Abstract
This research proposes a methodological approach allowing to study the representation and encoding of the different levels of visual processing by generating metameric visual stimuli – i.e., stimuli recruiting different populations of neurons at certain levels of processing but not others. To that end, we used fMRI datasets from natural pictures and developed an encoding model predicting fMRI activation for an image, based on the activation for the same image in a robust Resnet-50 pre-trained on millions of images. Using the encoding model, we then predicted the fMRI activations associated with an image X and find the image X’ (representing a metamer of X) using an Adam optimizer. The loss function minimized the distance between X and X' in some parts of the visual cortex (IT to V2) but maximized the distance in other parts of visual processing (V1). So, for the image X, we changed the loss function and calculated the images X', X'' and X''' which represent gradual metamers of image X for each part of the visual system, where X' is different from X in V1, X'' 'is different from X in V1 and V2, X''' in V1, V2 and V4. This approach allows a better understanding of the role of each level of perceptual processing by constructing a mapping of activated brain areas in a more interpretable space - that of stimuli - and make possible the development of more precise experimental protocols in visual neuroscience.