Plain wording

Shock expectancy ranked the cues by their current danger in every phase, and followed the cues through the reversal.

slug behavioral-learning-confirms-contingenciesrole empiricalpanel fig2A

Patterns for the two contexts grew more distinct during reversal than during acquisition, most clearly in prefrontal cortex.

slug context-specificity-increases-reversalrole empiricalpanel fig5B

Sharper prefrontal context coding revives the once-dangerous cue's acquisition pattern in ACC, the once-safe cue's in precuneus.

slug context-specificity-predicts-acquisition-reinstatement-regio…role empiricalpanel fig5Di

Even in wholly new contexts, sharper prefrontal context coding predicts reviving the newly dangerous cue in middle temporal gyrus.

slug context-specificity-predicts-reinstatement-new-context-mtgrole controlpanel fig5Diii

In the old context, sharper prefrontal context coding predicts dmPFC reviving the newly dangerous cue's reversal pattern.

slug context-specificity-predicts-reversal-reinstatement-dmpfcrole empiricalpanel fig5Dii

During acquisition, threatening cues drive more activity than safe cues in dACC, superior frontal gyrus and caudate nucleus.

slug cs-plus-univariate-fear-network-acquisitionrole empiricalpanel fig2Bi

During fear acquisition, different threatening cues come to share one pattern in the fear network instead of staying distinct.

slug cue-generalization-increases-acquisitionrole empiricalpanel fig3A

After reversal, the shared pattern for consistently threatening cues survives only in dACC, not across the wider fear network.

slug cue-generalization-limited-dacc-reversal-consistentrole empiricalpanel fig3Bi

After reversal, currently dangerous cues drive more fear-network activity than currently safe ones, whatever they meant before.

slug current-threat-activates-fear-network-reversalrole empiricalpanel fig2Bii

Back in the original context, dmPFC revives category-level patterns from acquisition rather than from the new-context test.

slug dmpfc-reinstates-acquisition-traces-generalizedrole empiricalpanel fig4Bii

Reversal learning lumps the currently dangerous cues together while keeping the cue whose threat value changed separate.

slug dual-strategy-reversal-generalize-plus-specifyrole synthesispanel fig4 (synthesis)

After reversal, the newly dangerous cue takes on the same shared fear-network pattern the originally feared cue had.

slug generalized-pattern-cs-minus-plus-reversalrole empiricalpanel fig3Bii

Prefrontal cortex codes contexts as distinct patterns, and that distinctness lets an old fear memory return in its own context.

slug hypothesis-context-specificity-supports-renewalrole hypothesispanel hypothesis

The fear network lumps currently dangerous cues together while higher-order cortex keeps the cues whose meaning changed apart.

slug hypothesis-dual-strategy-reversalrole hypothesispanel hypothesis

The fear network codes dangerous cues as a class, so any cue that becomes dangerous inherits the class representation.

slug hypothesis-fear-network-generalizes-threat-cuesrole hypothesispanel hypothesis

Inferior frontal gyrus revives cue-specific patterns from the most recent learning, the reversal phase, rather than earlier ones.

slug ifg-reinstates-reversal-traces-item-specificrole empiricalpanel fig4Bi

Individual cues keep their own stable patterns in temporal cortex through the shock-free tests, while class patterns do not.

slug item-stability-persists-test-phasesrole empiricalpanel fig3C, fig3D

During reversal, precuneus and prefrontal cortex hold a distinct, stable pattern for the cue whose threat value changed.

slug item-stability-precuneus-pfc-reversalrole empiricalpanel fig4

All pattern-similarity analyses use only trials without a shock, dropping the trials likely to carry the strongest fear responses.

slug lss-unreinforced-trials-onlyrole methodologicalpanel methods (RSA section)

In the two test phases no cue drove more overall activity than another, though participants still expected shocks differently.

slug no-bold-differences-test-phasesrole controlpanel fig2B (test phases)

During acquisition, a cue's own pattern was no more stable for threatening than for safe cues anywhere in the brain.

slug no-item-stability-difference-acquisitionrole empiricalpanel fig3A

Participants whose prefrontal cortex coded contexts more distinctly during reversal showed more fear renewal at test.

slug pfc-context-specificity-predicts-renewalrole empiricalpanel fig5

Prefrontal context patterns should grow more distinct during reversal than acquisition, when context begins to predict danger.

slug prediction-context-specificity-increases-during-reversalrole predictionpanel prediction

During reversal the newly dangerous cue should take on the same shared fear-network pattern the originally feared cue acquired.

slug prediction-cs-minus-plus-generalizes-reversalrole predictionpanel prediction

Threatening cues should share patterns more than safe cues do, in dACC, superior frontal gyrus, caudate and insula.

slug prediction-cue-generalization-fear-network-acquisitionrole predictionpanel prediction

Cues whose threat value changed should hold more stable individual patterns in precuneus, IFG and prefrontal cortex.

slug prediction-item-stability-changing-cues-reversalrole predictionpanel prediction

Participants whose prefrontal cortex codes contexts more distinctly should revive acquisition fear memories more strongly at test.

slug prediction-pfc-context-specificity-tracks-renewalrole predictionpanel prediction

Cues that were dangerous before but are safe now still stir the fear network weakly, a trace of the original fear memory.

slug prior-threat-activates-fear-network-weaklyrole interpretationpanel fig2Biii

The regions tested in the follow-up analyses were picked from significant clusters in the same data, which inflates significance.

slug rsa-roi-derived-from-searchlightrole methodologicalpanel fig4A (ROI definition procedure)

The study uses trace conditioning, a gap between cue and shock, assuming from prior work that this engages the hippocampus.

slug trace-conditioning-hippocampus-engagedrole scopepanel fig1 (methods)