Sensory neurons exhibit unpredictable spiking responses to fixed sensory input. Some of this response variability is correlated between neuronal pairs. This correlated variability (spike-count correlation; rsc) is thought to impair judgments about the stimulus, because these rely on statistical decisions based on pooling many sensory neurons. However, a portion of the correlated variability may in fact reflect variability in modulatory/feedback inputs. How such common inputs impact the neuronal representation of sensory events is poorly understood. An important experimental constraint can be provided by quantifying the portion of rsc that derives from feedback. We confine our analysis to the structured rsc thought to be important for perceptual judgments. In this dissertation, I present results from population recordings in macaque primary visual cortex (V1) while subjects perform a coarse orientation discrimination task using filtered noise stimuli. Given current models, the presence of choice-related activity (Choice Probability; CP) implies that rsc structure take a particular form. I empirically demonstrate that the rsc structure in area V1 (along the orientation dimension) is compatible with these predictions. However, I show quantitatively that this is mostly due to common inputs which change dynamically with the subject’s task (i.e. the set of orientations being discriminated), strongly implying a feedback origin. I also show quantitatively that CP is mostly due to the portion of rsc structure that changes dynamically with the task. This is compatible with the view that signals related to a subject’s perceptual choice are fed back to V1 neurons supporting that choice, and that CP is mostly due to this feedback. In addition, I show that the structured spike-count correlations that change dynamically with the subject’s task degrade the perceptual performance of an ideal observer of the V1 population. This is because they introduce fluctuations in the V1 representation that mimic the effect of changing the stimulus along the dimension being discriminated. I discuss what these results imply about the nature of the common feedback inputs that appear to underlie rsc structure, and discuss possible feedforward effects the resulting spikes may have on a perceptual decision.
Bondy, Adrian G.,
"Decision-Related Feedback Influences the Structure of Correlated Variability in Visual Cortex"
(2015).
Neuroscience Theses and Dissertations.
Brown Digital Repository. Brown University Library.
https://doi.org/10.7301/Z0D50KCG
