Abstract
Our internal representation of the visual world is not as detailed as we perceive it to be, as illustrated by the change blindness paradigm wherein people fail to detect large changes in a visual scene. Here we investigate at what level of visual processing this information is lost. Since it is well established that primary visual cortex (V1) is strongly modulated by contrast energy, we ask whether we can identify high contrast changes from V1 even if they are not detected by the subjects. We acquired fMRI data while participants viewed pairs of images (natural and synthetic) shown consecutively. The image pairs were either the same image or different images with a contrast change introduced in the second image. This gives us three conditions, both images at high contrast, both images at low contrast and alternating high and low contrast (change). To visualize the results we projected the fMRI activity of each recording site into visual space using their corresponding pRF (population receptive field) for each condition. We also predicted the visual space representation from the images based on the RMS contrast. We showed that the visual space representations are very similar between fMRI control conditions and RMS predictions. Next, we reconstructed the visual space representations evoked by the change condition as compared to control conditions. For the synthetic images, results from 4 subjects show that V1 visual space projections differ in the changed region for change condition as compared to the control conditions. In the natural images, we were able to identify the changes from the V1 visual space projections in some but not all images. We speculate that the nature of the change may influence our ability to detect the changes. To conclude, V1 can detect changes in a scene without reaching conscious awareness.
Meeting abstract presented at VSS 2018