A new interdisciplinary study, published on October 15, 2025, raises a provocative question: did natural lead exposure during the Pleistocene systematically hinder brain and language development in Neanderthals and other extinct hominids, while a tiny genetic modification gave modern humans a crucial advantage? Researchers from the University of California, San Diego, and an international team of collaborators argue that traces of lead were much more present and earlier than previously thought—millions of years before organized mining—and that a variant of the NOVA1 gene, unique to Homo sapiens, cushioned the neurotoxic effects, creating space for the development of complex language and social networks that marked the rise of our species.

Lead in fossilized teeth: a silent record of early exposure

The team analyzed 51 fossil and historical tooth samples from across Africa, Asia, and Europe, including members of the genus Homo (modern and archaic), early hominids like Australopithecus africanus, and extinct great apes like Gigantopithecus blacki. In as many as three-quarters of the samples, clear chemical “rings” of lead exposure were detected, and in G. blacki—a giant ape that lived about 1.8 million years ago—the most frequent patterns of acute exposure were recorded. Such patterns in tooth enamel create a chronology similar to tree rings: each red or saturated “band” marks an episode of intense contact with lead during growth phases.

Scientists expected the highest lead burden to come from antiquity and industrialization—Roman lead pipes, paints, and gasoline—but were surprised to find that exposure patterns in prehistoric teeth were often comparable to samples from individuals born in the mid-20th century. This outlines a natural, episodic source of lead: volcanic ash, dust from fires, mineralized groundwater, and lead-rich rocks in caves that hominids sought for shelter and water sources.

Why caves might be the culprit

The hypothesis of caves as exposure “hotspots” relies on geochemistry: in many karst and mountainous areas, where caves provided safety, fresh water, and a constant temperature, mineral veins with lead are present. Water percolating through such rocks can occasionally carry dissolved lead ions, and the deposition of dust and contact with contaminated sediments further increases the risk. Tooth enamel, the hardest tissue “archive” we have, reliably records these episodes in fine, rhythmic growth layers. Research on proteins and elements in ancient enamel has advanced dramatically in recent years, opening a window into the ecology and exposures of periods older than two million years.

A small genetic twist with big consequences: NOVA1

At the center of this new story is the gene NOVA1 (neuro-oncological ventral antigen 1), a regulator of neural differentiation and the formation of synaptic networks. Most modern humans have a “newer” allele that differs by just one base from the archaic variant recorded in Neanderthals and Denisovans. Back in 2021, it was shown that replacing the human variant with the archaic one in neural organoids—miniature “brains” grown from stem cells—changes the architecture, maturation, and synchronization of neural networks. The new study goes a step further: it simulated the effects of lead on organoids with both the human and archaic variants of NOVA1 and found that lead more strongly destabilizes the developmental programs of neurons in the archaic context.

Specifically, lead disrupted the expression of a series of developmental genes in both variants, but only the archaic NOVA1 variant triggered changes in the expression of FOXP2—a gene strongly linked to the motor aspects of speech and language processing. In humans with pathogenic FOXP2 mutations, severe difficulties in articulation and planning speech movements are observed. Although there has long been debate about how much FOXP2 is a “language gene” and to what extent it was shared with Neanderthals, the new regulatory perspective suggests that it's not all in the FOXP2 sequence itself, but also in how it is managed by a network of regulators in which NOVA1 holds an important hub.

From neurons to language: how environment modulates genetic potential

Lead is a potent neurotoxin that interferes with synaptogenesis, alters calcium signaling, and disrupts white matter development; the consequences are measurable in lower IQ scores, impaired emotional regulation, and long-term cognitive difficulties when exposure occurs in early childhood. The new organoid observations suggest that the modern human variant of NOVA1 mitigated some of the damage in neurons associated with speech and language planning circuits, while the archaic configuration, especially under conditions of episodic lead “stress,” led to faster early maturation but also less network complexity over time. Such a dynamic could explain why Neanderthal populations had a limited capacity for symbolic exchange and the transmission of complex information, despite evidence of abstract thinking and technical skill.

What teeth tell us about water, migrations, and social strategies

The tooth samples indicate that lead contact was “episodic”—appearing in bands, not as continuous saturation. This is important because such a pattern reflects behavior: seasonal searches for water in caves and crevices, migrations through geologically “diverse” terrain, and occasional drought crises when cave streams were the only source. If the modern human variant of NOVA1 provided resistance, even slight, then each wave of exposure would have “derailed” the developmental trajectory of language-cognitive networks less in our species. In populations with an archaic regulatory profile, the same waves could have had a cumulatively greater effect, reducing social cohesion and competitive range compared to groups with greater linguistic efficiency.

Controversies and limits of interpretation

The role of FOXP2 in human language is a subject of long-standing debate: today, it is considered one of many cogs in a broader assembly, more of a “pragmatic regulator” of speech-motor loops than the sole switch for “language vs. no language.” Accordingly, the authors also emphasize that the new hypothesis should not be read as deterministic: it is about the interaction of environmental stressors (lead), evolutionarily “ironed-out” regulatory networks (NOVA1), and a broader set of genes related to neurodevelopment and plasticity. The complexity is also supported by recent research on animal models and populations, which shows that small changes in “language” genes can alter vocalizations and communication patterns, but that the path from neuron to culture is multi-layered and contextual.

