Hermida's central complaint was that most models linking poverty to cognitive development are built almost entirely on evidence from wealthy, "extreme" contexts, and they just don't generalize across the range of socioeconomic and cultural settings that actually exist, especially across Latin America, where she's done most of her fieldwork.
Children raised in poverty show lower generosity and altered social preferences in some studies. The open question is whether that reflects a deficit, a genuine harm to development, or an adaptation, a sensible response to growing up in a resource-scarce environment. Current models mostly explain results at one extreme of the socioeconomic distribution and go quiet at the other. She was careful to say this framing isn't an excuse for poverty. Poverty is a violation of human rights regardless of whether you can tell an adaptive story about a particular outcome.
A point she kept returning to: it's not just how much money a family has, it's how unpredictable it is. Seasonal swings, day-to-day unpredictability for gig or informal workers. About 30% of some study populations report unpredictable income, and that share is far higher in parts of Latin America than in typical US comparison samples, yet volatility almost never gets measured alongside income level in the standard literature.
One striking result came from a study using children's human-figure drawings as a developmental proxy. Months of maternal absence due to labor migration were associated with roughly a 10% increase in the drawing score, the opposite of what a naive "absence hurts development" prediction would say. The explanation, once researchers looked closer, was remittances. Mothers send money home, children eat and live better, and that improvement outweighs the cost of the mother's absence. It's a clean example of context flipping the sign of an effect that looks obvious on its face.
Purchasing-power comparisons show large gaps between wealthy countries and Latin American countries at the bottom of the income distribution, but that gap almost disappears at the top. Inequality's cross-country signature is concentrated among the poor, not the rich. And in a study of about 131 five-year-olds comparing rural and urban children matched as closely as possible on a composite measure of unmet basic needs, communicative development still differed by rurality even after controlling for that measure. The standard SES index misses something real about living rural versus urban, even within the same country.
She closed with a set of recommendations: more diverse samples across socioeconomic contexts, more careful and nonlinear measurement of SES rather than a single linear composite, treating SES as something dynamic rather than fixed, and testing deficit-versus-adaptation models in genuinely different contexts instead of assuming a US-derived finding travels.
One audience member brought up a related finding from Brazil: children raised in chronic poverty show a flatter contrast-sensitivity function, meaning the effect of poverty reaches down into basic, low-level visual processing, not just "higher" cognition. Another question challenged whether an aggregate SES construct is even useful compared to studying specific mechanisms directly, stress, sleep loss, noise exposure. Hermida agreed this is a live, unresolved debate. Her own position is that aggregate measures are still useful for now, mostly because most current studies don't report enough detail to support the more mechanistic approach yet. A last question raised the issue of cultural bias in pencil-and-paper tasks that assume schooling and literacy. She acknowledged this as a real limitation, saying they try to use tasks common to every child's school experience and pilot them locally, while admitting no measurement tool can ever be fully culture-neutral.
Several specific percentages and sample sizes here are from memory or quick notes, so double-check before citing them anywhere.