Neuronal Group Selection and Spike Timing Dependent Plasticity

In 1987 Dr. Gerald Edelman published a book titled “Neural Darwinism: Theory of Neuronal Group Selection (Oxford paperbacks)” where he presented his idea that selective processes are fundamental to brain function. He argued that the gross connectivity and microstructure of the brain in conjunction with known synaptic interactions amongst neurons provide a substrate for the dynamic formation and maintenance of relatively small groups of neurons that form the information processing substrate in the brain. Today we’ll take a peak into some of what Dr. Edelman has been up to lately.

Distribution of neuronal groups across a simulated cerebral cortical surface.
Figure 1. Distribution of groups on the simulated ‘cortical’ surface containing 100,000 neurons. Neurons within each group are marked by the same color. Most groups contained less than 100 neurons. From “Spike-timing Dynamics of Neuronal Groups.” By Eugene M. Izhikevich, Joseph A. Gally, and Gerald M. Edelman. Cerebral Cortex Volume 14 No. 8, May 13, 2004.

Current models of group selection in Edelman’s laboratory derive from the simulation of cerebral cortex described in the paper “Spike-timing Dynamics of Neuronal Groups” published May 13, 2004 in Cerebral Cortex. The dynamics of this cortical model are governed by spiking neurons, axonal conduction delays, and spike timing dependent plasticity (STDP). In spike timing dependent plasticity, when a presynaptic neuron fires before a postsynaptic neuron, the synapse is potentiated. Conversely, when a postsynaptic neuron fires before a presynaptic neuron, the synapse is depressed. In the paper, the research team uses the model to describe the collective behavior and self-organization of spiking neurons to form neuronal groups.

The computer model of the cerebral cortex had 100,000 neurons. There were 80,000 excitatory neurons that made 8 million synaptic connections onto other neurons and 20,000 inhibitory neurons that made 500,000 synaptic connections. Synapses exhibited both short-term depression and facilitation, as well as long-term spike timing dependent plasticity. All synapses in the network underwent spike timing dependent plasticity all the time. Does this plasticity enable self-organization in the model?

The answer they found after running the simulations was yes. They showed that the interplay between delays and spike timing dependent plasticity gave rise to the spontaneous formation of neuronal groups defined as “small collectives of neurons having strong connections with matching conduction delays and exhibiting time-locked but not necessarily synchronous spiking activity” (see Figure 1 above). Individual groups may change over time, compete or cooperate with other groups, and even may go extinct.

Spike timing dependent plasticity is currently central in the thinking about neuronal group selection. One of the scientists, Dr. Eugene Izhikevich, who has worked in Edelman’s laboratory has also added methods in dynamical systems into the mix. Tomorrow we’ll look at his some of his recent work.

Other related blog posts:

Memory and the Precise Timing of Signals in the Brain