Abstract
Studies show that a successful visually guided behaviour depends on the target's spatiotopic position. Targets can be encoded and retrieved from memory (decoded) to perform different tasks. However, whether "this retinotopically encoded information transforms to spatiotemporal encoded representations before residing in memory" is not investigated thoroughly. This study aims to explore how accurate the retinotopic representation derives from spatiotopic encoding. We designed a task in which the spatiotopic and retinotopic encoding targets were evaluated on the same trials. The extra-cellular activity was recorded by inserting two 10x10 multi-electrode arrays in the dorsal and ventral circumvolutions of the lateral prefrontal cortex (area 8a, and 9/46, respectively) in two rhesus macaque monkeys. For example, an animal can remember the spatiotopic location of a stimulus in a particular position on display that leads to a success-based reward. However, does the animal keep the target's location as a retinotopically-encoded representation or a spatiotopic one? A classification-based decoding method was employed on single-cell recordings to show spatiotopic and retinotopic representations of a target. Albeit the results' accuracy in decoding retinotopic representations were higher than spatiotopic ones, the lower accuracy in spatiotopic location retrieval can be attributed to the specific brain areas where the cell recordings were performed. Additionally, our results indicate that spatiotopically-positioned targets are encoded as well as the retinotopic encoding. Different classification techniques were used (decision tree vs linear discriminant analysis), and the results show that each method responds better to a specific encoding in both spatiotopic and retinotopic frames of reference, respectively.