Neuronal Mechanisms of Attention
The phrase “visual attention” captures a broad range of phenomena. Most people consider visual attention to be an active cognitive process. For example, searching for a friend in a crowd involves attending to specific locations in space to locate the friend while also attending to specific visual features like the color of the clothing worn by the friend. When we study visual attention in the laboratory, we mainly study covert visual spatial attention, or the allocation of attention to a particular region of visual space that is away from the center of gaze. Covert visual attention is probably an important emergent property in animals, including humans, that care about facial expressions but are wary of making direct eye contact. We design visual attention tasks that involve shifts in the locus of covert spatial attention on different trials. Subjects are cued to attend to particular spatial locations and are rewarded for detecting subtle changes in visual stimuli at attended locations. In order to monitor attention behavior, the cue is occasionally invalid and subjects either miss the stimulus change all together, or react more slowly when detecting a stimulus change at an un-cued location. This type of cueing paradigm is called Posner cueing (Posner et al., 1980).
Covert spatial attention tasks have been utilized to study the impact of attention on neuronal activity throughout the visual system. Visual attention directed toward a stimulus overlapping the receptive fields of recorded neurons tends to increase the activity or firing rate of those neurons compared to conditions where the same stimulus is in the receptive field, but subjects are attending elsewhere. This attentional modulation of neuronal firing rate is characterized by an attention index, or the difference divided by the sum of neuronal activity across attention conditions (e.g. attending toward versus attending away from the stimulus in the receptive field). Interestingly, attentional modulation of neuronal firing rate scales along the visual hierarchy. Accordingly, attentional modulation of neuronal firing rate is modest in early visual structures such as the visual thalamus (LGN) and primary visual cortex (V1) and is more robust in higher visual cortical areas such as MT, V4 and FEF. However, in all of these visual structures, there is large variability in attentional modulation across individual neurons. This variability suggests that there may be rules governing whether and how much attention alters the activity of distinct sub-populations of neurons. Furthermore, attentional modulation of firing rate may not be the best readout of attention as the attention index is measured over long time scales (1-2 seconds) while attention modulation of actual neuronal activity may be highly dynamic. We previously discovered that attention can alter the efficacy of communication between neurons in the early visual pathways (Briggs et al., 2013) and we are interested in pursuing further explorations into the underlying biophysical, cellular, and circuit mechanisms by which attention alters neuronal activity.
Our most recent work illustrates the diversity of attention effects that can be observed in a single visual cortical area (Short et al., 2017). Specifically, we discovered that attentional modulation of V1 neurons is not uniform within the attended spatial region. Instead, attention selectively increases the firing rate of neurons that encode stimulus features that are important for successful completion of the task. These findings suggest that a general attention spotlight or spatial gain model of attention is too limited. Instead, models of attention must incorporate both spatial and feature attention components in order to accurately capture the variability in attentional modulation of neuronal activity, even for neurons responsive to the same attended spatial location. Future work will test this hypothesis explicitly by recording from the same neuronal populations while subjects perform multiple attention tasks involving detection of different visual stimulus features.
back to projects