Neuronal representations in the primary visual cortex (V1) are shaped by experience, through the integration of both external visual inputs and internal signals related to an animal’s behaviour. The activity of V1 neurons is thus influenced by a variety of factors when animals navigate through their environment. Using in vivo two-photon calcium imaging, my research group examined the activity of V1 neurons before, during and after head-fixed mice were trained using a visually-guided task in a virtual reality environment. We found a refinement of responses to both visual aspects of the task as well as reward-related responses in V1. These results provide additional evidence that neuronal activity in primary sensory areas is highly dynamic and changes with the behavioural significance of sensory inputs. I will present current projects of the lab related to the mechanisms underlying this plasticity.
When neuroscientists contemplate the coding of a tactile stimulus, it is natural for us to focus first on how the evoked neuronal activity underlies the perception of stimulus features. But a second percept accompanies the tactile experience – the feeling of time occupied by that stimulus. Logically, the “raw material” for both percepts must be the neuronal representation of the stimulus itself, and from this representation further streams of processing must lead to distinct percepts. We carried out psychophysical experiments in which human subjects and rats judged, on each trial, either the intensity of a vibration or the duration of a vibration. All subjects (both species) showed an interaction between the two percepts: a longer vibration feels stronger in intensity, and stronger feels longer. Further exploration of the interaction between the intensity and duration percepts allowed us to construct a computational framework whereby the vibration-evoked firing early in the processing stream is accumulated by two integrators, in parallel, each integrator giving rise to a distinct percept. Neuronal data from behaving rats seem to support this framework.