Prefrontal cortex dopamine

Dopamine in the prefrontal cortex (PFC) is essential for normal cognition and manipulations of it have profound effects on working memory and cognitive flexibility. It may also play an important role in psychiatric disease, especially schizophrenia, where hypodopaminergia in the PFC has been proposed as a cause of the difficult to treat negative symptoms. Despite its importance, the precise stimuli that elicit PFC dopamine release and its fundamental role in modulating PFC activity remain uncertain.

PFC dopamine is known to be strongly released following aversive stimuli, while simple rewarding stimuli, like a water droplet given to a thirsty mouse, release relatively little. Our lab is working to extend our knowledge of this system by recording PFC dopamine release in a broader range of situations, from simple rewarding and aversive stimuli to more complex stimuli, like social interactions and mixed-valence stimuli, that have a mixture of good and bad aspects. Our findings suggest that PFC dopamine responds to a broader array of stimuli than previously appreciated and its properties may fit within a modified version of the reward prediction error framework.

Dorsolateral striatum dopamine

Dopamine release in the dorsolateral striatum (DLS) has been linked with reward prediction errors, movement vigor, action selection entropy and movement initiation, but it remains controversial whether any one of these theories can explain all of the existing data. Our lab has been recording DLS dopamine during continuous running, either voluntarily initiated or driven by a treadmill, using a novel task that eliminates the anxiety responses mice show on a driven treadmill. Our findings suggest a novel view of the DLS dopamine system that is consistent with an RPE framework, but with a very narrow temporal window.

Transformers and the brain

Machine learning has been revolutionized by the discovery of transformers in 2017. These new neural networks use a mechanism called self-attention to allow for a more dynamic routing of information through a neural network and excel at sequence processing tasks like language comprehension and generation. The role of self-attention (in the machine learning sense) in the brain, however, remains uncertain. Recently, we proposed a new theory of apical dendrite computation that allows pyramidal neurons to perform the essential computations of self-attention in parallel by comparing spike trains in spines. We are currently investigating if this computation can be implemented more quickly and testing the hypothesis experimentally.

Mouse virtual reality for vision and whisking

Virtual reality setups have become a common feature of neuroscience laboratories, but the popular dome and LCD monitor-based systems have a number of weaknesses in their ability to completely control the visual stimuli delivered to the experimental subject, as well as being costly and difficult to incorporate with many experimental designs, due to their large size. We have been collaborating with the Schaffer/Nishimura lab to test miniaturized VR headsets designed for mice. In addition to being cheap and compact, we have found that they are far superior to our dome-based systems in their ability to evoke innate responses to looming stimuli, suggesting that they are more immersive.

Our lab has also been building a whisker based VR system that can represent complex 2D mazes for navigation by using a rotating whisker stimulator that can be rapidly repositioned to represent the closest wall to the experimental subject. We are currently testing our prototype to see if it can improve navigation in 2D virtual environments by adding whisker sensation.