Neuroscience · Cognitive Neuroscience
A spiking neural network model of the retina was trained either to represent incoming images or to predict scenes roughly 100 milliseconds into the future. The prediction-focused model reproduced several known retinal responses and matched how retinal output cells from different animal species respond to natural images and movies.
The structure of early training influenced how flexibly mice learned a complex odor-timing task. Recurrent neural networks showed similar effects, suggesting that well-structured practice helps the brain build knowledge that can be reused when tasks become more complex.
Monkeys learned to link arbitrary visual patterns with movements of the left or right hand. As learning progressed, Purkinje cells in two cerebellar regions increasingly distinguished specific visual cue–hand combinations, with the visual cue becoming identifiable before the choice.
A whole-brain model of 90 anatomically connected regions produced a wider range of activity patterns when each region included interacting excitatory and inhibitory cell populations. The researchers also found that the brain’s modeled wiring helped produce fast gamma rhythms whose amplitude changed with slower rhythms.
In trained mice, deep-layer neurons in the auditory cortex developed categorical responses to sounds that differed in their temporal rate. These neurons also carried strong signals related to the animals’ choices, linking sound classification with behavior.
Brain stimulation separated motion-discrimination accuracy from confidence in perceptual decisions. Strengthening one visual pathway affected both measures, while disrupting a parietal brain region changed confidence-related performance without altering motion discrimination.
In male mice, brief bursts of neuronal activity tracked which sensory cues predicted rewards and changed as those associations changed. Suppressing the bursts disrupted learning and task performance, providing causal evidence that they help drive associative learning.
Early blockage of NMDA receptors in male rats led to long-lasting changes in inhibitory interneurons in the visual cortex, including fewer parvalbumin-positive cells. When the rats later lived in enriched environments, parvalbumin cell counts rose and several plasticity-related protein markers increased, while NMDA receptor NR1 levels did not recover.
