Distributed Fading Memory for Stimulus Properties in the Primary Visual Cortex

Danko Nikolić,Wolf Singer,Wolfgang Maass,Stefan Häusler,Jonathan D Victor

doi:10.1371/journal.pbio.1000260

Abstract

Author SummaryResearchers usually assume that neuronal responses carry primarily information about the stimulus that evoked these responses. We show here that, when multiple images are shown in a fast sequence, the response to an image contains as much information about the preceding image as about the current one. Importantly, this memory capacity extends only to the most recent stimulus in the sequence. The effect can be explained only partly by adaptation of neuronal responses. These discoveries were made with the help of novel methods for analyzing high-dimensional data obtained by recording the responses of many neurons (e.g., 100) in parallel. The methods enabled us to study the information contents of neural activity as accessible to neurons in the cortex, i.e., by collecting information only over short time intervals. This one-back memory has properties similar to the iconic storage of visual information—which is a detailed image of the visual scene that stays for a short while (<1 s) when we close our eyes. Thus, one-back memory may be the neural foundation of iconic memory. Our results are consistent with recent detailed computer simulations of local cortical networks of neurons (“generic cortical microcircuits”), which suggested that integration of information over time is a fundamental computational operation of these networks.

Highlights

A number of analysis methods have been designed to investigate how neuronal spiking activity correlates to sensory stimulation or behavior and, as expected, many relations have been found
To emulate classification processes realized with leaky integrateand-fire (I&F) readout neurons, the time stamps of discharge sequences were first convolved with an exponentially decaying kernel that mimicked the time course of excitatory postsynaptic potentials (EPSPs) [38,39]
In a few additional analyses, we used support vector machines (SVM) with polynomial and radial basis function kernels to investigate whether classification performance improved with these more sophisticated, nonlinear transformations of the input variables

Summary

Introduction

A number of analysis methods have been designed to investigate how neuronal spiking activity correlates to sensory stimulation or behavior and, as expected, many relations have been found. The analysis method should be allowed no more time for accumulating evidence than is available for cortical neurons to accomplish their task. To fulfill these requirements, readout neurons need to be simulated on a computer and fed with data obtained in parallel recordings from a large number of cortical neurons, the assumption being that a certain fraction of these neurons provides input to neuronal classification

Methods

Results

Discussion

Conclusion