshalom Caspi, 1,2 Karen Sugden, 1 Terrie E. Moffitt, 1,2 * Alan Taylor, 1 Ian W. Craig, 1 HonaLee Harrington, 2 Joseph McClay, 1 Jonathan Mill, 1 Judy Martin, 3 Antony Braithwaite, 4 Richie Poulton 3 s = allele for short serotonin transporter l = allele for long serotonin transporter

60 Norm of reaction: Example #1 Influence of Life Stress on Depression: Moderation by a Polymorphism in the 5-HTT Gene Avshalom Caspi, 1,2 Karen Sugden, 1 Terrie E. Moffitt, 1,2 * Alan Taylor, 1 Ian W. Craig, 1 HonaLee Harrington, 2 Joseph McClay, 1 Jonathan Mill, 1 Judy Martin, 3 Antony Braithwaite, 4 Richie Poulton 3 s = allele for short serotonin transporter l = allele for long serotonin transporter

61 Norm of Reaction Even when there is a substantial genetic contribution to a psychological trait (e.g., depression), the amount of variation in the trait that is explained by genetic similarity (i.e., heritability) depends on the subjects environment (e.g., stressful events)

62 Norm of reaction: Example #2 Turkheimer et al. (2003) calculated IQ heritability for twins who differed in SES Sample median family income $22,000 (in 1997 dollars); 1997 US median $53,000

63 Norm of Reaction Does the norm of reaction imply that nature s role is somehow arbitrary, uncertain, or unreal? Consider PKU disorder In 100% of cases, the defective allele creates a defective enzyme that is from 0-50% effective as normal in breaking down phenylalanine In 100% of cases the build up of phenylalanine is poisonous to the brain and results in severe mental retardation Norm of reaction still applies: In an environment with NO phenylalanine, the defective gene won t make any difference in behavior In an environment WITH phenylalanine, the gene will make a huge difference in behavior

64 4. Child s Phenotype Child s Environment

65 4. Child s Phenotype Child s Environment Children are active sources of their own development in two ways: They actively evoke certain responses from others (e.g., calm, beautiful babies) They actively select surroundings and experiences conducive to their innate interests, talents, and personality characteristics. H 2 + E 2 < 1

66 Well, which is it...nature or nurture? Umm...both?

67 Three Laws of Behavioral Genetics 1) H 2 > 0: All traits are somewhat heritable. 2) H 2 > E 2 : Effect of shared genes is always greater than effect of shared environment. 3) H 2 + E 2 < 1: Shared genes and shared environment are not the only causes of individual differences in psychology.

68 Developmental Neuroscience So far we ve been talking about genes that vary from person to person, but genes shared by (virtually) all people do a lot of work recipe for bodies recipe for brains Developmental neuroscience is interested in the recipe (and cooking) of the nervous system