Neural circuits solve planning by combining future time points with attractor dynamics
Researchers demonstrate that combining neural representations of different future time points with attractor dynamics allows circuits to solve planning problems.
The model achieves planning by embedding how the environment is structured directly into synaptic connections to create an attractor network that infers desirable futures.
Recurrent neural networks trained by gradient descent on planning tasks learn solutions that match the spacetime attractor in representation, dynamics, and connectivity.
Networks exposed to different environment structures show a mechanism that quickly reconfigures the world model for planning without requiring changes to synaptic strength.
Sources behind this report
- eLife
Research · CC BY 4.0
Source published
Read by the newsroom
Report history
No earlier revision dates are retained here. This does not establish that the report has never changed; read any dated updates and corrections.
This brief was made by AI and checked against its cited public sources. How the newsroom works