Low socio-economic status (SES), and a conflictive, cold and unsupportive family environment in childhood have been associated with early adulthood hostility. However, it is unknown whether this association changes in magnitude with age from childhood to adulthood. We investigated whether childhood family factors (SES and parental child-rearing style) predicted differential development of offspring hostility and anger from early to middle adulthood.

Between 2041 and 2316 participants (age range 3–18 years at baseline) were selected from the longitudinal Young Finns study. The participants were followed for 27 years between 1980 and 2007. Childhood SES and parent's self-reported child-rearing style were measured twice: at baseline and 3 years after baseline. Hostility and anger were assessed with self-report questionnaires at 12, 17, 21 and 27 years after baseline.

Low parental SES and hostile child-rearing style at baseline predicted higher mean levels of offspring anger and hostility. Low parental SES and one of the hostile child-rearing style components (strict disciplinary style) became more strongly associated with offspring hostility with age, suggesting an accumulating effect.

Childhood family factors predict the development of hostility and anger over 27 years and some of these family factors have a long-term accumulating effect on the development of hostility.

The psychobiological model of personality by Cloninger and colleagues originally hypothesized that interindividual variability in the temperament dimension ‘harm avoidance’ (HA) is explained by differences in the activity of the brain serotonin system. We assessed brain serotonin transporter (5-HTT) density in vivo with positron emission tomography (PET) in healthy individuals with high or low HA scores using an ‘oversampling’ study design.

Subjects consistently in either upper or lower quartiles for the HA trait were selected from a population-based cohort in Finland (n = 2075) with pre-existing Temperament and Character Inventory (TCI) scores. A total of 22 subjects free of psychiatric and somatic disorders were included in the matched high- and low-HA groups. The main outcome measure was regional 5-HTT binding potential (BPND) in high- and low-HA groups estimated with PET and [11C]N,N-dimethyl-2-(2-amino-4-methylphenylthio)benzylamine ([11C]MADAM). In secondary analyses, 5-HTT BPND was correlated with other TCI dimensions.

5-HTT BPND did not differ between high- and low-HA groups in the midbrain or any other brain region. This result remained the same even after adjusting for other relevant TCI dimensions. Higher 5-HTT BPND in the raphe nucleus predicted higher scores in ‘self-directedness’.

This study does not support an association between the temperament dimension HA and serotonin transporter density in healthy subjects. However, we found a link between high serotonin transporter density and high ‘self-directedness’ (ability to adapt and control one's behaviour to fit situations in accord with chosen goals and values). We suggest that biological factors are more important in explaining variability in character than previously thought.