Next-generation methods: organoids, mass spectrometry, and the enamel “diary”

The study merges two research worlds. The first is the organoid approach, which uses induced pluripotent stem cells to grow structures that mimic the early development of the cortex and thalamus; in such models, it is possible to precisely change individual alleles (e.g., replace the human NOVA1 with the archaic one) and track the consequences in synapse formation, activity rhythms, and transcriptomic networks. The second is the forensic reading of tooth enamel: laser ablation and high-resolution mass spectrometry “read” layer by layer of growth, capturing trace elements (including lead) and thus reconstructing the calendar of exposure during an individual's early life. The combination of these two approaches has enabled a rarely seen causal narrative: the same environmental stressor was tested under controlled conditions on a genetically defined neural system, while fossils provided independent confirmation that this stressor was indeed present and relevant.

Neanderthals, social networks, and “chemical resistance”

There has been much debate about how capable Neanderthals were of complex symbolism and organization. Archaeological records attest to sophisticated tools, the use of pigments, and potential forms of symbolic behavior. However, the new hypothesis shifts the focus: it's not just about cognitive capacity, but about the resilience of neural networks to real, cyclical environmental shocks. If humans possessed a regulatory “cushion” against lead, the difference in communication efficiency—the speed of learning, the precision of information transfer, the stability of vocal-motor programs—could have become crucial in moments of social competition, expansion, and the exchange of knowledge over greater distances.

What was (and wasn't) said: between caution and excitement

The authors clearly emphasize the limitations: organoids are not “miniature people,” but reductive models of the early brain; the fossil record is fragmentary and depends on the preservation and representativeness of samples; and a correlation of exposure does not automatically imply causation in population outcomes. Nevertheless, the story fits into a broader trend in evolutionary neuroscience that, instead of searching for a single “magic gene,” investigates nodes in the network—regulator genes like NOVA1—and their interaction with the “real” environment that our ancestors found, not created. Such a framework helps explain how subtle differences in regulation could have amplified the effects of ubiquitous toxins and thus more finely channeled the development of cognitive systems.

Date and context: why October 15, 2025, is important

The publication of the paper in mid-October 2025 comes after a series of discoveries this year that have pushed the boundaries of readability of tooth enamel and ancient protein traces, as well as new findings on candidate genes involved in speech-language functions in experimental models. This convergence of techniques and themes creates an atmosphere in which it is possible to more calmly connect “wet” biological experiments with “dry” fossil records and thus rethink the selection pressures that shaped the human brain.

What this means for the clinical and educational spheres today

Understanding how variants like the modern NOVA1 modulate sensitivity to environmental toxins could open new avenues for developing preventive strategies in public health, especially in communities where infrastructure is old and lead is still present in water or paint. At the same time, in the neurodevelopmental clinic, it raises the question of individual differences in functional recovery after early exposure and potential biomarkers that could help identify children at greater risk for language difficulties and autism spectrum disorders. Although the path from the lab bench to therapeutic recommendations is long and cautious, such research clearly illustrates that “language” is not just a cultural competence, but also a biological infrastructure that the environment can either suppress or stimulate.

Key questions for future research

• How much did lead exposure vary between the regions and habitats that hominids inhabited, and was there a migratory advantage for groups that avoided geologically rich lead deposits?


• How does the modern human NOVA1 variant integrate with other regulatory hubs related to language (e.g., networks involving FOXP2 and a range of transcription factors), and does resistance differ between individuals and populations?


• Can additional “imprinted” records of lead exposure be found in other biotic markers (e.g., tooth root cementum, auditory ossicles) that would confirm the episodic pattern?


• What can integrated models—organoids, artificial neural networks, and computational simulations of population dynamics—tell us about the thresholds above which language learning is permanently compromised?

Why language is a “superpower”—and why chemistry matters

Language enabled the coordination of large groups, accelerated the exchange of technologies and stories, and increased the scope of planning and collective memory. If biochemical resistance—even if only slightly—protected these fragile networks from the “noise” caused by lead, the advantage multiplied every time a community had to quickly transmit knowledge about food, dangers, or tools. In this light, the tiny shift in the NOVA1 sequence may have indeed turned into a huge difference in the fate of species: not because it “gave language” itself, but because it allowed language to survive the real world of prehistory, with its dust, caves, and mineral waters that sometimes carry—lead.

Additional context for readers

For those who want to delve deeper into the topic, it is advisable to understand the basic concepts. Brain organoids are experimental models that reproduce the early stages of development and are used to examine genetic and environmental influences. FOXP2 is a transcription factor associated with articulation and language processing, but it is not the “only gene for language.” Small variants in candidate genes can measurably affect vocalizations in animal models, but human language arises from a network of hundreds of genes, experience, and culture. Finally, archaeochemical analysis of tooth enamel and proteomics extend our ability to read biological archives far into the past, to timeframes older than 2 million years.