Neuronal oscillations, attention, and normalization circuits

Attention aids our ability to perceive our external world in detail and to learn which aspects are relevant for successful behavior. Studies over the last 20-30 years have revealed how attention alters the processing of information at the neuronal level, but the studies have often generated conflicting results. Attention increases the neuronal activity of neurons that represent the attended object. Attention can also change the level of rhythmic activity between different neuronal circuits and thereby improve the communication between channels. This Rhythmic activity has been proposed to play an important role in cognitive functions such as attention, perception, working memory, and other cognitive functions. Disruption of this rhythmicity is associated with cognitive deficits as seen in schizophrenia, autism, Alzheimer disease and Parkinson's disease. Despite these insights, we nevertheless have a poor understanding how increases of neuronal activity or alterations of rhythmic activity are mediated or controlled. We still know little about the neuronal circuits involved. We equally have very limited insight why differences are found in different brain areas or why subtly different tasks can yield very different effects on these measures of brain activity. Recent theoretical models propose a mechanism to reconcile these conflicts. They propose that attention acts on inhibitory (normalization) circuits in the brain and thus controls activity levels and rhythmicity levels. These models, despite their explanatory success in silico, need experimental validation. By doing so, we will gain a more detailed understanding of different circuits involved in cognition, and also obtain a more mechanistic understanding in the genesis of cognitive dysfunctions.

Attention aids perceptual abilities. The neuronal underpinnings have been studied in detail, but often yielded conflicting results. Normalization models of attention reconcile many of these conflicts, with notable success in explaining the interactions between attention and stimulus contrast. Normalization models argue that attention affects neuronal processing by interacting with the stimulus and the suppressive drive: attention itself drives normalization. Neuronal responses are thus determined by the stimulus size and by the size of the attentional field. Normalization should thus affect different parameters of neuronal tuning functions in a manner similar to attention. The proposal cannot strictly be true, as normalization usually reduces neuronal activity, while attention increases it. However, across a neuronal population the effects of stimulus and attention driven normalization could still be correlated, even if the average sign of the effect differs between the two. Normalization models of attention equally predict that attention should increase oscillatory activity, as normalization circuits are a main driver of gamma frequency oscillations. Appropriate implementation of the normalization model of attention explains why attention has different effects on gamma oscillatory power in V1 than it does in V4, provided that stimulus size, size of the attentional focus, and receptive field size are accounted for. We aim to validate (or falsify) these predictions, by performing simultaneous recordings across cortical layers of V1 and V4, while stimulus size and the size of the attentional focus are parametrically varied, thereby assessing their effects on normalization circuits. We will gain a detailed understanding of the circuits involved in attention. We aim to understand why subtle task manipulations yield neuronal data sets, which so far make it impossible to propose a unified theory of the neuronal mechanisms underlying attention and cognition.

The research will mainly have an impact on the academic neuroscience/cognitive neuroscience community (see academic beneficiaries). It is unlikely to have an immediate or mid-term impact on treatment of patient groups suffering from cognitive dysfunction, as the research falls under the category of basic neuroscience. However, it may still inform clinicians, as it investigates mechanisms which are proposed to be the variables most profoundly and measurably affected in cognitive dysfunction at the neuronal level. Understanding how these variables can be influenced by stimulus and task conditions is important for diagnosis. The research may likewise have an effect on clinical practice, as it gives insight which stimulus and task conditions might best be implemented in clinical test conditions.
Beyond this, the research may have an impact on current attempts to build the next generations of robots, so called sentient machines. Building sentient machines (e.g. human companions: FET flagship proposal Robocom [http://www.robotcompanions.eu/], I am a member of the consortium) requires an in depth understanding of the principles that allow us to navigate our everyday life successfully (attention, learning, memory). To do so, it aim to harness the principles underlying neuronal communication in living systems and implement the flow of communication signals in artificial systems. An understanding at the level of detail aimed for in this proposal will aid this endeavor.