Fear prioritizes feedforward sensory coding at the expense of contextual processing in monkey V1

The primary visual cortex (V1) supports both rapid feedforward coding of local features and contextual processing shaped by recurrent and top-down feedback. While fear learning is known to prioritize threat-predictive signals in V1, it remains unclear whether this prioritization uniformly enhances all stages of cortical processing or selectively biases certain computations. We investigated how fear conditioning modulates V1 population coding for simple gratings versus global contours, which rely on distinct temporal stages of integration.

Monkeys were trained in a visual fear-conditioning paradigm. We utilized chronic microelectrodes to record V1 population activity while monkeys viewed conditioned and nonconditioned stimuli. Using a Fisher linear discriminant classifier, we quantified stimulus discriminability ( d ') during early and late response epochs to distinguish between feedforward and feedback-driven coding.

Fear conditioning produced a striking temporal and stimulus-specific dissociation. For gratings, conditioning significantly enhanced orientation decoding during the early response epoch, consistent with a rapid retuning of feedforward encoding. This early facilitation was also present for contour stimuli involving contextual integration. By contrast, during the late response epoch, fear conditioning significantly reduced decoding performance for contours, while leaving grating discrimination largely unaffected.

Fear conditioning does not act as a global gain control. Instead, it reconfigures V1 dynamics to prioritize rapid, feedforward threat detection at the functional cost of later, recurrent- and feedback-dependent contextual processing. This shift suggests an adaptive trade-off: emotional learning sharpens the detection of isolated predictive cues but reduces the fidelity of the complex spatial integration required for scene analysis.