Selective Attention

Cortical neurons in sensory areas show patterns of irregular firing that are, however, synchronized to, e.g., gamma oscillations. This regime of noisy oscillations is further reinforced by selective attention, which arises questions such as: why do regular and irregular components appear simultaneously? is this regime of noisy oscillations functionally relevant? are oscillations reducing noise?

Aiming to address these questions, I developed a neural circuit model of attentional processing (Ardid et al., J Neurosci 2007, 2010). According to this model:

• Noise from neural variability underlies multiplicative scaling of firing rate. Hence, a modest increase in input, such as that of top-down attentional modulation, is supra-linearly amplified. At the neuron level, this means that the higher the reference firing was in absence of attentional modulation, the stronger the output firing becomes for the target neuron with attention.
• At the population level, visual neurons, such as those in MT and V4, show divisive normalization. In the model this is achieved by global inhibition. One local excitation and global inhibition are combined in the model, the prediction is that the net effect from top-down attention is a signal-to-noise (S/N) enhancement. More precisely, the firing rate of each neuron is either multiplicatively or divisively scaled by a factor that varies according to the distance with respect to the attentional focus.
• Remarkably, the model shows that S/N enhancement is further reinforced in the regime of noisy oscillations. Hence, the model predicts that noisy oscillations represents the best of the two worlds: (1) noise underlies multiplicative scaling (the basis for attentional S/N enhancement), and (2) oscillations, instead of reverting noise, build upon it and reinforce the S/N enhancement.