NEURAL CORRELATES OF PREDICTIVE SACCADES IN YOUNG HEALTHY ADULTS
MetadataShow full item record
Our behaviour is guided by the ability to predict future events. The predictive saccade paradigm has been shown to be a valuable tool that uses eye movements to measure the control of predictive behaviour. In this task, subjects follow a visual target that alternates or “steps” between two fixed locations at either predictable or unpredictable inter-stimulus time intervals (ISIs). Response times can be measured by subtracting the time of saccade initiation from the time of target appearance. When the ISI is predictable, saccadic reaction times (SRTs) become predictive (SRT <100ms) within 3-4 target steps, but when the ISI is unpredictable, the SRTs remain reactive to target appearance (SRT >100ms). The goal of our study was to investigate neural mechanisms controlling prediction by contrasting areas in the brain that were more active for predictive (PRED) versus reactive (REACT) saccades in young healthy adults using functional magnetic resonance imaging (fMRI). fMRI analysis revealed two distinct neural networks more recruited for REACT and PRED tasks. We observed greater activation for the REACT task compared to the PRED task in oculomotor network areas including the frontal, supplementary, parietal eye fields, dorsolateral prefrontal cortex, thalamus, and putamen. These structures are all involved with the control of saccades. We also observed greater activation for the PRED task compared to the REACT task in default network areas, including the medial prefrontal cortex, posterior cingulate cortex, inferior parietal lobule, and hippocampus. These structures are known to be involved with passive thinking when subjects are not focused on their external environments. We also observed greater activation for the PRED task in the cerebellum (crus I), which may serve as the internal clock that drives the regular rhythmic behaviour observed for predictive saccades. In summary, our findings suggest brain activation in the PRED task reflects automated and motor-timed responses, while that for the REACT task reflects externally-driven responses. Therefore, the predictive saccade task is an excellent tool for measuring prediction involving fast internally-guided responses.