What if a fear memory could become broader or more specific without stress hormones being the thing that controls it? This study identifies a prefrontal circuit that separates memory specificity from memory strength. Forming memories of aversive experiences is adaptive for anticipating and responding to future threats.

The strength of the memory determines how robustly defensive behaviors are expressed, whereas memory specificity—the degree to which a memory is recalled in circumstances of varying similarity—optimizes how prior knowledge is used to minimize risks while maximizing opportunities. Fear memory generalization ensures defensive responses in similar threatening situations, whereas discrimination ensures animals do not display such behaviors in distinct safe situations.

The result is that animals do not display such behaviors in distinct safe situations, rather than treating those situations as dangerous. The standard model holds that stress hormones, particularly glucocorticoids, drive memory consolidation and promote generalization following aversive experiences.

Glucocorticoid administration enhances memory consolidation, while blocking glucocorticoid synthesis or receptor activation impairs it. The same brain regions implicated in learning and memory also regulate hypothalamo-pituitary-adrenal axis activation and glucocorticoid release.

When circuit manipulations alter both neural activity and glucocorticoid output, it remains unclear whether memory effects reflect hormonal changes, circuit mechanisms independent of hormones, or both. The prelimbic region of the medial prefrontal cortex regulates aversive memory and stress responses.

The rostral prelimbic region issues a prominent projection to the anteroventral bed nuclei of the stria terminalis, a pathway that coordinates behavioral and neuroendocrine responses to stress. Earlier work found that inhibiting avBST during memory consolidation enhanced memory strength and further elevated corticosterone, but blocking corticosterone synthesis prevented that memory enhancement.

So the rPL–avBST pathway was a candidate for modulating memory, but whether it acts through glucocorticoids or independent circuit mechanisms was unknown. The study used inhibitory avoidance discrimination learning in rats, where memory strength is reflected by latency to enter the dark compartment of the aversive context.

Memory specificity was measured by the contrast in latencies between the aversive context and a perceptually similar neutral one. The key question was whether the pathway modulates memory strength, specificity, or both, and whether those effects depend on glucocorticoids.

The reported result was that the rPL–avBST pathway bidirectionally modulates memory specificity without affecting memory strength, independently of glucocorticoids. Intersectional optogenetics selectively inhibited avBST-projecting rPL neurons immediately after inhibitory avoidance training, within the critical window for glucocorticoid-dependent memory consolidation.

Rats received retrograde AAV-Cre in avBST and Cre-dependent halorhodopsin or mCherry control in rPL, with bilateral optical fibers over avBST. In vivo optrode recordings confirmed that five hundred sixty-one-nanometer illumination suppressed avBST neuronal activity throughout the ten-minute stimulation period.

On day one, rats received inhibitory avoidance training with a zero point eight milliampere, one-second footshock, and immediately afterward the pathway was inhibited for ten minutes with laser illumination. On day three, retention was tested in the aversive context and a similar neutral context, using latency to enter the darkened compartment as the retention index.

rPL–avBST Halo rats showed prolonged corticosterone elevation relative to controls, extending through sixty minutes after training, with significantly greater total corticosterone output. During testing, control rats discriminated between contexts, showing high latency in the aversive context and low latency in the neutral context.

rPL–avBST Halo rats instead showed high latencies in both contexts, indicating generalization to the neutral context and significantly lower discrimination indices. Latencies to the aversive context were identical between groups, demonstrating that pathway inhibition affected memory specificity but not strength.

In panel F, the magenta Halo curve separates from the black control curve after the ten-minute sample and remains visibly elevated at the thirty-, sixty-, and ninety-minute timepoints, with an asterisk marking the sixty-minute comparison. Panel G shows the same pattern in the integrated area-under-the-curve bars, where the magenta bar is visibly taller and individual animal points are overlaid.

The augmented post-training corticosterone response and resulting generalization phenotype suggested that footshock-evoked glucocorticoids drove the mnemonic effect. The next test asked whether elevated corticosterone was sufficient to promote generalization.

Corticosterone administered immediately after inhibitory avoidance training elevated retention latencies to the neutral context relative to vehicle controls when tested two days later. During testing, sham plus mCherry and adrenalectomized plus mCherry controls discriminated successfully, showing high latencies to the aversive context and low latencies to the neutral context.

Adrenalectomized plus Halo rats failed to discriminate, displaying high latencies to both contexts—an eightfold increase in neutral-context latency relative to both control groups. Adrenalectomized plus Halo rats had significantly lower discrimination indices than all other groups.

Aversive-context latencies remained unchanged across all groups, demonstrating selective effects on specificity, not strength. These findings demonstrate glucocorticoid-independent control of memory specificity by the rPL–avBST pathway. In panel F, the magenta ADX-plus-Halo bar sits below the dashed zero-point reference and is visibly separated from the two control bars, which cluster near the upper part of the scale.

The individual points show the same group-level pattern, with the magenta observations concentrated at lower discrimination values. After showing that rPL–avBST inhibition promotes memory generalization independently of glucocorticoids, the study tested whether pathway excitation enhances memory specificity.

A modified inhibitory avoidance generalization task pre-exposed rats to the neutral context for one minute without footshock, then delivered a zero point eight milliampere, two-second footshock in the aversive context. This neutral-context pre-exposure combined with the longer footshock duration biases subsequent memory toward generalization.

In control rats, this task reliably produces high retention latencies to both contexts during testing. Post-training excitation did not affect corticosterone output, while control rats displayed high retention latencies to both contexts, indicating generalization as expected.

rPL–avBST ChR2 rats maintained low latencies to the neutral context and high latencies to the aversive context. Discrimination indices were significantly higher in the ChR2 group than in mCherry controls, indicating enhanced discrimination between the two contexts.

