Why Minds Break · Ian Pitchford · Chapter 6 of 10

Development Under Uncertainty

Name: Why Minds Break
Author: Ian Pitchford

Separates whether early conditions predict later vulnerability, whether development causes the link and whether it evolved as a forecast, then tests the leading proposals against the evidence.

The chapter keeps apart questions that are often merged: whether early circumstances predict later outcomes, whether development itself produces the association, whether that process evolved to prepare a child for the world to come, and whether it amounts to learning an expectation in the computational sense of Chapters 4 and 5. It stresses that population associations cannot explain one person's difficulties and that studying pregnancy places no blame on mothers. David Haig's account of pregnancy as a negotiation between partly divergent genetic interests, extending Robert Trivers, casts doubt on any assumption that mother and fetus share one agenda. Genomic imprinting gives that view its firmest support, while Haig's readings of gestational diabetes, pre-eclampsia and Prader–Willi syndrome remain functional hypotheses or conjecture.

David Barker's thrifty phenotype proposed that poor early nutrition alters the body in ways that become harmful when food improves, and low birth weight does predict later diabetes and heart disease. Related proposals extend the idea to social cues, among them Chisholm's account of attachment, Gluckman and Hanson's predictive adaptive responses and the Adaptive Calibration Model, and each needs testing channel by channel. Nettle, Frankenhuis and Rickard showed how unreliable cues limit forecasts across long intervals, and Wells read the thrifty phenotype as a maternal effect instead. Early programming happens; reading it as an evolved forecast is disputed.

Four channels are examined. Studies linking prenatal stress to schizophrenia point in different directions. Prenatal anxiety predicts childhood behaviour problems modestly, but IVF and Norwegian family designs attribute much of the link to shared genes. Father absence is weakly associated with earlier first menstruation, though a children-of-twins analysis points to family confounding. Attachment predicts some problems yet shows little stability over long intervals, and a 2025 pooled analysis found no sign of differential susceptibility in its main tests. Further sections cover male fragility and sex ratios, grandmothers, older fathers, breastfeeding and life-history profiles, where Del Giudice's differentiated model meets the objections of Stearns and Rodrigues.

The verdict is that the evidence settles the predictive question best, the causal one in part and the evolutionary one least. No study reviewed measured the expectations or confidence settings a predictive-processing explanation would need, so the idea that adversity installs highly precise priors stays a candidate rather than a finding. Someone who expects danger in a dangerous setting need not reason badly.

The chapter ends with a recurring problem: a response can alter what it predicts. An avoidant person who meets little harm may have prevented it, missed opportunities or achieved nothing. Taking Rosenström's finding that accumulated depression went with rising harm avoidance, it argues that studies must record the opportunities people forgo as well as those they accept.

Questions the chapter asks

Do difficult conditions before birth or in early childhood cause later mental health problems, or merely predict them?
Did children evolve to read early cues as a forecast of the world they will grow up in?
How can anyone judge whether a cautious response was justified when the caution itself changes what happens?

This is a summary of the chapter, written for this site; it is not an extract from the book. Coming soon from Amazon in Kindle, paperback and hardback editions.