PRF MODELS

Computational neuroimaging: quantitative models of human visual neurons

 Partecipanti

Mappa

 Word cloud

Esplora la "nuvola delle parole (Word Cloud) per avere un'idea di massima del progetto.

 Obiettivo del progetto (Objective)

'One of the most complex systems, often referred to as science’s last frontier, is the human brain. However, human neuroscience research is constrained; it has no or rarely access to invasive procedures that are widely used in animals. Invasive procedures allow measurements at much smaller scales, e.g. at level of individual neurons. Consequently, most knowledge of human neurons is extrapolated from animal experiments. Ultimately, at least some human and animal neuronal properties will differ, making human measurements at comparable scales essential. We propose to bridge this gap by coupling non-invasive human neuroimaging signals, measured at the millimetre scale, with neuronal properties, measured at the micron scale. We will use a new computational neuroimaging method, which measures human neuronal population properties that are close to those derived from invasive animal experiments (Dumoulin and Wandell, 2008). The utility of these methods extends from basic to applied neuroscience, as supported by initial observations in both rare, i.e. achiasma, and common disorders, i.e. macular degeneration a leading cause of visual impairment. Thus, this method has both fundamental and clinical applications. We will extend this method in three ways. First, we will extend this method from human population to single neuron estimates. We require that our estimates replicate well-established animal experiments and human behavioural data. Second, we will validate the measurements with predictions derived from an established theoretical framework. The ability to confirm basic observations is crucial for correct interpretations of potential differences. The current theoretical framework was established using artificial stimuli that are supposed to extrapolate to natural conditions. However, recent studies suggest that this extrapolation capability is limited. Third, we will extend this method to build more complex models of neuronal properties under natural viewing conditions.'

Introduzione (Teaser)

Most of our knowledge of the human brain is extrapolated from animal experiments. However, human and animal neuronal properties differ, making human measurements at comparable scales essential.

Descrizione progetto (Article)

The EU-funded project ?Computational neuroimaging: quantitative models of human visual neurons? (PRF MODELS) proposed to bridge this gap by coupling non-invasive human neuroimaging signals with neuronal properties.

The researchers used a new computational neuroimaging method which measures human neuronal population properties that are close to those derived from invasive animal experiments. The utility of this method extends from fundamental to clinical neuroscience, as is supported by observations in both rare and common disorders that cause visual impairment.

The PRF MODELS team first wanted their computation measurements to replicate well-established animal experiments and human behavioural data. They then validated their measurements with predictions derived from an established theoretical framework. Finally, they aimed to build more complex models of neuronal properties under natural viewing conditions.

Neuropsychological evaluation of two stroke patients revealed specific deficits in motion perception. These results provided evidence for several distinct mechanisms to process motion in the human visual system. The project also provided evidence for both stability and plasticity in a rare human congenital disorder known as achiasma.

The PRF MODELS method can be applied to basic and applied neuroscience, as indicated by initial studies of achiasma and more common disorders like macular degeneration.