How can you make decisions or perform calculations with dynamic vectors ?

Neural vectors have no reason to be static - their direction and amplitude can change over time. In this lecture, we will illustrate the concept of neural trajectory, i.e. the idea that, over time, the neural vector that codes for a representation can transform, shift or rotate, and that this dynamic trajectory reflects the progressive transformations of a mental object as it enters a computation or decision. The temporal decomposition of the evolution of neural activity during a simple decision reading enables us to distinguish between stages of perception and decision-making within the same neurons of the prefrontal cortex. Similarly, neuronal recordings in the human cortex enable us to follow, in real time, the evolution of error detection or sentence processing. This very recent research provides new visualizations of cortical state over time, and new models of logical computation with vectors.