CNS*2002 Workshops

University of Chicago, July 25, 2002

Neural Assemblies

Thomas Wennekers (MPI for Mathematics in the Sciences, Leipzig, Germany)
Fritz Sommer (University of Ulm, Germany)
Ad Aertsen (Albert-Ludwigs University Freiburg, Germany)

Mini-symposium, 1 session.

Donald Hebb (1949) introduced cell assemblies as a biophysical theory of psychology. According to Hebb, stimuli, objects, but also more abstract psychological entities like concepts, contextual relations, ideas, are represented in the brain by simultaneous activation of groups of neurons he referred to as cell assemblies. Cell assemblies may overlap, that is, single neurons may belong to many different cell assemblies. The definition of a cell assembly is based on connectivity; cells forming an assembly are connected by relatively numerous and/or strong mutual excitatory synapses. Thereby, cell assemblies are characteristic states for the neural network dynamics (attractors): if a sufficient number of cells in an assembly receives stimulation, all cells in the assembly tend to become coactivated. Such `ignition' of a cell assembly can be viewed as association between a stimulus and a previously learned long-term memory. Already James (1890) presumed that habits (long-term memories) might be formed by activity-dependent changes of synaptic efficacies. In his theory of cell assemblies Hebb postulated specifically that synapses get strengthened when both connected cells are activated simultaneously within a certain time window. This learning mechanism, widely known as `Hebbian learning rule' has been confirmed experimentally. For short term memory cell assemblies provide a simple model as well: the mutual excitation can support active maintainance of a memory after the cueing stimulus has vanished. Cell assemblies might either be localized within a cortical column/area using short-range connectivity or spread over different cortical and subcortical areas/structures using long-range projections. The latter describes associative interactions between different cortical systems.

Since Hebb and James, a multitude of experimental and theoretical studies have targeted different questions about the nature of neural assemblies:

  • Neuronal coding in spatio-temporal coordination of single-unit activity has been studied in experiments with multiple-electrode recordings. The questions that have been addressed include detection, characterization and functional interpretation of coordinated neuronal firing.
  • Neural associative memories (of the `classical' Willshaw- or Hopfield-type) have been proposed as mathematical models describing the essential working principles of cell assemblies. Analysis of these models allows to characterize the type of neural computation possible on the basis of cell assemblies. Recent associative memory studies have reflected the biophysics of the cortex more realistically by including for example, spiking or conductance based neurons, learning rules based on synaptic physiology, various sources of noise. These studies have addressed questions about the computational function of biophysical structure and processes, like the role of feedback connectivity, spike correlations, periodic activity and synchronization, persistent activity, etc.
  • Cell assemblies play a role in a variety of models for higher cognitive functions involving the interplay between several cortical and/or subcortical associative structures. The focus of these cell assembly models is to study the nature of context and concept representation, information fusion, memory-consolidation, memory-mood interaction, or attentional and neuromodulatory influences on memory systems. As a more recent trend orbito-frontal control mechanisms are integrated into such large scale models, too.

The present workshop aims to provide a platform at this year's CNS meeting for the presentation and discussion of recent progress in the field of neural assemblies. Contributions focusing on all aspects of neural assemblies are welcome, ranging from experimental issues and data analysis methods to modeling and theory.

Suggested topics:

  • coding: anatomical vs functional connectivity, functional segregation vs integration population coding
  • dynamics: oscillations, synchronization, spike-patterns, pattern-sequences synaptic plasticity, temporal learning rules
  • memory: long- and short-term memory in associative structures, concepts, context, multi-modal integration attentional, motivational, emotional, neuromodulatory influences on memory processes
  • behavior: large-scale coordination of memory processes, orbito-frontal cortex models, cortico-hippocampal interplay, integrative systems

Confirmed speakers:

  • Ad Aertsen (University of Freiburg)
  • Chip Levy (University of Virginia)
  • Fritz Sommer (University of Ulm)
If you would like to contribute to the workshop, please send a brief description of topics addressed in your talk as soon as possible to

Suggested reading:

  • Donald Hebb: The organization of behavior. Wiley, 1949.
  • Guenter Palm: Neural Assemblies: An alternative approach to artifical intelligence. Springer, 1982.
  • Guenther Palm and Ad Aertsen: Brain Theory. Springer, 1986
  • George L. Gerstein, Kyle L. Kirkland: Neural Assemblies: technical issues, analysis, and modeling. Neural Networks 14, 589-598, 2001.