ACMD Seminar: Neural Circuit Models for Evidence Accumulation and Graded Information Transfer in Decision Making

Lindsey Brown
Postdoctoral Research Fellow, Princeton Neuroscience Institute, Princeton University

Tuesday, Oct. 15, 2024, 3:00-4:00 PM ET (1:00-2:00 PM MT)

Abstract: Often decisions are made by accumulating evidence for different possible options. Classic models of decision-making predict that neural activity will ramp persistently with evidence across the trial, consistent with many experimental recordings. However, recent data in both rodents and primates suggests that this information is not always persistently represented by the same population of neurons throughout the trial and instead shifts between populations necessitating new models. 

In the first part of this talk, I will focus on a navigation-based accumulation of evidence task in mice, where regions across the brain show sequential activity. We develop two classes of models that can accumulate evidence through sequences by gating a subset of neurons to be active at each position and introducing feedforward connections between positions to transfer this information to the next active position. Within a position, evidence is encoded either in the amplitude of two competing populations of neurons or in a “bump” of activity whose location corresponds to the evidence. Thus, these model classes make different predictions about how evidence is encoded, either monotonically or non-monotonically. We tested these predictions in neural data and found that despite similar choice selective sequences, different brain regions had different underlying evidence representations. Cortical regions showed mainly monotonic evidence tuning, while the hippocampus displayed non-monotonic evidence tuning, as would be needed for a cognitive map. 

In the second part of this talk, I will show that these models generalize to other decision-making contexts where changes in which population carries the information are not predetermined, unlike ordered sequences. A recent experimental study in monkeys has shown that graded evidence information shifts suddenly between populations of lateral intraparietal neurons when the reference frame is shifted by a saccadic eye movement. By modifying our previous model to include all-to-all connections between neurons of the same choice preference, our model captures the transfer of graded information between any pair of neurons in the population. This work suggests a broader principle for how graded information can be transferred within the population without synaptic rewiring to support decision making across different task demands.

Bio: Lindsey Brown is a postdoctoral research fellow at the Princeton Neuroscience Institute, where her work focuses on neural circuit models for decision making. She earned her PhD in applied mathematics from Harvard University, where her work applied model predictive control to shifting the phase of the circadian oscillator. She is a 2024 recipient of the Burroughs Wellcome Fund Career Award at the Scientific Interface.

Host: Anne Talkington

Note: This talk will be recorded to provide access to NIST staff and associates who could not be present to the time of the seminar. The recording will be made available in the Math channel on NISTube, which is accessible only on the NIST internal network. This recording could be released to the public through a Freedom of Information Act (FOIA) request. Do not discuss or visually present any sensitive (CUI/PII/BII) material. Ensure that no inappropriate material or any minors are contained within the background of any recording. (To facilitate this, we request that cameras of attendees are muted except when asking questions.)

*Safety Precaution: The hallway leading from the Courtyard to the exit closest to B-111 and B-113 will be used by contractors to move debris, machinery, and other supplies, as well as will be heavily trafficked by the contractors throughout the process. Be aware of the safety precautions posted during this time.

Note: Visitors from outside NIST must contact Meliza Lane at least 24 hours in advance.