Technical Program

Paper Detail

Paper: PS-1B.59
Session: Poster Session 1B
Location: H Fläche 1.OG
Session Time: Saturday, September 14, 16:30 - 19:30
Presentation Time:Saturday, September 14, 16:30 - 19:30
Presentation: Poster
Publication: 2019 Conference on Cognitive Computational Neuroscience, 13-16 September 2019, Berlin, Germany
Paper Title: Testing burst coding models of working memory with spike trains from primate prefrontal cortex
Manuscript:  Click here to view manuscript
License: Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 Unported License.
Authors: Daming Li, Yale University, United States; Christos Contantidinis, Wake Forest University, United States; John Murray, Yale University, United States
Abstract: Working memory (WM) is the brain’s ability to actively maintain information over a span of seconds, for which a hallmark neuronal correlate is stimulus-selective persistent spiking activity in prefrontal cortex. It was recently proposed, based on local field potential analysis, that this persistent activity might be an artifact of trial averaging, and that WM is instead subserved on single trials by sharp intermittent bursts of activity. However, this proposal remains untested on single-neuron spiking activity. We analyzed a doubly-stochastic statistical model of neuronal spiking to derive testable predictions for how burst-coding proposals impact measures of spiking variability, such as Fano factor. We tested these predictions with multiple datasets of single-neuron spike trains recorded from macaque prefrontal cortex during different WM tasks. Neurons exhibit a global decrease in Fano factor during the delay, relative to foreperiod, and few well-tuned neurons exhibit substantial WM-dependent burstiness. Furthermore, we analyzed computational models of WM circuits, which can subserve WM by persistent attractor dynamics, or by burst coding via short-term synaptic plasticity, demonstrating analyses that dissociate distinct circuit mechanisms. In summary, prefrontal spiking variability supports theoretical frameworks of persistent activity supporting WM, and provides strong constraints on proposals for WM coding through intermittent single-neuron bursting.