Almost two centuries ago, the science of phrenology tried to identify locations on the brain that were related to specific cognitive functions. Nowadays, we know that the basic methods and implications of phrenology were wrong. However, the basic idea that the brain is divided up in regions or local networks that perform a specific function is still prevalent in the field of neuroscience. For example, lesions in specific parts of the visual cortex can result in loss of specific subtasks like movement perception or color perception, suggesting that these subtasks are performed by separated local networks. Moreover, research using fMRI, a technique that is capable of recording localized brain activity in humans, has identified a variety of cognitive functions that are processed specifically in certain regions of the brain.
The overwhelming amount of research on localized processing might suggest that the brain is organized in small separated networks with specific tasks that communicate solely the outcome of their local processes to one-another. In such a system, specific features of for instance an apple, like color, shape, taste and the sound it makes when it falls from the tree, will be processed in different local neuronal networks. The question that arises here is how we are able reliably create a representation of an object when its features are processed independently.
One hypothesis that addresses this question suggests that the features of, for instance, this apple are not processed completely independently. Neurons processing different features of a visual scene, interact with other neurons processing features of the same object even if these neurons code for information in other modalities. For the apple it could mean that neurons coding the movement of the falling apple might enhance information processing in neurons coding for sound made by the falling apple. This way, neurons coding information from different sensory modalities are able to group their activity together in a single representation covering the whole range of features and aspects of the represented object. This process could even extend beyond simple sensory modalities and involve networks coding for representations like reinforcement value and emotional significance.
We intend to investigate this process of integration of information across modalities by studying neuronal networks while they integrate this kind of information real-time. Our aim is to not only to show the large-scale existence of this selective phenomenon in neuronal networks, but also to investigate the microstructure of information processing and the dependence of this process on specific neurotransmitters and receptor-proteins.
To study the phenomenon of sensory integration, we are converting a two-photon laser scanning microscope for use in vivo. Using the technique multi-cell bolus loading (Stosiek et al., PNAS, 2003) we aim to study activity in large groups of neurons in the upper layers of the visual and adjacent cortices. The microscope setup and some preliminary images from pilot experiments are shown on this page.
This page was last updated on 5 nov 2010