Skip navigation and jump directly to page content
Indiana University Bloomington
Choose site to be searched
Type search terms

Q733 Colloquium

Colloquia occur: Selected Mondays at 4:00 pm - 5:00 pm - Room PY 101.
Colloquia titles will be posted as they become available.

Organizer: Rob Goldstone
Phone: 855-4853
Email: rgoldsto@indiana.edu

Fall 2014 Q733 Colloquia

  • Sep 8, 2014 - Danielle Bassett, University of Pennsylvania
  • Sep 15, 2014 - Thomas Serre, Brown University
  • Sep 29, 2014 - Uri Hasson, Princeton University
  • Oct 13, 2014 - Todd Braver, Washington University
  • Nov 3, 2014 - Yael Niv, Princeton University
  • Dec 1, 2014 - Daphne Maurer, McMaster University
  • Dec 8, 2014 - Matthew Botvinick, Princeton University

Abstract

Sep 8, 2014: Danielle Bassett, University of Pennsylvania
Title: Q733 Colloquium
Abstract: Human learning is a complex phenomenon requiring network-wide flexibility to adapt existing brain function and precision in selecting new neurophysiological activities to drive desired behavior. Using functional connectivity measurements of brain activity acquired from initial training through mastery of a simple motor skill, we investigate the properties of brain network dynamics that predict individual differences in learning. Functional interactions between brain regions co-evolve with one another during learning in distributed patterns that decrease in size with practice, indicating the emergence of an autonomous subgraph whose dynamics no longer depends on other parts of the network. This consolidation of network dynamics is mirrored in higher-level summary statistics describing the modular organization of the brain, a property that plays a critical role in the selective adaptability present during evolution, development, and optimal network function. Our results indicate that more flexibility during early practice sessions, which we measure by the allegiance of nodes to modules, predicts more extensive learning in later practice sessions. Flexibility is greatest in a periphery of high-level processing regions whose connectivity changes frequently, and is least in a relatively stiff core of output regions whose connectivity changes little in time. The temporal core-periphery structure of human brain dynamics provides a fundamental new approach for understanding how separable functional modules are linked. This, in turn, enables the prediction of fundamental capacities, including the production of complex goal-directed behavior, in humans.

Sep 15, 2014: Thomas Serre, Brown University
Title: Q733 Colloquium
Abstract:

Sep 29, 2014: Uri Hasson, Princeton University
Title: Hierarchical process-memory: a new ecologically and biologically plausible memory of systems model of neural processing
Abstract: Traditional models of memory dissociate memory from processes. Such tendency is rooted in the analogy between computers’ architecture and the brain, which dissociate the central processing units (CPUs) which carry out the basic computations from the memory units. Based on such conceptualization, many empirical studies focus on simple delay periods (e.g. match-to-sample tasks) in which memory has to be actively maintained but not processed and cases in which the integration between past and present information is undesirable (e.g. when remembering a target word in a list of distractors). However, such models are not applicable to the majority of processes in which the past and present are in continuous interaction. In real life, memory is integral to all neural processes across multiple timescales, as the past and present converge continuously in the processes of incoming information. For example, a phoneme only achieves full meaning in the context of a word, a word only achieves full meaning in the context of a sentence, and a sentence only achieves full meaning in the context of a conversation. The capacity to integrate information over multiple timescales is, therefore, functionally essential, and memory is central to all neural processes. In my talk I will outline a new model for memory that resists the tendency to separate memory from process. The model argues that all neural circuits, ranging from early sensory areas to high order areas, has the capacity to accumulate information over time. Memory is intrinsic to each and any neural circuit, and is essential for its ability to process information. Furthermore, our data suggest that the process-memory timescale increases from early sensory areas to high order areas. For example, some areas have a short (tens of milliseconds) process-memory timescale (e.g. which enables the integration of few phonemes over time for detecting a word). Other areas have a mid (few seconds) process-memory timescale (e.g. which enables the integration of few words over time for parsing a sentence). While other areas have a long (hundreds of seconds) process-memory timescale (e.g. which enables the integration of sentences over time for comprehending a narrative). Our hypothesis, that each brain area accumulates information over its preferred timescale, suggests that memories of the recent past are not stored in a few localized working memory buffers, but rather are distributed in an organized hierarchical topography throughout the nervous system.

Oct 13, 2014: Todd Braver, Washington University
Title: Q733 Colloquium
Abstract:

Nov 3, 2014: Yael Niv, Princeton University
Title: Q733 Colloquium
Abstract:

Dec 1, 2014: Daphne Maurer, McMaster University
Title: Q733 Colloquium
Abstract:

Dec 8, 2014: Matthew Botvinick, Princeton University
Title: Q733 Colloquium
Abstract:


Previous Q733 Colloquia