Divisive normalization and neuronal oscillations in a single hierarchical framework of selective visual attention

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual Attention

Background:

• Divisive normalization models commonly use spike rate modulations to study covert attention.
• Top-down attention also increases neuronal oscillation synchronization, especially in gamma-band frequencies.
• Integrating spike rate and oscillation mechanisms into a unified attention framework is lacking.

Purpose of the Study:

• To develop a unified framework for attention by expanding the normalization model.
• To incorporate a multi-level hierarchy and time dimension to simulate attentional effects.
• To integrate oscillatory phase entrainment (communication-through-coherence) to maintain neuronal stimulus representations.

Main Methods:

• Expanded the normalization model of attention with hierarchical structure and time.
• Simulated a cascade of normalization models across cortical areas.
• Incorporated oscillatory phase entrainment (communication-through-coherence hypothesis).

Main Results:

• A simple cascade model led to signal degradation and reduced stimulus discriminability.
• Oscillatory phase entrainment prevented signal degradation, ensuring stable neuronal representations.
• The unified Hierarchical Normalization and Oscillation (HNO) model reproduced spatial and temporal attentional modulation aspects.

Conclusions:

• Divisive normalization and oscillation models can be complementary in explaining selective visual attention.
• The HNO model provides a unified account of neural mechanisms underlying attention.
• The model predicts a latency effect on neuronal responses due to cued attention.