This chapter overviews memory and learning from four different perspectives. First and foremost, it reviews the philosophical models of human memory. In this regard, it examines Atkinson and Shiffrin's model, Tulving's model, Tveter's model, and the wellâknown parallel and distributed processing (PDP) approach. The chapter also gives an overview of the philosophical research results on procedural and declarative memory. Second, the chapter is concerned with coding for memory and memory consolidation. Third, the chapter is concerned with a discussion on cognitive maps, neural plasticity, modularity, and the cellular processes involved in shortâterm memory (STM) and longâterm memory (LTM) formation. Finally, the chapter deals with the scope of brain signal analysis in the context of memory and learning. Possible scope of computational intelligence techniques on memory modeling is also appended at the end of the chapter.
1.1 Introduction
The human nervous system comprises several billions of neurons spread across the brain, spinal cord, and the rest of our body. These neurons collectively and/or independently participate in the cognitive processes undertaken by the brain. Usually, the efferent neurons receive stimuli from the receptors present in the cell membranes and carry the electrical activation due to the stimuli to the brain to recognize and interpret the stimuli. The brain in turn generates response through afferent neurons to trigger specific localized organs for its activation. Consider, for example, the experience of touching a hot body by a twoâyear old baby. Presume that the baby has no prior experience to touch a hot body. As she touches the hot body accidentally/incidentally, the efferent neurons present in the receptor (neurons) of her skin receives thermal stimulation, the electrical activation of which reaches her brain, and the motor command generated by the brain is then transferred to her limbs to withdraw her hand. The firstâhand experience of the baby is unconsciously recorded in her brain, to provide her a cautionary support to avoid similar incidents in future. A natural question that appears before us is where does the baby save her learning experience? How does she automatically retrieve her knowledge to avoid similar situations in future?
The book aims at offering answers to the previously mentioned queries and the like by analysis of the acquired brain signals/images during memory formation (encoding) and memory recall stages in adults. Although very little of human memory encoding and recall processes are known till this date, it is almost unanimously accepted that the human memory is distributed in the cortex with localized activities in certain brain regions. For instance, the hippocampal region, residing in the medial temporal lobe, is found to have good correlations with relatively permanent LTM. Two other forms of shortâduration memory are also reported in the literature [1,2]. They are popularly known as STM and working memory (WM). It is known that STM can hold information for few minutes only [1â3], unless it is refreshed periodically. The WM, on the other hand, provides a support to human reasoning and apparently looks like cache memory in computer systems. It may be remembered that the central processing unit (CPU) in the computer receives and saves information from the cache while executing a program segment. Although major part of a selected program resides in the system random access memory (RAM), the cache saves only fewer bytes of storage currently under execution. The cache is designed with high speed logic circuits, such as emitterâcoupled logic or integrated injection logic (I2L) [4â6] to maintain parity in speed with the processor. Similarly, the brain performs reasoning time efficiently, which is often bottlenecked by relatively low speed LTM. The WM thus bridges the speed gap between human reasoning system and the LTM access, which usually is sluggish with respect to our speed of logical reasoning.
The book is all about WM and STM encoding and recall, with a small coverage on interactions between the WM and the STM. Although there is a magnificent reporting on memory encoding and recall, most of the research outcomes are based on behavioral experiments on humans [7]. Thus the existing research results cannot offer the cognitive basis of memory encoding and recall. With the advent of modern brain imaging and signal acquisition equipment, it is now possible to make a thorough study on memory encoding/recall processes. Although such study provides a more scientific basis to understand the memory encoding and retrieval processes, they too are not free from limitations. For instance, the existing nonâinvasive techniques mostly rely on scalp potential and thus can hardly capture single neuron activation. So, the analysis is undertaken on the local response of a group of neurons. Second, while administering memory activity, the other activities of the neuron also appear on the scalp and thus act as noise input to the memory study. Elimination of the noise is not easy here as the noise distribution often falls in the same frequency spectra used by the memory system.
The mystery of memory formation largely relies on the regulatory and control mechanism of the cellular proteins. A brief review of molecular biology reveals that the neuronal cells, like any other cells in the human, contain deoxyribonucleic acid (DNA) double helix comprising several millions of four bases (adenine [A], guanine [G], thymine [T], and cytosine [C]). These four bases have an apparently random (positional) occurrence in the individual string of a DNA. Small sequences of such bases on the DNA that are responsible for inheritance of genetic materials from parents to children are called genes. The neuronal cells containing DNA double helix thus contain genes, which often translate to form cell proteins. The protein formation by DNA and particularly genes is a twoâstep process. In the first step, the DNA translates to ribonucleic acid (RNA), and in the second step, the RNA transforms into proteins. These cell proteins help in permanent/semiâpermanent encoding of the acquired information in the LTM. How the protein help in encoding is a complex biochemical process, very little of which is known at present.
This chapter is organized into 11 sections. In Section 1.2, a philosophical survey to memory is undertaken. Section 1.3 is concerned with the brainâtheoretic interpretation of memory formation. This section also takes into account the experimental perspectives of memory and learning. It includes both surgical and therapeutic experiments on memory encoding by considering plasticity and stability issues of memory and learning. Sections 1.4, 1.5, 1.6, 1.7, and 1.8 are concerned with cognitive maps, neural plasticity, modularity, and the cellular process behind STM formation and LTM formation, respectively. Section 1.9 deals with brain signal analysis in the context of memory and learning. Section 1.10 examines the scope of mathematical/computational models of memory and learning. Section 1.11 reviews the scope of the book. This section also provides a summary of the work presented and future directions of research in memory and learning.
