development and interpersonal relationships fall 26' - unit 1

Monadic Bias

psychology’s historical habit of studying individuals rather than relationships (dyads). Sears (1951) argued for dyadic research. Bersheid (1999) says psychology is individualistic in two ways: it looks for laws governing a single person, and it looks for causes inside the individual (traits, genes, attitudes)

What does a relationship study, according to Berscheid?

The “oscillating rhythm of influence” in two people’s interactions. The relationship doesn’t reside in the individual, it is inherently temporal, and it is invisible (you see it only through its effects)

Three reasons to study relationships

conceptual (relationships are “natural categories”; need to belong)

functional (what relationships do for development and health)

empirical (“surprise” findings where individual phenomena can’t be explained without relationships)

Conceptual reason: Sedikides et al. (1993)

Memory for information about an individual is tied to couple hood. People more often correctly matched information to people known to be in a couple, and more often confused information about the two members of a couple. So pairs are a “natural” social category

“First Memories” data

Roughly 80% of first memories involved a relationship, and about 20% did not. Parents were the most common relationship (about 38), then siblings (about 17). Supports relationships as central to development

Need to belong (Baumeister & Leary)

A drive for frequent, emotionally positive interactions with at least a few people, within relationships marked by “persistent caring”

Evidence for a belongingness need

Humans form relationships easily, prefer liking to disliking, show distress and protest at the end of relationships, and process information consistent with relationship concerns, and feel positive emotion when belonging increases and negative when it decreases, and substitute new relationships for unfulfilling ones

Functional reason: health and mortality

Mortality and morbidity vary with relationship status. Risk is linked to smaller networks, infrequent contact, and disrupted relationships (death, divorce, enforced separation). House et al.: low social integration is a risk factor with an age-adjusted relative risk exceeding that of cigarette smoking. Example: the Donner Party

Caveat on the Mortality/morbidity findings

They are correlational, and the mechanisms aren’t known. Reis et al. propose three pathways:

direct biological effects

health-related behaviors

coping/appraisal

Social isolation: Thompson & Melzack (1956)

Scottish terrier puppies were raised in opaque cages. At 7-10 months, the isolated pups were more wary of other dogs and less dominant when competing for a bone, and more agitated by novel stimuli

Harlow’s isolate monkeys

At 3 months, they were terrified, stereotyped, and soon recovered. At 6 months, they avoided peers and were victimized. At 12 months, they were withdrawn and apathetic. Harlow first thought 6 months was a critical period, but Harlow and Suomi’s younger-peer therapy (3-month-old “therapist” monkeys) showed slow recovery even for 12-month isolates

Privation vs. deprivation; human isolation evidence

Orphanage research: Goldfarb (post-Depression American orphanages) found children institutionalized past age 3 were socially immature, had lower IQs, were more dependent, and had more behavioral problems. Romanian orphanages: late-adopted children can still form attachments but have more insecurity. Most adopted children are not at risk. Variability is the most pervavsive outcome, and timing matters

Why is deprivation harmful? Can children recover?

Two hypotheses: maternal deprivation (no primary attachment figure) and social stimulation (no synchronous, contingent interaction). Recovery is possible, especially for non-abused, less deprived children placed with edicated, affluent, loving, responsive parents

Empirical reason: Landauer et al. (child compliance)

Three mothers and their preschoolers were paired in all possible combinations. Child disobedience was greater with their own mothers and was uncorrelated across the adult partners. Compliance is a property of the relationship, not just the child

Empirical reason: Miller & Norman (1975) (from Berscheid)

Participants who actually interacted (prisoner’s dilemma) attributed more responsibility and disposition to themselves than passive observers did, and had stronger affect. This reversed the predicted actor-observer pattern, so even a minimal relationship changed the result

Why lab findings may not transfer (Berscheid)

Lab encounters usually involve strangers with no prospect of future interaction and little emotional import. Real relationships have both, and emotion is experienced most often and most intensely in close relationships

The scientific method

a value system or attitude about what counts as acceptable information: theory, hypotheses, observations

Reliability (2 types)

Consistency. Inter-rater (do different raters agree?)

