How does the cerebello-thalamocortical pathway support flexible, cognitive behavior?
Our movements are smooth, coordinated, and adaptable to new environments. Decades of research on the cerebellum suggests that it is critical for these aspects of motor control. The cerebellum forms an internal model of the body to predict the sensory consequences of actions. These predictions enable us to quickly correct errors in our movements and adapt to new environments when these predictions fail. My postdoctoral work tests the hypothesis that the cerebellum plays a similar role in cognitive function by using internal models to correct and adapt internal thought processes. It leverages our knowledge of the neural mechanisms of sensorimotor timing, a novel task design, large-scale electrophysiology, perturbations of neural activity, and computational modeling to understand how the cerebellum interacts with neocortex to support cognition. I have developed a timing adaptation task for monkeys in which they must build an internal model to predict events in the environment and use these predictions to error-correct and adapt their internal time-keeping. As monkeys perform this task, I record activity in the dentate nucleus—a primary cerebellar output—while simultaneously recording in its downstream targets in the thalamus and sensorimotor neocortex. Ultimately, my goal is to reveal fundamental principles about how the cerebellum interacts with the neocortex in service of smooth, coordinated, and adaptable cognitive processes.
How does the brain commit to a decision?
The brain makes decisions by accumulating evidence until there is enough to stop and choose. Neural mechanisms of evidence accumulation are established in association cortex, but the site and mechanism of termination are unknown. In my PhD work, I showed that the superior colliculus (SC) plays a causal role in terminating decisions and provided evidence for a mechanism by which this occurs. We recorded simultaneously from neurons in the lateral intraparietal area (LIP) and SC while monkeys made perceptual decisions. Despite similar trial-averaged activity, we found distinct single-trial dynamics in the two areas: LIP displayed drift-diffusion dynamics and SC displayed bursting dynamics. We hypothesized that the bursts manifest a threshold mechanism applied to signals represented in LIP to terminate the decision. Consistent with this hypothesis, SC inactivation produced behavioral effects diagnostic of an impaired threshold sensor and prolonged the buildup of activity in LIP. The results reveal the transformation from deliberation to commitment.
Paper: https://www.cell.com/neuron/fulltext/S0896-6273(23)00400-2
How can we use behavioral data to identify a decision-makers strategy?
Many tasks used to study decision-making encourage subjects to integrate evidence over time. Such tasks are useful to understand how the brain operates on multiple samples of information over prolonged timescales, but only if subjects actually integrate evidence to form their decisions. This project explored the behavioral observations that corroborate evidence-integration in a number of task-designs. I found that several commonly accepted signs of integration were also predicted by non-integration strategies. Furthermore, an integration model could fit data generated by non-integration models. I then identified the features of non-integration models that allowed them to mimic integration and used these insights to design a motion discrimination task that disentangled the models.
Paper: https://elifesciences.org/articles/55365.
What is the nature of single-trial dynamics that underlie a decision process?
Decisions are well-characterized by models of noisy evidence accumulation. In the brain, neurons in area LIP display signatures of this evidence accumulation process in trial-averaged data. Evidence accumulation is a stochastic process. Thus, averaging across trials obfuscates the latent dynamics that hypothetically give rise to the decision, though this has been necessary historically because the number of simultaneously recorded neurons was limited by recording technologies. In close collaboration with postdoc Natalie Steinemann, I used newly-developed macaque neuropixels probes to record simultaneously from hundreds of neurons in LIP as animals made perceptual decisions. These large-scale recordings allowed us to observe the dynamics of LIP population activity on single-trials. These dynamics closely matched those predicted by models of noisy evidence accumulation (e.g. drift-diffusion models) and could be used to accurately predict the animals’ choices and reaction times on single trials.
Paper: https://elifesciences.org/articles/90859
Development of large-scale recording technology for use in non-human primates.
High-density, integrated silicon probes have transformed systems neuroscience in small animal models. These probes enable large-scale neural population recordings with single cell resolution. However, existing technologies have provided limited functionality in nonhuman primate species such as macaques, which offer close models of human cognition and behavior. Through a large collaboration between IMEC and the Howard Hughes Medical Institute, I was involved in designing and testing Neuropixels 1.0-NHP, a new generation of probes for use in non-human primates. The Neuropixels 1.0-NHP is a high channel-count linear electrode array designed to enable large-scale, simultaneous recording in superficial and deep structures within the macaque brain. Specifically, I developed a protocol for reliably and safely inserting these delicate probes into the macaque brain, assessed their performance and longevity, and designed custom hardware that interfaces the probes with commonly used, commercially available microdrives. In our paper, my colleagues and I demonstrated recordings from thousands of neurons within a single session and large-scale recordings from dozens of cortical and subcortical regions. We also demonstrated examples of the new classes of experiments that can be achieved with these probes. All of the protocols and hardware designs have been released open-source to the primate neuroscience community in order to facilitate fast, wide-spread adoption of this transformative recording technology.
Paper: https://www.nature.com/articles/s41593-025-01976-5
gabrielstine.com 