Abstract
fMRI has been widely used for basic and clinical studies. While cortical neural activity varies with changes of cortical depth of less than 0.2 mm, the BOLD signal averages over voxels of 1–3 mm on a side. This leads to an open question: does BOLD pool brain activity evenly from all layers in a cortical region or it is dominated by signals from some particular layers?
We tried to answer this question by measuring the laminar patterns of Gamma-band activity in Macaque primary visual cortex (V1). The rationale was: if gamma-band activity in the local field potential (LFP) is related to BOLD, as several studies (Logothetis, 2001 & Goense and Logothetis, 2008) have claimed, the laminar pattern of gamma activity should indicate which cortical layers contribute to the BOLD signal.
We measured the LFP in V1 visually driven by a large patch (2–3 deg in radius) of sinusoidal grating drifting in different directions (0 to 360 deg, 30 deg intervals) at high contrast (99%) as well as at zero contrast (blank screen). With track reconstruction, we identified the cortical depth for each recording site, and then studied the visually-driven activity in the gamma-band as a function of cortical depth.
From 315 recording sites in 6 anaesthetized monkeys, we found gamma activity was strongest in layer 4B and layer 2/3. Gamma-power was generally weaker in the deep layers with a sharp drop in gamma power at the border of layer 4Cα and 4Cβ: layers 4Cβ, 5 and 6 showed weaker gamma power.
Our results suggest that 1) V1's BOLD signal will be dominated by gamma-power in layers 2/3 and 4B; 2) the BOLD signal reflects activity in the output instead of the input layers; 3) the BOLD signal is most influenced by activity in cortical layers with strong recurrent connections.
This work was supported by NIH-EY001472, NIH-EY007158, NSF-0745253, the Robert Leet and Clara Guthrie Patterson Trust Postdoctoral Fellowship, and the Swartz Foundation.