A journey through neural space: Neuronal population correlates of visual processing and perception

Single neurons can often show strong correlations with an animal’s behavioral output, but it is currently unclear what role population-level interactions play in visual processing and perception.
To address these questions, I performed several two-photon calcium imaging experiments in layers 2/3 of mouse primary visual cortex (V1). In Chapter two, I present evidence that signals related to the sensory input may be anatomically differentiated from fluctuations in population activity, and highlight the importance of correlations for the information content of neural codes. Chapters three and four describe the results from an experiment in which mice were trained to perform a visual detection task, while the concurrent activity of ~100 neurons was recorded. Two main conclusions could be drawn from these data: 1) rather than the hitherto assumed importance of mean neuronal activity, it appeared more pivotal for the behavioral detection of visual stimuli that neurons in V1 responded heterogeneously; and 2) specific sequential patterns in neuronal activity were consistently correlated with stimulus detection. This suggests that temporal properties of neuronal activity may play a larger role in sensory processing than is often thought. Finally, chapter five describes how multidimensional population codes may present a more parsimonious explanation for many experimentally observed phenomena in neuronal activation patterns. From a multidimensional perspective, neural codes are more temporally stable, more information efficient, and more robust to trial-by-trial fluctuations in activity. Conceptually integrating these results in chapter 6, I conclude that some seemingly disparate experimental observations may be reconciled by adopting a population-level perspective.