Test-retest: (stable across time?)

Validity (types)

Accuracy. Convergent (concurrent or predictive; agrees with measures it should)

Discriminant (doesn’t correlate with what it shouldn’t). Also ecological validity: does it generalize to natural settings?

Self-report measures

Example: PREPARE, attachment style measures like Fraley’s ECR-R (avoidance and anxiety). Strengths: quick, standardized format.

Limitations: presentation biases, developmental change

Clinical interview (example: adult attachment interview)

Semi-structured, standard probes with contingent follow-ups.

Strengths: flexibility, depth

Limitations: information may not generalize across participants, and interviewer effects (“leading the witness”). The AAI infers a “working model of relationships” (inferred experience scales and coherence of mind) from transcripts

Observational measures

Naturalistic or structured

Examples: preschool observation. Markman-Cox marital task, strange situation

Strengths: direct behavior, ecological validity if naturalistic

limitations: low base rate behaviors, time sampling, reactivity, and weaker ecological validity if structured

Coding approaches

Microanalytic (FACS: 44 action units coded second by second). Macro/clinical (global ratings of relationship quality

Physiological/biological measures

Cortisol, autonomic measures (EKG, SCL, blood pressure, respiration), EEG, DNA.

Strengths: relatively objective, can triangulate

Limitations: time-consuming, “methodological sanctification,” need psychological theory to interpret.

Example: Caspi’s MAOA finding

Correlations cs. experimental designs

Correlational: lets you study things you can’t ethically randomize and describe natural history, but correlation does not imply causation

Experimental: the only way to establish causation, and randomization reduces confounds, but ecological validity suffers (Bronfenbrenner’s “science of strange behavior in strange situations with strange adults”).

Example: Liebert & Baron (1972) violent clip vs. track meet, with children less helpful after the violent clip

Cross-sectional, longitudinal, cross-sequential

Cross-sectional: different ages at one time

Longitudinal: same couples across time

Cross-sequential: multiple cohorts across time, separating developmental effects from cohort effects

Triangulation

The “truth” emerges from multiple methods and informants that converge on a coherent (not necessarily identical) story.

Example: informants agree only modestly (r about.30), more for externalizing than internalizing, more among informants who share a relationship with the target

Adult romantic relationship study (multi-method)

Samples: dating, engaged, older married, gay, and lesbian couples.

Summary: dating couples were satisfied; committed couples (especially same-sex female) were effective in observations; and committed couples (especially married men) showed the least autonomic reactivity

Non-independence: the problem

Statistics assume independent observations, but dyadic data are non-independent by definition. Violations can inflate Type I error or produce Type II error. ICC for negative emotion across partners was .36 and for positive emotion

Non-independence: solutions

Use the “best” informant, use an average score, use different scores, or use one partner’s data as an “independent” predictor (problematic in the indistinguishable case). The best-developed solution is the APIM

APIM (actor-partner interdependence model)

Uses hierarchical linear modeling to estimate actor effects and partner effects within non-independent units (dyads).

Actor effect: a person’s own predictor related to their own outcome

Partner effect: the partner’s predictor relates to the person’s outcome

Interdependence: four defining dimensions

Frequency of interaction (high)

Strengths of impact (high)

Diversity of contexts/issues (high)

Duration (long)

(Berscheid et al.)

