Overview

There are two main findings from (Liu, Hashemi-Nezhad, and Lyon 2015) in the the primary visual cortex (V1) using anesthetized cat. First, that contrast invariance orientation tuning depends on having a stimulus that extends beyond the CRF. If the stimulus is optimized for the CRF, then the tuning width decreases with lower contrast (illustrated in Figure 3 of the paper). The orientation tuning profile is invariant when the stimulus extends to the surround, but when is only covers the CRF.

The second main finding (illustrated in Figure 4 of the paper) is that contrast invariance appears with the large stimulus because the tuning width decreases in the high contrast stimulus when the surround stimulus is added to the CRF stimulus. The tuning width for the low contrast conditions on average stays the same with or without the stimulus in the surround (although individual cells may be facilitated or suppressed).

This results of (Liu, Hashemi-Nezhad, and Lyon 2015) are difficult to reconcile with classical results and, for me, indicate that a better measure of contrast-invariant orientation tuning is needed. This paper should definitely be read for anyone interested in this feature.

Stimulus and Methods

For the main experiment, they have two contrast conditions (low and high) that are defined for each neuron and two size conditions (CRF and CRF+ECS) that are defined for each contrast (and neuron). The smaller of the two sizes, the CRF / patch condition, is defined as the size that produces the largest response from the cell. The larger size, the CRF+ECS (extraclassical surround) condition, is defined by the size that produces the maximum suppression.

The paper almost exclusively reports the half-width at half height (HWHH). This is half the width of the (fitted) orientation tuning curve that elicits half of the maximum response of that tuning curve.

Discussion

The paper states in the discussion that most other papers on this topic did not use the optimally sized stimulus, hence why they report different results. They do point out that (finn_contrast-invariant_2007?) did use a similar CRF condition, but reported different results presumably because they used patch clamping. In Supplemental Fig. 3 of Finn et al., there are some extracellularly recorded neurons that reportedly are more consistent (I haven’t checked results).

Deep anesthesia is known to change the properties of ECS of early visual neurons. It is unclear to me how much the results from anesthesized animals can be generalized to the normal awake state.