Thalamocortical transformation of visual signals: role of geniculate orientation biases and intracortical circuits in orientation selectivity of striate cortical neurons

© 2013 Dr. Sivaram ViswanathanOne of the fundamental questions in our understanding of the visual system is posed by various neurons along the afferent visual pathway, from eye to the brain, which differ in their response properties. While the neurons in the retina and the lateral geniculate nucleus (LGN) of cats and most mammals respond reasonably to diffuse patches of light, most primary visual cortical neurons (striate cortex) respond only to elongated stimuli of particular orientation. Hubel and Wiesel’s Nobel prize winning discovery of this orientation selective property of striate cortical neurons has opened up an enormous field for investigation. They proposed that striate cortical orientation selectivity arises from excitatory convergence of several non-oriented LGN afferents whose receptive fields are arranged along the long axis of the striate cell’s receptive field. The emergence of orientation-tuned responses among these striate cortical neurons from relatively poorly tuned thalamic neurons has been a topic that courted an intense debate. This comes from the findings of alternative models of orientation selectivity that have shown that the excitatory input from the LGN is almost circular, that the major excitatory drive comes from within the cortex and intracortical inhibition may have a role in the generation of orientation selectivity. These models claimed that inhibitory circuits within the visual cortex are required to sharpen the broadly tuned input from the LGN. One such model, called anisotropic LGN-driven recurrent model (ALD-R model), suggests an alternative scheme that exploits the orientation biases present in the responses of subcortical neurons, with the intracortical inhibitory and excitatory networks sharpening this broadly tuned subcortical inputs at the cortical level. Several experiments were designed in this study to test the predictions of the ALD-R model, including the role played by geniculate orientation biases and non-specific inhibition in the orientation tuning of cat striate cortical neurons. Firstly, the contribution of broadly tuned inhibition on geniculate orientation biases was tested using a novel electrical stimulation protocol within the LGN. The effect of this LGN electrical stimulation was also studied in the topographically corresponding striate cortical region. Our results showed that the non-specific inhibition brought about by electrical stimulation within the LGN resulted in significant sharpening of the LGN orientation biases as well as a mild broadening of the orientation selectivity of neurons in the corresponding striate cortical region. The effect of non-specific inhibition generated by this electrical stimulation paradigm was also compared with pharmacologically induced inhibition on geniculate relay cells and the results were comparable. These findings add support to the ALD-R model, where even non-specific inhibition acting on geniculate orientation biases could lead to sharper orientation tuning similar to those seen among striate cortical neurons. The dependence of striate cortical orientation preferences on LGN orientation biases was tested in the second experiment, where simultaneous paired extracellular recordings were performed from a geniculate cell and a cortical cell, which had overlapping receptive fields. The rationale behind this study is that a geniculocortical pair with matched orientation preferences would show higher coherence than a pair that has near orthogonal orientation preferences. The degree of coherence between the spikes from each neuronal pair was compared with the differences in optimal orientations exhibited by each geniculocortical pair. The results support the prediction of ALD-R model that orientation preferences of LGN neurons can predict the orientation preferences of striate neurons that they purportedly project to. The next experiment tested whether contrast invariance of orientation tuning, a property that is often pointed out as a drawback of Hubel and Wiesel’s model, is a subcortical property that gets transferred to striate cortical neurons. Extracellular responses of LGN neurons for varying stimulus contrasts were studied and their orientation sensitivities for the high and low contrast stimuli were compared. The results show that cat LGN neurons exhibit contrast invariance of orientation tuning and striate neurons could acquire this property from the geniculate excitatory inputs. Together, the results of the above studies substantiate the claims of ALD-R model where broad orientation selectivities present in geniculate inputs predict striate cortical orientation tuning and also other properties such as contrast invariance. The contribution of intracortical inhibition, especially those that are broadly tuned or non-specific is pivotal for the generation of orientation selectivity of striate cortical neurons