Interdependence’s correlates

Correlated .04 with emotional tone and .20 with subjectve closeness

Exchange vs. communal

Exchange: behavior is governed by the balance of costs and benefits

Communal: behavior flows from an obligation to respond to the other’s needs. Clark and Mills distinction

Relationships vary in…

Exchange vs. communal orientation, how voluntary they are (e.g., parent-child is least voluntary; friendships and “going steady” more so), and provisions

Complication: change over time

Provisions of relationships (Weiss)

Attachment, reliable alliance, enhancement of worth, social integration, guidance, opportunity for nurturance

The “problem” of development

A given relationship’s developmental significance can change dramatically over time, and different kinds of relationships can serve the same purpose at different times

Attachment

An affective connection through time serves two functions:

secure base and safe haven

It is probabilistic, not a theory of everything, not relevant to all relationships or all aspects of them, and not “bonding”

Klaus & Kennell (1976) Bonding hypothesis vs. attachment

Bonding: focus on maternal affect and cognition in the first minutes/hours of life, deterministic, tested with “at-risk” mothers

Attachment: an affective bond developing through time, probabilistic

Harlow’s surrogate mother study

A reaction to drive-reduction theory. Monkeys reared with wire and cloth surrogates spent far more time with cloth mothers, regardless of which one fed them. So feeding (drive reduction) isn’t the primary “adhesive” of early bonds

Bowlby

Interested in juvenile thieves’ developmental histories. Synthesized ethology, psychodynamics, and cognitive science into attachment theory, with the infant “pre-adapted” for bonds. Attachment’s proximity-seeking is seen as evolved to secure the infant’s safety

Normative stages of attachment

Asocial (0-6 weeks)

Indiscriminate attachments (6 weeks - 7 months)

Specific attachment (7-9 months)

Multiple attachments (a hierarchy of attachment figures)

Two ways to measure individual differences

Attachment Q-set (Waters; naturalistic observation)

The Strange Situation (Ainsworth; lab)

Strange Situation structure

Eight 3-minute episodes.

Infant (I)

Caregiver (C)

Stranger (S)

Experimenter (E)

Separation 1 (episodes 3 to 4), Reunion 1, Separation 2 (infant alone, episode 6), Reunion 2 (episodes 7 to 8).

Coding focuses on reunion behavior

Secure (B)

Uses the parent as a secure base for exploration

Separation distress is effectively alleviated by the caregiver

About 65% of infants

Associated with sensitive, responsive caregiving

Anxious/Avoidant (A)

Avoids the caregiver at reunion after stressful separation

About 20-25%

Developmental history: rejection

Anxious/Resistant (C) - also ambivalent

Poverty of exploration

Can’t be settled

Seeks yet resists comfort (Hyperactivation)

About 10-15%

Developmental history: inconsistent responsiveness

Disorganization (D)

A breakdown of strategy in the caregivers’ presence, and an “irresolvable dilemma” (the caregiver is both comfort and fear)

Shows confused or contradictory behaviors, fearful expressions, frozen postures

All infants form attachments to caregivers; they differ in quality, not presence

Linked to maltreatment

Disorganization: why it matters (Bernard et al.)

Disorganization at 18 months predicted dissociative symptoms through adolescence (Carlson) and externalizing problems (Lyons-Ruth)

Parental insensitivity alone doesn’t strongly predict disorganization, but frightening behavior does

Parental Sensitivity and the sensitivity hypothesis

Security is promoted by sensitive caregiving, meaning responsiveness to distress cues

Sensitivity as synchrony (Isabella & Belsky 199)

Distinct forms of asynchrony map onto A vs. C De Wolff & van IJzendoorn, (1997) meta-analysis, 66 studies: moderate association, r=.24

Temperament vs. attachment: the position

Both are systems for interpreting individual differences, but attachment researchers view them as empirically orthogonal and have largely rejected a role for endogenous origins of (in)security

Sroufe (1985)

Argued that early temperament contributes minimally to attachment security

Logical argument: secure children are behaviorally heterogeneous, with as much variation in separation distress within the secure group as between secure and anxious groups

Also set the empirical challenge: shows that behavioral differences in the early weeks relate to later classification. Reviews suggest the evidence is lacking

Twin logic (ACE model)

A (additive genetic) is estimated by doubling the MZ-DZ correlation difference.

