What Happens Before "Next" Happens: The Role of Top-Down Brain Oscillations In Visuospatial Predictive Learning
Year: 2019
Type: dissertation
Abstract: The brain constantly predicts the forthcoming sensory events and compares its prediction with incoming sensory information then it updates its internal representation based on the learning signals (i.e., the prediction error) driven by the difference between its prediction and incoming sensory inputs. If the brain uses predictive learning, it needs to differentiate predictions from the actual outcome. Hence, a fundamental question is how the brain distinguishes these two different states (predictions and actual inputs). We hypothesize that the predictive learning occurs based on a temporal organization. First, the brain generates its predictions. Later in time, the brain experiences the actual sensory stimulus. The brain then updates its internal representation as a function of the prediction error (the difference between predictions generated earlier in time and actual outcome occurred later in time). We hypothesize that, based on the biological evidence, the brain’s predictions are generated in 100 ms time-windows. We have designed sequential predictive learning experiments where subjects have implicitly learned what will happen next. Our behavioral results showed a significant learning trend over time. To investigate whether top-down modulation of brain oscillations accounts for this learning, we have recorded EEG signals. Our findings indicated that when participants covertly deployed their attention to the lateralized predicted targets, theta ITC meaningfully changed in EEG sensors over occipital cortex. Notably, these meaningful changes occurred prior to the onset of the potential target stimulus specifically during the 100ms pre-stimulus time-window and have a meaningful information of the behavioral responses of the upcoming stimulus. Furthermore, our results showed that the brain state of theta ITC 100ms prior to the presentation of the predicted trials is almost same as the brain state of 100ms post-stimulus suggesting that theta ITC carries the predicted information of an upcoming event that is reflecting top-down signals in the absence of the visual input.
Author: Zolfaghar, Maryam
Institution: University of Tehran
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7053472836