Mathematics and Statistics Colloquium (Mar 26)
Join the Department of Mathematics and Statistics for a colloquium on Thursday, March 26 from 4:00 pm - 5:00 pm in IES 110. (Reception beforehand at 3:00pm on the 5th floor of BVM Hall).)
Speaker: Jorge Jaramillo (University of Chicago)
Title: Understanding working memory through a dynamical analysis of a neural circuit model
Abstract: Computational neuroscience uses mathematical models, computer simulations, and data analysis to understand how neural circuits give rise to cognition and behavior. One approach is to combine biological insight with dynamical systems theory to explain how networks of neurons encode, store, and manipulate information over time. In today's talk I will focus on working memory, a fundamental cognitive function whose disruption is implicated in a wide range of psychiatric disorders.
Working memory is the ability to sustain information beyond the presentation of a stimulus. A key neural mechanism thought to support working memory is persistent activity, the sustained firing of neurons after a stimulus has disappeared, which maintains a representation of that stimulus in the absence of external input. In this work, we study one property of working memory: robustness to interference from distractors. To this end, we use a simple neural circuit model that captures two core interactions between neural populations: recurrent excitation, where neurons reinforce each other’s activity and cross inhibition, where competing populations suppress one another’s activity. These interactions are summarized by the recurrent strength parameter JS, which allows the network to sustain persistent activity.
By analyzing the neural circuit dynamics, we show that increasing JS enlarges the memories' basin of attraction—the set of initial brain states that converge to a stable memory representation—thereby making working memory more robust to distractor interference. These results illustrate how the structure within neural circuits shapes the stability of maintained information and helps explain how the brain reliably preserves memories over short periods of time.
Bio: Jorge Jaramillo is an Assistant Professor in the Department of Neurobiology at the University of Chicago. Before joining UChicago, Dr. Jaramillo was a Junior Group Leader at the European Neuroscience Institute and Campus Institute for Dynamics of Biological Networks in Göttingen, Germany (2021-2023) and a postdoctoral associate in the lab of Prof. Xiao-Jing Wang at the New York University Center for Neural Science (2015-2020). He received a PhD in Computational Neuroscience as a Bernstein Fellow at the Humboldt University in Berlin, Germany under the supervision of Prof. Richard Kempter (2014).
Join the Department of Mathematics and Statistics for a colloquium on Thursday, March 26 from 4:00 pm - 5:00 pm in IES 110. (Reception beforehand at 3:00pm on the 5th floor of BVM Hall).)
Speaker: Jorge Jaramillo (University of Chicago)
Title: Understanding working memory through a dynamical analysis of a neural circuit model
Abstract: Computational neuroscience uses mathematical models, computer simulations, and data analysis to understand how neural circuits give rise to cognition and behavior. One approach is to combine biological insight with dynamical systems theory to explain how networks of neurons encode, store, and manipulate information over time. In today's talk I will focus on working memory, a fundamental cognitive function whose disruption is implicated in a wide range of psychiatric disorders.
Working memory is the ability to sustain information beyond the presentation of a stimulus. A key neural mechanism thought to support working memory is persistent activity, the sustained firing of neurons after a stimulus has disappeared, which maintains a representation of that stimulus in the absence of external input. In this work, we study one property of working memory: robustness to interference from distractors. To this end, we use a simple neural circuit model that captures two core interactions between neural populations: recurrent excitation, where neurons reinforce each other’s activity and cross inhibition, where competing populations suppress one another’s activity. These interactions are summarized by the recurrent strength parameter JS, which allows the network to sustain persistent activity.
By analyzing the neural circuit dynamics, we show that increasing JS enlarges the memories' basin of attraction—the set of initial brain states that converge to a stable memory representation—thereby making working memory more robust to distractor interference. These results illustrate how the structure within neural circuits shapes the stability of maintained information and helps explain how the brain reliably preserves memories over short periods of time.
Bio: Jorge Jaramillo is an Assistant Professor in the Department of Neurobiology at the University of Chicago. Before joining UChicago, Dr. Jaramillo was a Junior Group Leader at the European Neuroscience Institute and Campus Institute for Dynamics of Biological Networks in Göttingen, Germany (2021-2023) and a postdoctoral associate in the lab of Prof. Xiao-Jing Wang at the New York University Center for Neural Science (2015-2020). He received a PhD in Computational Neuroscience as a Bernstein Fellow at the Humboldt University in Berlin, Germany under the supervision of Prof. Richard Kempter (2014).