Learning is not only a matter of strengthening activity in one brain region. Goal-directed behavior depends on distributed networks linking cortex, basal ganglia, thalamus, hippocampus and other structures. How these large-scale interactions reorganize as an animal learns a task remains a central question in systems neuroscience.
In collaboration with Fritjof Helmchen and colleagues at the University of Zurich, we combined high-density multi-fiber photometry with information-theoretic analyses to study how large-scale brain networks reorganize during learning. By recording activity across multiple cortical and subcortical regions while mice acquired a tactile discrimination task, we could follow how communication within cortico-basal ganglia-thalamo-cortical circuits progressively shifted from reward-related responses toward stimulus-driven, task-relevant dynamics.
A striking result is that, as performance improved, many brain regions shifted their peak activity in time. Early in learning, activity was more strongly aligned with reward-related action. With learning, peak activity moved toward the reward-predicting stimulus. In other words, the network gradually reorganized from reacting to outcomes toward anticipating task-relevant information.
To quantify directed interactions between regions, the authors used transfer entropy. This revealed that functional networks spanning basal ganglia, thalamus, neocortex and hippocampus grew and stabilized with learning, especially around stimulus presentation. The internal globus pallidus, ventromedial thalamus and several frontal cortical areas emerged as important hub regions within this learning-related network organization.
The broader message is that learning is not simply the acquisition of a stimulus-response association. It is a dynamic network process. As the animal becomes better at the task, the brain establishes more stable and task-appropriate mesoscale interactions, allowing sensory information, action selection and reward expectation to be coordinated more efficiently.
A useful way to summarize the paper is: learning reshapes not only what individual brain regions encode, but when and how they communicate.
This work highlights the value of multi-region recordings for understanding behavior. By combining dense optical recordings with information-theoretic measures of inter-regional influence, the study shows how distributed brain networks progressively organize themselves during task learning.
To know more:
- Sych, Y., Fomins, A., Novelli, L., & Helmchen, F. (2022). Dynamic reorganization of the cortico-basal ganglia-thalamo-cortical network during task learning. Cell Reports, 40, 111394. https://doi.org/10.1016/j.celrep.2022.111394.