Kagan’s longitudinal study
Jerome Kagan (1929–2021) spent his career at Harvard studying the nature and the limits of temperament. After earlier work and a 1980s pilot, he and Nancy Snidman launched the now-classic longitudinal study in 1989. They screened roughly 500 four-month-old infants and then followed many of the same children at ages 1, 2, 4½, 7, 11, and 15 – with Carl Schwartz later continuing the work with neuroimaging into early adulthood.
The headline finding is durable: infants who reacted most intensely to novelty were significantly more likely to become behaviorally inhibited toddlers, shy school-age children, and socially reticent or anxious adolescents. The bias was still measurable, decades later, in how their brains and bodies responded to unfamiliar things.
Infant temperaments act to limit what children will become; they do not guarantee a particular personality.
– Jerome KaganWhat the screening actually measured
The 45-minute battery was designed to be mildly novel, never frightening: a still mother’s face, three moving mobiles, recorded speech, popping balloons, and cotton swabs dipped in dilute alcohol held briefly near the nose. What the researchers scored was the combination of two responses:
- Motor activity – vigorous arching of the back and pumping of the arms and legs.
- Distress – frequency of crying and fretting.
This produced three rough groups among the four-month-olds:
Crucially, “high reactive” is the infant marker. The childhood phenotype it tends to grow into – wariness, freezing, and a long latency to approach new people and objects – is called behavioral inhibition (BI). The two terms describe the same thread at different ages.
Predisposition, not destiny
This is the most important and most often-mangled point. The prediction is probabilistic. Across the longitudinal follow-ups, only about a third of children showed fully consistent behavior at every age. Many high-reactive infants became measured, thoughtful adults with no anxiety disorder at all; some children with no early markers developed anxiety anyway.
Kagan described the temperamental bias as “a constraint on what is possible rather than a determining force.” It tilts the odds. It does not write the story. Population-level findings should never be read as a forecast for one individual child.
The neighboring concepts
High reactivity sits inside a crowded neighborhood of related ideas. They overlap, but they are not interchangeable, and the evidence behind them differs a lot.
Behavioral inhibition (BI)
The toddler/childhood phenotype downstream of high reactivity: wariness, freezing, and slow approach to novelty. This is the line with the strongest longitudinal evidence.
The Behavioral Inhibition System (BIS)
Jeffrey Gray’s broader neuropsychological construct – a system that generates anxiety and risk assessment in response to conflict and punishment cues. It overlaps with Kagan’s BI but is a wider, more theoretical idea.
The Highly Sensitive Person (HSP) / Sensory Processing Sensitivity (SPS)
Elaine and Arthur Aron’s trait (1997), emphasizing depth of processing, ease of overstimulation, emotional reactivity, and sensitivity to subtleties. Aron places it in 15–20% of people. It is measured by self-report rather than infant laboratory observation, and its evidence base is thinner and more recent than Kagan’s – best treated as a complementary, partly overlapping framework.
Orchid and dandelion (differential susceptibility)
W. Thomas Boyce and Bruce Ellis’s idea that a minority of “orchid” children are biologically more reactive to both adverse and supportive environments – they wilt in harsh conditions but flourish in nurturing ones. Recent work suggests this sensitivity is continuous rather than a sharp two-type split.
Thomas & Chess and Rothbart
The New York Longitudinal Study introduced the “slow-to-warm-up” cluster (~15%) and the enduring concept of goodness of fit. Mary Rothbart’s influential model organizes temperament around reactivity and self-regulation (effortful control) – high reactivity in Kagan’s sense maps roughly onto high negative affectivity in hers.
A side-by-side comparison
| Concept | Core idea | How it’s measured | Evidence base |
|---|---|---|---|
| High reactive / BI | Biologically based wariness toward novelty | Infant lab observation; behavioral coding | Strongest – decades of longitudinal & neuroimaging data |
| BIS (Gray) | Anxiety system responding to punishment cues | Self-report scales (e.g., Carver & White) | Strong theoretical & psychophysiological base |
| HSP / SPS (Aron) | Depth of processing & sensory/emotional sensitivity | Self-report (HSP Scale) | Smaller, newer; overlaps with neuroticism/introversion |
| Orchid / dandelion | Heightened plasticity to both harm and care | Reactivity markers × environment | Growing; susceptibility likely continuous, not bimodal |
| Slow-to-warm-up | Initially withdrawn, slow to adapt, low intensity | Parent-report dimensions (NYLS) | Foundational; introduced “goodness of fit” |
Category or dimension?
Kagan often spoke of high reactivity as a category – a distinct kind of child. Most subsequent evidence, though, supports a dimensional view: reactivity varies along a continuum, and “high reactive” describes the extreme tail. The samples for the most-replicated extreme groups are modest in size, and the original cohorts were predominantly middle-class and Caucasian, from the Boston area. Cross-cultural studies find similar reactivity distributions, but the social meaning of inhibition varies – shy children are viewed more positively in some East Asian contexts than in U.S. peer groups.
Next: the coordinated biological profile behind all of this – what’s actually different in a high-reactive amygdala, heart rate, and stress hormones. Read the biology →