Researchers at MIT found that roughly a quarter to a third of all synapses in the adult mouse cortex are functionally silent, lacking the receptors needed to transmit signals under normal conditions. These dormant connections sit on thin, finger-like protrusions called filopodia, and the team’s data suggest they form a hidden reserve the brain can activate when it needs to learn something new. The finding challenges a long-held assumption that silent synapses are mainly a feature of early development, and it raises a provocative question: does the adult brain keep a large stock of unused wiring precisely so it can rewire without building new connections from scratch?
What is verified so far
The core dataset comes from MIT neuroscientists Dimitra Vardalaki, Kwanghun Chung, and Mark Harnett, who used a tissue-expansion technique called eMAP combined with super-resolution protein imaging to examine individual synapses in the primary visual cortex of adult mice. According to their Nature study, the team measured 2,234 synapses on layer 5 pyramidal neurons and found that approximately 25% lacked AMPA receptors, the proteins required for standard excitatory signaling. Without AMPA receptors, a synapse can possess the structural hardware of a connection but remain electrically quiet during routine neural activity.
An MIT institutional summary of the work described the proportion as “about 30 percent of all synapses in the brain’s cortex,” a figure that rounds up from the raw receptor counts and extends the claim beyond the single cortical layer the team directly sampled. The study also reported that many of these silent synapses were perched on filopodia, slender extensions that are structurally distinct from the mushroom-shaped dendritic spines where most active synapses reside. That structural detail matters because filopodia are typically associated with developing brains, not adult tissue, which is part of what made the finding unexpected.
Vardalaki, Matthew Yaeger, and Harnett later authored a review in the Annual Review of Neuroscience that placed the 2022 results inside a broader evidence base. In that synthesis article, the authors argue that silent synapses represent a latent reservoir of plasticity in adulthood, a pool of connections that can be “switched on” through activity-dependent processes involving NMDA receptors and downstream signaling cascades. The review notes that this reservoir could help reconcile how adult brains manage to learn rapidly without constantly growing and pruning large numbers of new synaptic structures.
The mechanistic picture centers on the interplay between AMPA and NMDA receptors. At silent synapses, NMDA receptors are present and can respond to glutamate, but they are blocked by magnesium ions at resting membrane potentials. When a postsynaptic neuron is sufficiently depolarized-typically by activity at neighboring synapses-the magnesium block is relieved, calcium flows through NMDA channels, and intracellular signaling pathways are triggered. Those pathways can rapidly insert AMPA receptors into the postsynaptic membrane, converting a previously silent contact into an active one. This AMPA insertion process is a well-established model for how long-term potentiation (LTP) strengthens specific synapses during learning.
What remains uncertain
The most visible tension in the data is the gap between the two headline numbers. The Nature paper’s direct measurements put the silent fraction at roughly 25% of 2,234 synapses in one cortical layer. The MIT press description generalized that figure to “about 30 percent” across the cortex. Whether the proportion holds in other cortical areas, deeper layers, or different cell types has not been confirmed by the same imaging method. Readers should treat the 30% figure as an extrapolation rather than a direct measurement.
All of the synapse counts come from adult mice, not humans. No primary human cortical tissue counts or receptor maps have been published using the same eMAP technique, so the relevance of these proportions to the human brain remains an open question. Similarly, the study did not include behavioral or learning-task experiments that would directly link the activation of filopodia-based synapses to specific memory formation. The structural evidence is strong, but the functional proof, showing the same silent synapses waking up during a defined learning event in a living animal, has not yet been reported in these papers.
There is also a definitional wrinkle. “Silent synapse” can refer to two different biological mechanisms. The MIT work focuses on postsynaptic silence, where the receiving side of the connection lacks AMPA receptors. But synapses can also be silenced presynaptically, when the sending terminal fails to release neurotransmitter vesicles. These two forms of silence likely serve different functions, and the relative prevalence of each within the same circuits has not been quantified in the MIT dataset. As a result, the overall landscape of silent connectivity in the adult cortex remains only partially mapped.
Another open issue is how stable filopodia-based synapses are over time. The structural snapshots provided by eMAP reveal many filopodia in adult tissue, but they do not show whether those structures persist for weeks, appear transiently during specific experiences, or fluctuate with neuromodulatory states such as stress or sleep. Longitudinal in vivo imaging, combined with functional readouts, will be needed to determine whether the same silent synapses are repeatedly recruited or whether the brain continually generates and prunes this reserve pool.
How to read the evidence
The strongest piece of evidence is the Nature paper itself, which provides direct protein-level imaging of individual synapses and receptor distributions. That dataset is reproducible in principle because the eMAP expansion protocol and imaging pipeline are described in detail, and the authors report clear criteria for classifying synapses as AMPA-positive or AMPA-negative. The Annual Review of Neuroscience article by the same MIT authors is a secondary synthesis; it offers interpretation and context but does not add new experimental data. Both publications are peer-reviewed and represent the highest-quality sources currently available on this topic.
Supporting work from other labs adds plausibility. Research in the anterior cingulate cortex of adult mice has shown that previously weak or undetectable synaptic responses can be recruited pharmacologically through pathways involving cyclic AMP, with a dependence on calcium-permeable AMPA receptors. Those results, obtained in a distinct cortical region, suggest that the capacity to awaken dormant synapses is not confined to the visual cortex. Earlier hippocampal experiments in slice preparations and cultures demonstrated that synapses initially lacking AMPA-mediated currents can acquire them after patterned stimulation, reinforcing the idea that postsynaptically silent contacts are a general feature of excitatory circuits.
At the same time, readers should resist overextending the conclusions. The current evidence firmly supports the existence of a substantial population of AMPA-lacking synapses on filopodia in the adult mouse visual cortex and indicates that such synapses are poised to become functional under the right conditions. It does not yet establish how many of these synapses are actually recruited during everyday learning, how their prevalence varies across species, or whether manipulating them would enhance or disrupt cognitive performance. Until longitudinal and behavioral studies catch up with the structural work, claims about a “hidden reserve” of learning capacity in the adult human brain should be viewed as informed hypotheses rather than settled facts.
For now, the most cautious reading is that adult cortical circuits are more structurally flexible than previously assumed. Instead of relying solely on gradual strengthening and weakening of already-active synapses, the brain appears to maintain a stockpile of silent connections that can be rapidly enlisted when new information must be encoded. If future research confirms that this mechanism operates broadly across regions and species, it could reshape theories of how learning, memory, and recovery from injury proceed in mature nervous systems-and highlight silent synapses as potential targets for therapeutic intervention.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