C (shared environment) is the MZ correlation minus A.

E (non-shared environment plus error) is 1 - A - C.

Heritability is estimated by whether MZ twins are more similar than DZ twins

ECLS-B security data

Birth cohort 2001, N10,688

Same-sex twin pairs of known zygosity (24-month assessment: 168 MZ and 311 DZ pairs)

Security measured with a 45-item AQS adaptation

ICC MZ =.65, DZ = .62 (not significantly different), so A is about 0

Shared environment accounts for about 63% of variance; nonshared environment about 37%

Roisman & Fraley (2006): infant-caregiver relationship quality

9-month twins (172 MZ, 333 DZ pairs)

Observed with NCATS teaching mask

MZ and DZ correlations were virtually identical

ACE estimates: A=.01, C=.40, E=.59

Best-fitting model had no genetic component (CE)

The relationship quality is environmental, shared and non-shared

Why Roisman & Fraley challenge Turkheimer's "first law" The "law"

The "law" is that all human behavioral traits are heritable

Infant-caregiver relationship quality is a rare exception

Limits: it's one age and one domain; shared environment effects don't identify mechanisms (they may not be about parents); twin design can't rule out genetic variation in parents (Children-of-Twins design suggested)

Molecular genetics (SECCYD)

N = 711, SSP at 15 months, genotyped the "usual suspect" polymorphisms (dopamine, serotonin, oxytocin)

Mean effects were distributed around .00 for both security and disorganization; no replicable sensitivity x genotype interactions

Small-N molecular studies are hard to replicate in larger-N efforts

Overall: no aggregate genetic influence on security; Sroufe was prescient.

Bokhorst et al. (2003) and Waller & Shaver (1994)

Bokhorst: 157 twin pairs, MZ no more similar than DZ on security

Waller & Shaver: love styles may be a rare phenotype with little or no genetic variance

The four hypotheses

Universality (all infants become attached to specific caregivers)

Normativity (most infants are secure, using the caregiver as a secure base and safe haven)

Sensitivity (security depends on sensitive caregiving)

Competence/secure base (security scaffolds later competence: peers, adult relationships)

Ainsworth's Uganda study (1954-55)

28 infants: 16 secure, 7 insecure (cried a lot, ambivalent), 5 "nonattached" (probably avoidant in the SSP).

Showed attachment even with multiple caregivers; what mattered was continuity and quality of interaction, not the number of caregivers

Universality: key evidence

Every cross-cultural study found similar attachment behavior, and every culture studied shows the three patterns (even avoidant, though rarely in Israel)

In multiple-caregiver societies (Hausa, Efé, Gusii), infants still showed a preferred figure

Behavioral expressions vary, but the secure-base pattern doesn't (Gusii handshake vs. Western hug)

Normativity: key evidence

Secure was the majority in every study.

Examples: Gusii 61%, Dogon 69%, Khayelitsha 72%, Beijing 68%, city Israel 72%, Tokyo 61%, Sapporo 68%, West Java 57%

Mothers and experts in China, Colombia, Germany, Israel, Japan, and the US all describe the ideal child as secure (Posada et al. 1995; Japan r = .86)

Cultural exceptions to normativity

Communal kibbutzim: only 48% secure (vs. 80% family-based sleeping), with more ambivalent classifications (inconsistent responsiveness at night)

Israel overall overrepresents ambivalent and underrepresents avoidant

Sapporo (Takahashi) had no avoidant infants

Dogon: no avoidant, 24% disorganized

Sensitivity: key evidence

Supported in 12 studies and indirectly in 2

Examples: Gusii (maternal age, household density), Uganda (Peterson: maternal affect predicted 32% of AQS variance), Khayelitsha (sensitivity predicted security independent of postpartum depression, which was 35%), Tokyo (Vereijken: r of .59 or higher), Osaka and Sapporo (AAI predicting child security).

