While it is well understood that cortical neurons receive only a fraction of their inputs from upstream sensory areas, a majority of studies of sensory processing – both experimental and computational – necessarily focus on the role of feedforward processing in vision. Having studied feed-forward processing extensively, I would say it has been more out of a necessity than choice. Experiments can only record from a very small fraction of neurons in a given area (and much less so from other higher areas that provide feedback connections), and even if they could, such inputs are typically outside of experimental control. While a number of ideas have been put forward about the role of lateral and feedback connections, such ideas are therefore difficult to validate experimentally.
As a result, we have been developing approaches to study sensory processing in behaviorally relevant contexts, and use both behavioral observations (eye movements, task, decision) and recorded "network activity" through local field potential recordings to gain insight into the interplay of stimulus-processing and large network- and behavioral context. We hope that by doing so, we will gain insight into how recurrent inputs influence visual cortical computation, and set a novel foundation for understanding the role of network activity, including feedback from higher areas, in shaping sensory processing in the cortex.
- Cui Y, Liu L, McFarland JM, Pack CC, Butts DA (in prep) Delta-band oscillations correlate with perceptual decisions and decision-related activity of sensory neurons. [See Cosyne 2015 poster]
- Cui Y, Liu L, McFarland JM, Pack CC, Butts DA (2016) Inferring cortical variability from local field potentials. Journal of Neuroscience 36: 4121-4135. [Journal website; also see SFN 2014 poster]
- McFarland JM, Bondy AG, Saunders R, Cumming BG*, Butts DA* (2015) Saccadic modulation of stimulus processing in primary visual cortex. Nature Communications 6: 8110. [collaboration with B. Cumming lab, NIH] [Full text]