|
The first part of the talk will be about self organisation of neuronal activity in the liquid state machine while the second part will report on synchronous spiking activity revealed in simultaneous recording from up to 61 cells that is stimulus specific and task correlated and were recorded in the pre-frontal cortex and motor cortex from awake monkeys as well as from the visual area A17 from cats. The first part: The liquid state machine had been proposed as a promising computational neuronal model. However, so far the idea of the liquid state machine were based on fixed and random networks and therefore havenąt incorporated any kind of plasticity that can shape the networks structure as well as the network activity. We demonstrate that the combination of two kinds of plasticity, first the spike timing dependent plasticity (STDP) that changes synaptic strength, and second the intrinsic plasticity (IP) that changes the excitability of individual neurons to maintain homeostasis, lead to computational properties for time series processing that are more optimal than in case of random networks. The second part: In the second part we are presenting results revealed by a new method called NeuroXidence that help resolving one of the most intensively discussed topics in the recent years: The importance of synchronous neuronal firing for information processing in the neuronal system. NeuroXidence has been designed to deal with four main properties of neuronal spike trains that make the analysis of coordinated firing events difficult: These are variability, short timescales, history dependencies, and auto-structures of the spiking activities, as well as the rareness synchronous firing events. We demonstrate both on toy-data and single-unit activity recorded in cat and monkey cortex, that NeuroXidence discriminates reliably between significant and spurious events that occur by chance. We present data recorded from up to 27 cells in pre-frontal cortex of two monkeys during a short term memory task that support strongly the hypothesis that synchronous activity is involved in the maintenance in the short term memory. In a second dataset we demonstrate that synchronous activity recorded in the monkey motor cortex is related to expectancy. And in a third data set recorded in cat area A17 from up to 61 cells simultaneously we demonstrate that the formation of synchronous neuronal cell assemblies is stimulus specific. Taking together the robustness of the new method NeuroXidence, and the experimental results we present strong evidence that stress the importance of synchronous activity for information processing in the mammalian cortex. This work was supported by the Hertie and VW foundation. |
|