The experiments reported in this thesis lend strong computational support to the theory that tilt aftereffects result from Hebbian learning of the strengths of lateral connections between neurons. Furthermore, the aftereffects occur as a result of the same decorrelating process that is responsible for the initial development of the orientation map. This process tends to deemphasize constant features of the input, resulting in short-term perceptual anomalies such as aftereffects. The same model should also apply to other aftereffects and to simultaneous tilt illusions.
Because RF-LISSOM is a computational model, it can demonstrate many phenomena in high detail that are difficult to measure experimentally, thus presenting a view of the cortex that is otherwise not available. For instance, this thesis showed direct visualizations of aftereffects as they were occurring in the simulated cortex, making it clear exactly which processes contributed to the effect. This type of analysis can provide an essential complement to experimental work with humans and animals.
RF-LISSOM is the first model to provide a comprehensive and fundamental account of how both cortical structure and function emerge by Hebbian self-organization in the primary visual cortex. It is also the first to show how both indirect and direct tilt aftereffects could arise from simple, biologically plausible mechanisms in the primary visual cortex. Thus a single simple model may explain an unprecedented number and scope of cortical phenomena, which contributes substantially to our understanding of the cortex.