Principal Investigator: Carlo Umilta
A Neurophysiologically-Motivated Computational Model of Spatial Attention in Vision and Action
Traditionally, selective attention is considered as a supramodal system acting as a controller increasing the efficiency of all the basic sensorimotor systems. For instance, a well-known metaphor of visual spatial attention is that of a searchlight, which, by "illuminating" specific regions of spatial maps, renders more efficient processing of the stimuli failing on those regions. Even recent connectionist models, which describe attention as a differential activation of feature-specific representations or pathways in the system, assume ".. some kind of "Device" which can locate and activate relevant representations in the relevant module." (Heslenfeld et al., 1997).
This view is challenged by a theory known as the "premotor" theory of attention, first proposed by two of the applicants. The premotor theory maintains that attention derives from activation of the same circuits that determine perception and motor activity, that is, those involved in sensorimotor trans-formations. For instance, preparation of motor programs related to eye movements plays a central role in orienting spatial attention. The premotor theory has gained evidence from a number of recent neurophysiological, psychological, and neuroimaging studies.
The aim of the present project is to develop a neurobiologically plausible computational model of the neural circuits involved in space coding and spatial attention. The model will incorporate strong design principles derived from neuroanatomical, neurophysiological (e.g., single neuron recordings in the monkey) and brain imaging studies. This will allow to formally test some basic assumptions of the premotor theory of attention and will provide detailed simulations of a number of behavioral studies, including reaction time experiments in humans.
The model will serve as the basis for driving further research in both
the neurophysiological mechanisms and the behavioral correlates of attention.
The premotor theory of attention was originally formulated to account
for orienting attention to a spatial location. One important aim of the
present project is to formally investigate whether the same principles
can account for attention to graspable objects, that is, whether preparation
to interact with tridimensional objects determines a facilitation of the
detection and discrimination of their perceptual characteristics related
to the prepared action. In this regard, the results of the project of
Prof. Rizzolatti and his team (to which the present project is bundled)
should provide further constraints to the model. They will use a combination
of methodological paradigms, that is, reaction time (RT) experiments,
recordings of event related potentials using combined EEG-MEG technique,
and transcranial magnetic stimulation. The modeling efforts will contribute
to improve the design of the behavioral investigation and the model's
predictions will drive new experiments.
