Sequential temporal anticipation characterized by neural power modulation and in recurrent neural networks
People and trained neural networks perform the same listening task to test how we anticipate important moments.
Dynamical systems, statistical models & neural networks
Computational methods
Computational methods turn ideas about the mind into models we can test. Dynamical systems, statistical models, and neural networks help us explore how behavior and brain activity change over time.
How it works
A computer model expresses an idea as a set of rules. We can run it, change its assumptions, and compare its predictions with observations. Statistical models help us describe patterns and uncertainty in measurements.
A dynamical system describes how something changes from one moment to the next: how activity settles into a stable state, repeats in a rhythm, or shifts after new information arrives. Neural networks offer another way to model learning, by adjusting connections between simple computing units. These approaches can also work together.
What it helps us ask
How might feedback stabilize a movement? What allows us to anticipate an important moment, or adapt when a rhythm changes? We compare model predictions with behavior and brain recordings, then change parts of a model to see what matters. A successful match gives us an explanation to test further; it does not mean the model works exactly like a brain.
See it in our work
Toward reliable signals decoding for electroencephalogram: A benchmark study to EEGNeX
EEGNeX compares neural-network designs for finding useful patterns in EEG recordings.