The familiar picture of memory centres on neurons. Yet the branching astrocytes woven among them also influence how a learned experience shapes behaviour. In a 2026 mouse study, changing astrocyte signalling altered fear responses and the neural activity supporting them. The finding places communication between different cell types at the heart of the story. [2]

Olena Bukalo and colleagues published the peer-reviewed research online in Nature on 11 February 2026. The experiments examined the basolateral amygdala, or BLA, in laboratory mice. [1]


Meet the other cells in the circuit

Astrocytes are part of the nervous system’s glial cells. Glia include several distinct cell types, so the term is broader than astrocyte. These cells have often been introduced as the support system that allows neurons to work, but their close involvement with synapses also makes them relevant to information processing. [3]

A neuron’s familiar rapid electrical spike is called an action potential. Astrocytes do not communicate through that same pattern of spikes. They can nevertheless participate in signalling, including changes in calcium inside their cells. Electrical silence in the neuronal sense therefore does not mean biological inactivity. The important question is what their activity contributes to a circuit. [3] [1]

What it means to retrieve and update learned fear

The researchers paired a sound cue with footshock during conditioning. Later, they presented the cue without the shock. They measured freezing, a defensive behaviour, while also investigating cell activity. The methods report three cue–shock pairings and specify male mice. [1]

Freezing provides an observable response, not a direct account of an animal’s subjective experience. For interpreting a memory experiment, it helps to distinguish learning an association, expressing a response when it is recalled, and changing future responses after new experience. A lower response during one test does not, by itself, reveal which of those processes changed.

Repeatedly encountering a previously threatening cue without its expected outcome can produce extinction. Behavioural research has shown that extinction commonly involves new, context-sensitive learning, while the original association can survive. Fear can consequently return under some circumstances. This is why “learning safety” and “erasing a memory” are not interchangeable descriptions. [4]

Recording activity, then testing its contribution

The team used calcium-sensitive fluorescent indicators to monitor astrocytes. These indicators report changes associated with calcium activity; they do not display the contents of a memory. The initial photometry experiment included 17 mice; other experiments used separate cohorts. Recording was combined with targeted genetic and chemogenetic tools that changed astrocyte calcium dynamics. Chemogenetic tools place engineered receptors in selected cells so an administered compound can alter their activity. [1]

The central result went beyond a correlation: two different manipulations shifted fear recall in opposite directions, while both disrupted extinction learning. This makes the timing and pattern of astrocyte signalling important, rather than suggesting that simply increasing or decreasing activity is always beneficial. [1]

The effects extended to neurons. Disrupting astrocyte signalling changed fear-related neuronal patterns and how information was relayed toward the prefrontal cortex. The university’s research account describes this as a contribution both to representing the learned situation and to the circuitry that helps produce a defensive response. [2]

Together, the measurements address different levels of explanation. Behaviour establishes what the animal did. Cellular recordings show activity associated with the task. Perturbations test whether changing one component alters the outcome. Agreement across these approaches is more informative than a striking cell image or a behavioural difference considered on its own.

How this fits earlier astrocyte research

A separate study by Michael R. Williamson and colleagues, published online on 6 November 2024 and in a January 2025 Nature issue, examined learning-associated astrocyte groups in the mouse hippocampus. Reactivating those groups stimulated recall, and the researchers identified a role for the transcription factor NFIA. That study concerned a different brain region and experimental approach, but also implicated astrocytes in memory-related circuit function. [5]

Ken-ichi Dewa and colleagues reported another aspect of the problem in a study published online on 15 October 2025. They found that experience could leave multiday changes in subsets of astrocytes, with later recall engaging signalling linked to memory stabilization. Their work connected changes in gene regulation and noradrenaline responses with the effects of repeated experience. [6]

These studies should not be treated as measurements of one universal astrocyte mechanism. Different tasks, brain regions and timescales can expose different contributions. Together, they motivate a more specific question: which astrocytes participate, in which circuit, at which stage of learning? That is more useful than replacing a neuron-only account with an equally broad astrocyte-only account.

What the new study does not establish

The featured experiments were performed in male laboratory mice, so they do not establish identical effects in females or humans. A conditioned sound cue is also a restricted model of experience. Human autobiographical memories involve additional contexts, meanings and histories. Extending the findings requires testing, rather than assuming that every kind of memory uses the same arrangement. [1]

The authors and their institution discuss possible relevance to persistent, inappropriate fear. That is a research direction. Targeted laboratory manipulations do not establish a safe therapy, an effective clinical intervention or a way to remove an unwanted memory. Their value here is in helping identify cellular contributions that future research can investigate. [2]

The next questions are more precise

An important unresolved issue is how astrocytes influence the relevant neurons at the molecular level. The authors discuss candidate neurotransmitter and neuromodulator pathways, but additional experiments are needed to separate their contributions. [1]

Useful follow-up questions include whether the same relationships appear across sexes, how different astrocyte subgroups respond, and which signalling changes are necessary at each stage of extinction. These are proposed next steps, not additional findings. For neuroscience and biotechnology, the practical opportunity is to develop measurements and interventions precise enough to test those questions.

The memorable result is a cellular partnership: understanding a neural circuit can require studying the cells that shape its activity as carefully as the neurons carrying its signals.

Sources and references

  1. Bukalo, O. et al. Astrocytes enable amygdala neural representations supporting memory. Nature 652, 434–441 (2026; published online 11 February).
  2. Rajalakshmi, N. The brain cells long called 'support' found to be critical for aversive memory. University of Arizona News, original research-institution report (2026; 20 February).
  3. Queensland Brain Institute What are glia?. University of Queensland, Brain Basics (undated; accessed 29 October 2026).
  4. Bouton, M. E. Context and behavioral processes in extinction. Learning & Memory 11, 485–494 (2004).
  5. Williamson, M. R. et al. Learning-associated astrocyte ensembles regulate memory recall. Nature 637, 478–486 (2025; published online 6 November 2024).
  6. Dewa, K.-i. et al. The astrocytic ensemble acts as a multiday trace to stabilize memory. Nature 648, 146–156 (2025; published online 15 October).