Heterogeneous self-couplings enable scale-invariant memory and cognitive flexibility without criticality

Heterogeneous self-interactions in a recurrent neural network generate scale-invariant memory in subcritical regimes, without assuming criticality, and promote spontaneous hierarchical organization.

• Why it matters: Power-law statistics are often linked to self-organized critical states, but it is unclear whether scale-invariant dynamics can arise without criticality while still supporting flexible computation.
• What they did: The authors analyze a recurrent neural network with heterogeneous self-couplings across nodes, then test short-term memory and the Wisconsin Card Sorting Test for cognitive flexibility and rule switching.
• The result: Heterogeneous self-interactions alone yield scale-invariant dynamics and comparable task performance to models using heterogeneous time constants, with more parsimonious synaptic modifications.

AI summary of the abstract · may contain errors · not medical advice · read the paper.