Date created: | Last Updated:

Category: Project

Description: Intuitive explanations of the visual system of the brain are challenged by the psychological observation of illusory conjunctions. Theories and computational models exist both for regular object recognition and for illusory conjunctions. But how does the brain transition from the state of illusory conjunctions to correct object recognition? My master thesis addresses this question by investigating the time course of object perception and the transition from illusory conjunction to object recognition. To this end, I propose a neural network architecture and simulate it with spiking neurons. The main result is that bumps form first independently in different feature maps in the primary visual cortex V1 giving rise to illusory conjunctions. Within around 100 ms of visual stimulus presentation, the different feature maps communicate via a location map in MT such that they converge to the representation of the same object. With its biological plausibility, mathematical foundation, computationally efficient implementation, my spiking neuron network is a step toward an understanding of vision. It could help building a neuromorphic visual system.