Xin-Yue Yang^1,2,3,4, Fang Fang^1,2,3,4; ¹School of Psychological and Cognitive Sciences and Beijing Key Laboratory of Behavior and Mental Health, Peking University, Beijing, China, ²Key Laboratory of Machine Perception (Ministry of Education), Peking University, Beijing, China, ³Peking-Tsinghua Center for Life Sciences, Peking University, Beijing, China, ⁴IDG/McGovern Institute for Brain Research, Peking University, Beijing, China

Continuous training on a visual task improves corresponding visual abilities, but the mechanisms of post-training wakeful consolidation and how to intervene during this stage remain unclear. Previously, we demonstrated that repetitive grating stimulation at 10 Hz after training on an orientation discrimination task enhanced wakeful consolidation. This enhancement was absent without prior training or when repetitive uniform-disk stimulation was used. Additionally, augmented visual evoked potentials and increased glutamate-GABA concentration ratio in the visual cortex accompanied this enhancement, suggesting the involvement of long-term potentiation (LTP). However, the link between these findings and underlying neural circuit mechanisms remains unexplored. To address this, we developed a Habbian recurrent network model to replicate key findings through simulations. In the model, spiny stellate (SS) cells integrate inputs from thalamocortical neurons and project to superficial pyramidal (SP) neurons and parvalbumin-positive interneurons (SI). Recurrence is introduced as SPs and SIs mutually project onto both themselves and each other. All synapses are Hebbian-plastic, with LTP driven by repeated presynaptic contributions to postsynaptic firing. Orientation preferences were assigned using Gaussian initial weights, with reduced sigma values for the connections between neurons preferring the orientation used in visual training, a process known to sharpen tuning curves. Iterative orientational inputs simulated the repetitive sensory stimulation, with network properties statistically compared pre- and post-iterations. Results showed that both feedforward synaptic weights and SP tuning curves demonstrated enhanced orientation discrimination following the LTP simulation. The enhancement diminished without prior tuning-curve sharpening and was further weakened with uniform noise inputs, which simulate uniform-disk stimulation. Additionally, elevated glutamatergic presynaptic activity relative to GABAergic activity accompanied the enhancement. The model successfully reproduced behavioral and neurometabolic findings, supporting the hypothesis that LTP underlies enhanced wakeful consolidation induced by repetitive sensory stimulation, and offering insights that could guide future research into wakeful consolidation mechanisms.