[Paper Review] Attentional modulation in layer 4 of the visual cortex could be mediated by interneurons with complex receptive field characteristics

Paul Tiesinga, Calin Buia|ArXiv.org|Nov 8, 2006

Neural dynamics and brain function56 references3 citations

TL;DR

The paper proposes that attentional modulation in layer 4 of the visual cortex arises from distinct populations of inhibitory interneurons: complex RF interneurons that increase firing with attention and simple RF interneurons that decrease firing, with the former driving network synchrony via gamma-band oscillations. This mechanism enables contrast-invariant orientation tuning and attention-induced increases in eLFP coherence without changing excitatory cell firing rates.

ABSTRACT

Many neurons in the visual cortex are orientation-selective, increase their firing rate with contrast and are modulated by attention. What is the cortical circuit that underlies these computations? We examine how synchrony can be modulated by the excitability of interneurons, in a model layer 4 network displaying contrast-invariant orientation-tuning. We did not find parameter settings for which the standard ring model (Somers et al, 1995), which contains only cells with simple receptive fields (RF), behaved appropriately. Only when interneurons with complex receptive fields were included, similar to those found recently in cat primary visual cortex (Hirsch et al, 2003), did the network behave appropriately. A critical feature in the model was that complex interneurons projected to simple interneurons but the simple interneurons did not project back to them. The network was switched from the non-attended state to the attended state by increasing the depolarization of the complex interneurons. In addition to contrast-invariant orientation tuning, the model reproduced the following experimental results: (1) the gamma-frequency range coherence between the estimated local field potential (eLFP) and spike trains of excitatory cells was higher in the attended state than in the non-attended state, but the firing rates of the excitatory cells did not vary between states; (2) the gamma-frequency-range power in the eLFP increased with contrast. The model predicts that there are two populations of inhibitory cells, one with complex RF characteristics whose firing rate increases with attention and the other with simple RF characteristics whose firing rate decreases with attention.

Motivation & Objective

• To determine the cortical circuit mechanism underlying attentional modulation of orientation-selective neurons in layer 4 of the visual cortex.
• To investigate why standard ring models with only simple inhibitory interneurons fail to reproduce attentional modulation.
• To test whether interneurons with complex receptive fields—recently observed in cat V1—can enable attention-dependent network synchrony.
• To predict differential attention-dependent firing rate changes in two interneuron populations: complex RF interneurons increase, while simple RF interneurons decrease with attention.

Proposed method

• Constructed a biologically constrained ring model of layer 4 cortical microcircuit with 21 columns, each containing 84 excitatory and 21 inhibitory neurons.
• Incorporated LGN inputs with orientation-tuned receptive fields via 3x1 subfields aligned with cortical neuron preferences, using realistic synaptic weights and delays.
• Modeled two types of inhibitory interneurons: simple RF (SI) and complex RF (CI), with distinct connectivity: CI → SI only, not reciprocal.
• Simulated attention by increasing depolarization of CI interneurons, mimicking top-down or subcortical input.
• Measured network dynamics via spike trains, estimated local field potential (eLFP), and coherence between eLFP and spike activity.
• Used parameter sweeps to compare network behavior in non-attended (asynchronous) vs. attended (synchronous) states, assessing gamma-band power and coherence.

Experimental results

Research questions

• RQ1 Why do standard ring models with only simple inhibitory interneurons fail to reproduce attentional modulation of synchrony?
• RQ2 Can interneurons with complex receptive fields enable attention-dependent modulation of gamma-band synchrony in layer 4?
• RQ3 How do the firing rates of simple and complex inhibitory interneurons change with attention in this network?
• RQ4 Does attention increase coherence between excitatory spiking and eLFP without altering excitatory firing rates?
• RQ5 How does contrast affect gamma-band power in the eLFP, and is this consistent with experimental data?

Key findings

• The model only reproduced attentional modulation when interneurons with complex receptive fields (CI) were included, particularly when they projected exclusively to simple interneurons (SI).
• Increasing depolarization of CI interneurons switched the network from a non-attended to an attended state, inducing gamma-band synchrony without changing excitatory cell firing rates.
• In the attended state, coherence between eLFP (from CI cells) and excitatory spike trains increased with contrast, matching experimental observations.
• Gamma-band power in the eLFP increased with stimulus contrast, consistent with in vivo recordings in V1.
• The model predicted that attention increases firing rates in complex RF interneurons while decreasing them in simple RF interneurons.
• Contrast-invariant orientation tuning was maintained due to a balance between increased inhibition from CI to excitatory cells and decreased inhibition from SI to excitatory cells with attention.