Abstract
Introduction: Associative learning is an essential brain process in which different items are learned to be associated. Although some studies have found that learning of a single feature (perceptual learning) involves the early visual cortex (e.g. Shibata et al, 2011, Science), there is no clear evidence that associative learning of different visual features occurs in the region. Here, via a new fMRI decoded neurofeedback method, we tested whether associative learning of color and orientation can be created in the early visual cortex. Methods: First, based on fMRI signals in V1/V2 that responded to each of red and green grating, a classifier was constructed to optimally classify these signals into red or green. During the following training, a physically presented achromatic vertical grating was associated with V1/V2 activation patterns corresponding to red. While the achromatic grating was presented, participants were instructed to regulate brain activity to make a to-be-displayed disk as large as possible. FMRI signals in V1/V2 were inputted to the classifier, which determined the "red" likelihood. Unbeknownst to participants, the determined likelihood was then reflected in the updated disk size. After the training, chromatic psychometric functions were measured for both the trained and untrained orientations. Results: Participants learned to induce activation patterns in V1/V2 that are similar to the patterns evoked by red gratings while a vertical achromatic grating was presented. Importantly, inductions of these activation patterns paired with the vertical achromatic grating led participants to more likely perceive the trained vertical grating as red than green. This associative learning persisted at least 3 to 5 months. In contrast to the McCollough effect that shows complicated association of orientation and the color complementary to an adapted color, the present results provide the first direct evidence that the early visual cortex has the capability in forming long-lasting associative learning.
Meeting abstract presented at VSS 2016