One failure: first Sapporo (Nakagawa et al.), probably low power (N under 30) and self-conscious mothers

Competence: key evidence (weakest support)

Gusii: secure attachment to nonmaternal caregivers predicted Bayley scores; attachment to mother predicted nutritional status, so effects were domain-specific

Kibbutz: attachment to the metapelet predicted kindergarten competence; the extended attachment network predicted best.

Dogon: security linked to better nutrition.

Takahashi (Sapporo) found early differences disappeared by year 3

Conclusion: cross-cultural support is still insufficient

Attachment networks and the integration vs. hierarchy models

Integration: multiple secure attachments compensate for insecure ones linearly

Hierarchy: the mother (primary caregiver) matters most

Kibbutz data fit integration in kindergarten, but hierarchy later

Rothbaum et al.'s critique of attachment theory

Claims core propositions are culturally relative

Argues Japan is a challenge to universality, normativity, and sensitivity, and that amae resembles ambivalent behavior (so ambivalence is culturally appropriate in Japan)

Also claims there are distribution differences across cultures (e.g., Germany, Japan)

van IJzendoorn & Sagi-Schwartz's rebuttal

Rothbaum overlooked key evidence

Vereijken's Q-sort: amae and attachment were orthogonal, and amae resembled dependence

Behrens (2004): amae has a negative connotation and lacks biological roots; calling them competing concepts is a "category error."

Only the first Sapporo study supports Rothbaum, and it had procedural problems (crying episodes lasting 2 minutes, not 20-30 seconds)

Rothbaum et al. (2000), main thesis

Japan follows a path of "symbiotic harmony" (union, others' expectations, stability, assurance); the US follows "generative tension" (reunion, personal preferences, transferability, trust)

The differences lie in the meaning and dynamics of relationships, not their strength

Bronfenbrenner's ecological model

Microsystem: immediate contexts (family, school, church, daycare, neighborhood play areas)

Mesosystem: links among microsystems (home-school)

Exosystem: settings the child doesn't experience directly but that affect development (parent's work, a plant closing that cuts school funding)

Macrosystem: cultural/subcultural context embedding the others

Kelley et al.'s model

A close relationship (P-O interaction) is shaped by and shapes both individuals (P, O) and society

Context (society) influences relationships, and vice versa

Direct vs. indirect context-relationship links

Direct: contextual norms, expectations, and values determine which relationship dimensions vary and how

Indirect: more complex connections that show mechanisms, usually mediational or moderated

Mediation

A third variable (M) accounts for the correlation between two variables: I leads to M leads to D.

A mediator can reduce (attenuate) the I-D correlation

Moderation

A moderator changes the relation among variables; the I-D link holds for one group but not another (e.g., boys vs. girls)

Moderators can be discrete or continuous

Culture as context: co-sleeping (Morelli et al.)

Mayan families predominantly co-slept; US families used separate rooms as "independence training."

Cultural prescriptions directly shape relationship behavior

Challenge: defining and measuring culture

Berscheid's "environment as ether" argument

Relationship researchers treat a relationship's environment as undifferentiated surround

Examples of environmental effects: alternative partners in the area (Udry; South & Lloyd), network approval of the relationship, economic strain (Conger), job stress (Repetti)

Karney & Bradbury's vulnerability-stress model: fate depends on couple vulnerabilities x environmental stressors

What is ABC?

Attachment and Biobehavioral Catch-up: a 10-session, manualized home intervention for parents at risk of neglecting young children (referred through child protective services)

Targets: provide nurturance when children are distressed, follow the child's lead with delight, and don't be frightening

Uses in-the-moment feedback and video feedback

ABC results

120 children, randomized to ABC or the control (Developmental Education for Families, DEF)

Disorganized: 32% ABC vs. 57% DEF (d = .52)

Secure: 52% vs. 33% (d = .38)

The security effect didn't hold when excluding children over 24 months; the disorganization effect did