PROJECT SUMMARY/ABSTRACT
The objective of the proposed research is to 1) understand if activity of various cell types in DLPFC is consistent
with computational models of decision-making, and 2) investigate if there is a functional hierarchy in macaque
monkey dorsolateral prefrontal cortex (DLPFC) during decision-making. Decision-making is our ability to choose
and perform appropriate actions based on sensory cues and context to achieve behavioral goals. DLPFC is a
key decision-making brain region which receives inputs from higher-order sensory areas and sends outputs to
premotor cortices. DLPFC is also implicated in disorders of decision-making such as schizophrenia but currently,
it is unknown how circuit-level disruptions in this disorder result in its perturbed function. Specifically, we do not
understand 1) how different cell types in DLPFC mechanistically enable decision-making and 2) the spatial
organization of function in this area. We address these questions by combining in vivo neurophysiology with
computational modeling in DLPFC of monkeys performing our novel red-green reaction time checkerboard
discrimination tasks. Our first aim tests three emerging hypotheses of the role of cell types during decision-
making: 1) are inhibitory cells choice-selective or -nonselective during decision-making? 2) Do inhibitory neurons
exhibit contra-specific rather than ipsi-specific connectivity? 3) Do either somatostatin- (SST+) or parvalbumin-
positive (PV+) support decision-making through choice-selectivity? In this first aim, we will first build several cell-
type specific computational models of decision-making to establish predictions of neural dynamics during the
task. We will then use laminar multi-contact electrodes to record single neurons from DLPFC of monkeys
performing our novel red-green reaction time checkerboard discrimination task. With these recorded responses,
we will adjudicate between the models. Finally, we will examine each unit’s electrophysiological properties
(waveform shape and firing rate patterns) with our recently developed and validated approach that allows us to
delineate putative cell types (PhysMAP). With these identified populations, we will test predictions made by our
cell type-specific mean-field models of decision-making. In our second aim, we will test three competing theories
this time regarding the nature of functional gradients in DLPFC: 1) the “modular hypothesis” that different areas
are qualitatively distinct in their representations; 2) the “gradient hypothesis” that different areas contain the same
information which differs quantitatively in the “amount”; or 3) the “hierarchical hypothesis” that “upstream” areas
contain all decision variables whereas "downstream” areas only contain action-oriented variables. We will record
from DLPFC across several anatomical axes during two variants of the checkerboard task. We will then use
decoding and population analysis approaches like demixed principal component analysis to understand the task
representations of DLPFC at each location. Impact: This project will elucidate the in vivo spatio-temporal and
microcircuit dynamics in a clinically relevant brain area. Such data is a prerequisite for future development of
circuit-level therapeutics for mental illness and brain machine interfaces for recovery following brain injury.
