Chapter 6 - Computations in memory systems in the brain

Abstract

This chapter describes memory systems in the brain based on closely linked neurobiological and computational approaches. The neurobiological approaches include evidence from brain lesions, which show the type of memory requiring each of the brain systems to be considered; and analysis of neuronal activity in each of these systems to show the kind of information represented in them and the changes that take place during learning. The computational approaches are essential to understand the retrieval of circuitry and stored memories, the capacity of each memory system in the brain, the interactions between memory and perceptual systems, and the speed of operation of the memory systems in the brain. The chapter also describes the ways to combine evidence from neurophysiology, the effects of lesions, and computational neuroscience to understand the working of some memory systems in the brain. Evidence from single-neuron (and multiple single-neuron) recordings is an important part of this approach because it is at the neuronal level that the information is being communicated between the computing elements of the brain and the neurons. Here, the spiking activity of neurons is recorded and analyzed to determine the computational inputs and outputs pertaining to a particular area. The computational model then simulates the activity of the different neurons in the area; utilizing evidence based on the connectivity of the neurons, the synaptic modification rules, and a theory of computation, and can be tested by direct comparison with the neuronal activity recorded in a brain area.