by Andrea Lavazza
Kaganovsky and colleagues genetically modified mice so that much of the cerebral cortex and hippocampus would not develop. They then transplanted human neural organoids into the space left vacant. This is a remarkable scientific achievement, intended to provide a living model for studying brain development and disease and for developing therapies. It is also a potentially controversial experiment. Previous studies generally placed relatively small grafts into an otherwise intact animal brain. The human cells became vascularized, connected with the host’s neural circuits and sometimes contributed to behavioural responses. The new study, published in Nature on 16 September 2026, represents a structurally different step.
The researchers produced “apallial” mice in which large parts of the neocortex and hippocampus were absent. Their total brain tissue was reduced by approximately half, although other brain structures remained. Soon after birth, four human cortical organoids were transplanted into the space created by this loss. The grafts grew, developed several human cortical cell types and connected with the mouse nervous system. The mice retained broadly normal movement, but differences emerged in coordination, working memory and spontaneous behaviour. In the analysis of spontaneous behaviour, animals with human grafts occupied an intermediate position between control and apallial mice. The model was also used to study the effects of oxygen deprivation on developing human neural tissue.
Ethical discussion of neural chimeras has generally asked whether human cells might give an animal new or enhanced cognitive capacities. The advisers consulted for this study also considered the possibility of altered or improved capacities, and the researchers performed extensive behavioural testing. The present model, however, raises a somewhat different ethical concern. The experiment first prevents much of the mouse cortex and hippocampus from developing and then introduces immature human tissue into the resulting neural space. It therefore changes the structures that may support not only cognition but also suffering and the expression of distress.
The apallial mouse cannot simply be treated as an ordinary mouse with one isolated impairment. The loss of much of the neocortex cannot be assumed to eliminate conscious pain, because cortical and deeper brain structures may both contribute to pain and awareness. Nothing reported so far suggests that the xenocortical mice acquired human-like capacities. Yet we do not know whether the graft partly restores, reorganizes or otherwise transforms capacities affected by cortical depletion.
Sentience should therefore not be treated here as a fixed property of the host species. It may change during development, transplantation and graft maturation. Not every such change would be ethically objectionable. Three possibilities matter: the animal might experience greater suffering; retain the capacity to suffer while losing familiar ways of expressing distress; or develop unfamiliar forms of experience whose implications for welfare are difficult to interpret. In each case, ordinary behavioural indicators could become misleading. The ethical concern is therefore not uncertainty alone, but the risk of underestimating harm. This should support repeated welfare assessments and more cautious criteria for stopping the experiment as the graft matures.
The study’s tests of sensitivity to heat and touch are relevant but do not settle this issue. Withdrawal responses were similar across the control, apallial and xenocortical groups, but such responses do not establish that the felt or emotional dimension of pain remained unchanged. Farahany and Seth also highlight that xenocortical mice spent less time in the centre of a darkened chamber than either comparison group and stood on their hind legs less often than controls. These behaviours are conventionally treated as possible signs of anxiety, although they do not establish conscious distress.
Existing animal-research principles remain applicable, but this model may require them to be used in a more dynamic way. Assessment of severity should consider the creation of the apallial condition itself, not only transplantation and behavioural testing. Welfare should be reassessed as the graft develops rather than assumed to remain constant throughout the experiment. As Farahany and Seth also propose, monitoring should occur at pre-specified intervals, and oversight committees should have the authority to require changes, pause or stop the study if familiar behavioural signs become less reliable indicators of distress. The researchers adopted substantial safeguards, including supplementary food, delayed weaning, anaesthesia and pain relief. But the breeding protocol also involved culling most littermates with an intact cortex to reduce competition for resources. Ethical evaluation should consider this total burden, including the production and maintenance of the model.
The study also complicates the idea that organoids will reduce reliance on animal experimentation. Here, organoids become the basis for creating a new and highly specialized animal model. This need not imply that the study violated the principles of replacement, reduction and refinement. It nevertheless shows that organoids cannot automatically be counted as replacements when they are used to expand what can be done with animals.
The unusually broad oversight surrounding this study is a strength. Yet this case also shows why multidisciplinarity does not necessarily amount to ethical pluralism. The article and its ethics supplement do not provide a full list of the ethicists involved. Farahany and Seth now report that Farahany served on the independent, uncompensated external advisory board. Ethical positions on the use of human neural organoids and related chimeras vary considerably, ranging from concerns about ethical hype and premature alarm to more precautionary approaches to consciousness, animal welfare and moral status.
Ethical pluralism would test the assumptions underlying experimental design. Publishing committees’ reasons, evidential thresholds and significant disagreements would make that pluralism assessable. If oversight focuses mainly on enhanced or human-like capacities, it may overlook different questions: the deliberate creation of neural impairment, possible changes in suffering and the reduced reliability of ordinary welfare indicators.
Kaganovsky and colleagues have developed a model of considerable scientific interest and have devoted unusual attention to its ethical implications. Precisely for this reason, the study should be taken as the beginning of the discussion. When neural structures are first depleted and then partly reconstructed, sentience becomes a moving target. Ethical oversight should follow that trajectory.
Author: Andrea Lavazza
Affiliation: Associate Professor of Moral Philosophy at Pegaso University, Naples, Italy
Competing interests: The author declares no conflicts of interest.
AI use disclosure: GPT-5.6 Sol was used to assist with translation and editorial revision. The author reviewed and takes responsibility for the final text.