Animals solve new, complex tasks by reusing and adapting prior knowledge. This flexibility depends not only on the content of experience but also on its structure. Early training curricula are especially important: poorly structured experiences can hinder abstraction and limit generalization. However, the neural mechanisms through which experience shapes future learning remain unclear. Here, we trained recurrent neural networks (RNNs) on an odor timing task used to study complex timing behavior in mice and then tested the model predictions with mouse behavior and medial entorhinal cortex recordings. Without structured early experience, both RNNs and mice developed rigid, error-prone strategies, whereas structured training promoted neural activity reflecting the task's temporal structure. Using dynamical systems analysis, we examined how different training curricula shaped network dynamics and whether these dynamics supported abstraction and generalization as task complexity increased. These findings demonstrate that the structure of prior experience governs how flexible, generalizable knowledge emerges in biological systems and computational models.
John C. Bowler, Dua B. Azhar, Cambria M. Jensen et al.· Nature Neuroscience· 0 citations
The medial entorhinal cortex (MEC) has classically been viewed as a spatial coding region, but growing evidence indicates that it also contains temporal representations. However, how spatial and temporal codes are organized within the MEC remains unclear. Here we recorded MEC neurons with high-density silicon probes while mice performed a MEC-dependent timing task and a virtual reality spatial navigation task in a similar head-fixed setup. We found that spatial and temporal representations were partially overlapping but systematically biased across the MEC population. Grid cells and non-grid cells with strong spatial tuning were less likely to show reliable time-locked activity during the timing task. In contrast, neurons with weaker spatial tuning more flexibly shifted their coding scheme to temporal coding during timing task. Moreover, spatial tuning strength and its negative relationship with temporal tuning were preserved in a distinct open field environment, indicating that the coding preferences of individual neurons are constrained by a stable network-level organization. Together, these findings suggest MEC is organized along a coding gradient, ranging from dedicated stable spatial coding neurons to more flexible spatial or temporal coding neurons which represent information according to cognitive demands.
Hyun-Woo Lee, John C. Bowler, James G. Heys· bioRxiv· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