When the same cohort was retested on day twenty-eight, rPL–avBST ChR2 rats maintained the discrimination pattern, though to a lesser degree, while mCherry controls continued to show comparable high latencies to both contexts. Discrimination indices remained higher in the ChR2 group than in mCherry controls, though the difference did not reach statistical significance, with a p value of zero point one.

The attenuated difference primarily reflected a time-dependent reduction in aversive-context latencies in the ChR2 group rather than increased generalization to the neutral context. Panel L shows the corresponding ChR2 discrimination bar still visibly above the dashed zero-discrimination line, whereas the mCherry bar is closer to it.

IL–avBST excitation did not affect training latencies or post-training corticosterone output. During testing, IL–avBST ChR2 rats showed reduced latencies to the neutral context and improved discrimination indices compared with mCherry controls, while aversive-context retention latencies were comparable between groups.

Together, these findings establish bidirectional, glucocorticoid-independent control over memory specificity during consolidation, with convergent effects of IL–avBST excitation suggesting a broader role for prefrontal inputs to avBST in shaping context discrimination.

In panel F, the ChR2 group’s neutral-context points cluster near the bottom of the axis, while its aversive-context points remain near the upper range, producing a visibly wider separation than in the mCherry group. Panel G shows the corresponding group mean rising from roughly one-half to around four-fifths of the discrimination scale, with an asterisk marking the comparison.

The study complemented the causal manipulations with neural activity correlates of discrimination, first using Fos immunohistochemistry after inhibitory avoidance discrimination testing. On day three, rats showed expected discrimination, with high retention latencies to the aversive context and low latencies to the neutral context.

The number of Fos-immunoreactive cells in avBST ninety minutes after testing positively correlated with discrimination indices, whereas Fos counts in adjacent anterodorsal BST showed no correlation with discrimination. Additional analyses confirmed that avBST Fos tracks context-specific discrimination rather than latency more generally.

In panel E, the orange avBST points follow a clear upward regression, with r equal to zero point seven one and p equal to zero point zero zero two, while the gray adBST fit is nearly flat. The Fos images above show the corresponding contrast visually: sparse labeled cells in C and noticeably denser labeling in D.

Fiber photometry recorded calcium transients in avBST as a proxy for neuronal activity, using jGCaMP8s expression and optical fibers implanted in avBST. The recorded rats showed high retention latencies to the aversive context and low latencies to the neutral context, confirming discrimination behavior.

Calcium transients in avBST differed between contexts around light-to-dark entry, with greater responses during the neutral than the aversive context. Bootstrap analysis confirmed that neutral-context calcium responses were elevated from four seconds before to eight seconds after light-to-dark entry, with no comparable elevation in the aversive context.

Across a twenty-second window centered on light-to-dark entry, activity showed a context-specific pattern, with greater neutral-context responses concentrated around the transition. Peak calcium responses were higher in the neutral than the aversive context.

This context-specific engagement of avBST during retrieval paralleled behavioral discrimination and was sustained beyond compartment entry, rather than being restricted to the transition itself. Figure six first shows the intersectional strategy used to record rPL neurons that project to the avBST, with fluorescent labeling confirming expression and axon targeting.

Behaviorally, rats showed longer retention latency in the aversive context than the neutral context, indicating discrimination. During retrieval, these projection neurons displayed context-dependent calcium activity, with stronger responses in the neutral context across the light-to-dark transition, linking the rPL–avBST pathway to memory specificity.

The findings reveal a distinct, glucocorticoid-independent mechanism by which the rPL–avBST pathway controls memory specificity. The pathway bidirectionally modulates contextual discrimination during the post-training consolidation window: inhibition promotes generalization, whereas excitation enhances discrimination.

These effects are dissociated from glucocorticoid output and memory strength, showing that circuits regulating stress hormone release can shape memory through mechanisms operating in parallel to, rather than via, hormonal signaling. Fiber photometry further revealed context-specific activity in both avBST and avBST-projecting rPL neurons during memory retrieval, consistent with continued engagement in context discrimination beyond the consolidation window.

The prefrontal cortex has established roles in fear acquisition, expression, and extinction, although its role in consolidation is less well characterized. The findings identify a distinct role for the rPL–avBST pathway in shaping memory specificity during the post-training consolidation window.

Optogenetic inhibition of the pathway promoted generalization, whereas excitation enhanced discrimination. These effects were specific to the immediate post-training window, with manipulations delivered three hours after training producing no change in memory specificity.

Although avBST-projecting rPL neurons collateralize to vlPAG, post-training excitation of the rPL–vlPAG pathway did not influence subsequent generalization, indicating that the consolidation effects are specific to the rPL–avBST arm. Fiber photometry showed context-specific activity in avBST-projecting rPL neurons during retention testing, with greater responses in the neutral than the aversive context.

These findings identify boundary conditions on the standard model: under specific circumstances, circuit-level mechanisms can shape memory consolidation through pathways operating in parallel to, rather than via, glucocorticoid signaling. The findings do not diminish the broader importance of glucocorticoids in mobilizing energy and coordinating the body's response to stressors.

It remains to be determined why a pathway controlling memory specificity also regulates glucocorticoid output, and whether these glucocorticoid-independent mechanisms generalize to other forms of aversive learning. Exogenous corticosterone promoted generalization in this task, preserving the relevance of glucocorticoid-targeting approaches while underscoring the need for circuit-level interventions.

The rPL–avBST pathway bidirectionally controls whether an aversive memory generalizes or discriminates, independently of glucocorticoids and without changing memory strength. That separates circuit control from the standard hormone-centered model.