Abstract
Attention greatly influences sensory neural processing by enhancing firing rates of neurons that represent the attended stimuli and by modulating their tuning properties. The cholinergic system is believed to partly mediate the attention contingent improvement of cortical processing by influencing neuronal excitability, synaptic transmission and neural network characteristics. Here, we used a biophysically based model to investigate the mechanisms by which cholinergic system influences sensory information processing in the primary visual cortex (V1) layer 4C. When including a cholinergic drive, we found significant sharpening in orientation selectivity, desynchronization of LFP gamma power and spike-field coherence, decreased response variability and correlation reduction mostly by influencing intracortical interactions and by increasing inhibitory drive. Our results indicated that these effects emerged due to changes specific to the behavior of the inhibitory neurons. The behavior of our model closely resembles the effects of attention on neural activities in monkey V1, suggesting precise mechanisms through which cholinergic modulation may mediate the effects of attention in the visual cortex.
Introduction
While it is well-established that attention improves sensory processing—allowing us to "filter" the world—the underlying biological machinery remains a subject of intense debate. Specifically, researchers have long looked at the cholinergic system, the network of neurons using the neurotransmitter acetylcholine (ACh), as a primary mediator of attention.
However, experimental results have been inconsistent. Some studies suggest ACh sharpens the "tuning" of neurons (making them more selective to specific stimulus features like orientation), while others suggest it merely scales up their firing rates. Furthermore, the circuit-level mechanisms—how individual synaptic changes lead to large-scale network desynchronization and reduced noise—are difficult to observe directly in living subjects. This study uses computational modeling to reconcile these contradictions and map out the specific roles of excitatory and inhibitory neurons in the attentive brain.
Methodology
The researchers constructed a recurrent spiking ring network model representing layer 4C of the macaque primary visual cortex. The model consists of 1,000 excitatory pyramidal neurons and 250 inhibitory interneurons (a biologically realistic 4:1 ratio), arranged in a ring where each neuron has a preferred stimulus orientation. Connectivity follows a "Mexican hat" profile—strong local excitation and broader lateral inhibition.
Cholinergic modulation was modeled biophysically by reducing the conductance of slow potassium currents ($g_{KS}$), which increases neuronal excitability, and modulating synaptic efficacy (strengthening thalamo-cortical inputs while suppressing certain intra-cortical connections). As a modeling study, the parameters (e.g., membrane capacitance, synaptic time constants) were derived from established electrophysiological literature on the macaque V1 to ensure biological validity.
Results and Discussion
Sharpening of Orientation Selectivity
A key finding was the significant sharpening of orientation tuning curves. Orientation tuning widths were sharpened by 19.9% under high ACh conditions and 12.7% under low ACh conditions.
Interestingly, this sharpening was almost entirely dependent on the inhibitory neurons. When cholinergic drive was applied only to excitatory neurons, firing rates increased, but selectivity did not improve. Selectivity only sharpened when the inhibitory drive was augmented, highlighting the critical role of inhibition in attentional modulation.
Network Desynchronization and Reliability
ACh significantly reduced gamma-band (30–50 Hz) oscillations and spike-field coherence (SFC). In neural coding, a desynchronized state is often more efficient for representing complex information because it reduces redundant noise across the population.
The model also showed a significant decrease in the Fano factor, indicating that ACh makes neural responses more reliable and predictable across trials. Furthermore, ACh reduced "noise correlations," effectively increasing the signal-to-noise ratio of the entire population.
Reconciling the "Gain vs. Sharpening" Debate
The model suggests that conflicting results in previous literature might be due to different baseline states or concentrations of ACh. The study found that ACh's effects are significantly more pronounced when the network's baseline activity is high, providing a potential explanation for why some experiments see simple gain while others see tuning changes.
Conclusion
This biophysically grounded model demonstrates how acetylcholine replicates the effects of attention in V1. By showing that cholinergic modulation sharpens neuronal orientation tuning and desynchronizes network activity primarily by enhancing inhibitory drive, this work provides a mechanistic explanation for how neuromodulators bridge the gap between cellular changes and cognitive behavior. Future research could expand this model to include more diverse cell types and investigate the temporal dynamics of ACh release relative to stimulus presentation.
Summary updated