Expanding the field of view: from classical to extra-classical receptive fields, neurons to populations, and voxels to sensors

Vision is an important sense for humans and its understanding has a profound impact on our lives. Every neuron in the visual cortex processes the information from a small region within the visual space, known as the receptive field (RF) of that neuron. Neurons also interact with one another through intercellular connections. Sometimes, stimuli falling outside the classical RF of the neuron can influence the response of these neurons through such intercellular connections. Such region in the visual field in which the presence of a stimuli by itself cannot elicit a response from the neuron but can modify its response to stimuli falling inside the classical RF is called its extra-classical RF. Extra-classical interactions are usually observed using contextual stimuli such as natural images. In natural images, extra-classical interactions play a crucial role in object recognition and scene segmentation. In humans, RF properties of a single neuron are not generally studied since invasive electrophysiology techniques are not feasible to be performed on healthy individuals. Hence, non-invasive technique called functional Magnetic Resonance Imaging is employed. However, spatial resolution of fMRI recordings is in the range of ~1 mm and thus contain the activity of millions of neurons. Hence the RF properties studied in humans are called as population RF (pRF). Also, fMRI measures the neural activity as a change in the blood oxygenation resulting from such activity. So, fMRI measurements are called as Blood oxygenation level dependent (BOLD) signals. There are other modalities for the measurement of neuronal activity such magnetoencephalography (MEG), electroencephalography (EEG), eCOG (electrocorticography) which also indirectly measure different properties of the activity of a population of neurons. The first section of the project introduces a novel forward model that can predict the MEG recordings to a visual stimulation using a pRF model built from fMRI. This gives confidence in a common underlying neural computation in both measurements and confirms that pRF models built using fMRI reflect neural computations. pRF properties are estimated using fMRI recordings by building models that best describe the computations performed by the population of neurons in response to visual stimuli. These models are called as the pRF models. PRF models provide detailed information about the pRF properties such as position, RF center size, RF suppressive surround size, spatial summation and connectivity. with the help of carefully designed experimental stimuli and tasks, we can use pRF models to study different properties of visual processing in both healthy and diseased population. In the second section of the project, we investigate the extent to which the pRF properties reflect the RFs of the individual neurons and their interactions. We divide this into two parts. In the first part, we found the evidence that the effects in single neuron level is disappearing at the population level due to various intercellular interactions. For this, we used stimuli with different spatial frequencies that can excite sub population of neurons selective to a specific spatial frequency. In the second part, we used contextual stimuli (natural images) to investigate the effect of the extra classical RF interactions on the pRF properties. Here, we were able to identify extra-classical RF interactions at the population level. Finally, in the third section, we use pRF models to investigate the abnormal visual processing in patients with schizophrenia that leads to visual hallucinations. We showed evidence for a decrease in the RF surround size in patients with schizophrenia with the visual hallucinations compared to those without visual hallucinations. We believe that the decreased surround size causing an imbalance in the inhibitory mechanism could be a reason for visual hallucinations in patients with schizophrenia