Memory is a large topic, built on the fundamental idea that the experiences one has can change the nervous system, so that behavior and mental activity can later be different as a result of what came before. Yet, memory is more than a record of personal experience. Humans can learn and then teach what they have learned to others, thereby making it possible to transmit information from one generation to another.

In the twentieth century the study of memory became part of the domains of both biological and psychological science. Work has proceeded at several levels of analysis – from questions about the cellular and molecular events that underlie synaptic change to questions about complex behavior. Between these poles are other important questions, such as what brain systems are important for memory and how they operate to support memory. As we enter the new millennium, biology and psychology have converged on a number of fundamental questions about memory. Is memory one thing or many? If there are different kinds of memory, what are their operating characteristics? Where in the brain do the important events occur? Where is memory stored? What happens at the level of individual cells and synapses?

The modern era of memory research can be said to have begun in 1957 when the effects on memory of medial temporal lobe resection were described in a patient who became known as HM. HM exhibited profound forgetfulness against a background of largely intact intellectual and perceptual functions.

Our understanding about the brain system that mediates memory began in the 1950s with the landmark case study of patient HM (Scoville & Milner, 1957). To relieve epilepsy that was intractable to pharmacological intervention, surgeons removed a large part of this patient's temporal lobes, including the amygdala, part of the hippocampus, and the cortex immediately surrounding the hippocampus and amygdala. Following surgery, HM exhibited a severe amnesia, leaving nonmemory aspects of intelligence and cognition intact. This observation demonstrated that memory could be separated from other cognitive functions and that structures of the medial temporal lobe are critical to memory.

While the early neuropsychological reports clearly pointed to the importance of the temporal lobes in memory, there was debate over precisely which temporal structures were important. Because the available clinical cases did not provide highly specific anatomical resolution, efforts were made to develop animal models in which experimental brain lesions could be performed with the necessary anatomical specificity. However, the early efforts to model amnesia in monkeys and rats did not yield a consistent pattern of severe and selective amnesia, precluding useful insights into the anatomical identification of the memory system. With hindsight, it is now clear that the difficulty in characterizing the brain system responsible for memory arose for two reasons (Eichenbaum et al., 2000). First, while the memory deficit following medial temporal damage was initially thought to be global in nature, it is now understood that damage to the medial temporal region causes amnesia that is limited to a specific domain of memory, and that other brain systems mediate other types of memory.

Understanding of sensory processes has been an ongoing concern for centuries. Consider the study of vision, for example. The early Greeks knew that seeing began in the eye and that the eye was filled with fluid – the humors. But they had no good sense of the eye's optics or of the retina's role; indeed, they thought the retina's job was to provide nutrients for the vitreous. Aristotle believed the humors were photoreceptive. The Pythagoreans believed that rays emanated outward from the eye to the external environment, leading us to wonder today as to what explanation they had in mind for why the world gets dark at night.

In 1604, one hundred years before the publication of Newton's Optics, the German astronomer Johannes Kepler launched a new era in vision when he noted that the eye worked as an optical instrument focusing an image sharply on the retina, observed that the retinal image is in fact oriented upside down and backwards, determined that the lens refracted light, and pinpointed the cause of myopia (before Kepler, people used spectacles to improve their vision but had no idea why curved glass sharpened their sight). There was plenty even Kepler failed to grasp however, including the actual workings of the retina. The discovery of photoreceptors – rods and cones – was over a century away, and when Treviranus officially discovered them in 1834, he misread the orientation of the retina, believing it was installed backwards in the eye (surely the light-sensitive photoreceptors could not be pointing away from the source of light and toward the dark interior of the eye socket!) This echoes the belief, apparently held by some in eras long past, that there exists a second lens in the eye that rectifies the visual image – for as everyone knows, we do not see our world upside down and backwards.

Introduction

A full understanding of the mammalian brain mechanisms underlying a higher cognitive phenomenon like learning and memory requires identification of relevant events or processes occurring at multiple levels of complexity; from molecular, synaptic, and cellular levels to neuronal ensemble and brain systems levels. This is an enormous challenge for brain researchers because cognitive phenomena can be monitored only at the level of a live animal's behavior, while many of the analytical methods for the underlying mechanisms are carried out using in vitro preparations and effective in vivo methods are limited. How can we be sure that the events or processes identified by in vitro methods or by even some in vivo studies are causally related to the animals' behavioral phenotype? For simpler invertebrate systems, molecular genetics has been effective for this purpose. Organisms harboring a mutation in a specific gene can be subjected to a variety of in vitro and in vivo analyses including behavioral tests, and deficits or impairments detected at different levels of complexity can potentially be bound together using the mutation as a connecting thread.

Background

Experimental strategy

For the analysis of more complex mammalian systems, however, additional tricks are necessary. One significant trick would be to restrict the mutation spatially and temporally. For instance, if one can restrict deletion (i.e., null mutation) of a specific gene to a particular type of neuron present in a particular area of the brain and only to a late phase of the animal's life, one can expect that the resulting deficits or impairments would be much more specific.

Introduction

Attention is relatively easy to define subjectively as in the classical definition of William James (1890) who said: “Everyone knows what attention is. It is the taking possession of the mind in clear and vivid form of one out of what seem several simultaneous objects or trains of thought.”

However, this subjective definition does not provide hints that might lead to an understanding of attentional development or pathologies. The theme of our chapter is that it is now possible to view attention much more concretely as an organ system. We follow the Webster dictionary definition of an organ system: “An organ system may be defined as differentiated structures in animals and plants made up of various cell and tissues and adapted for the performance of some specific function and grouped with other structures into a system.”

We believe that viewing attention as an organ system aids in answering many perplexing issues raised in cognitive psychology, psychiatry, and neurology. Neuroimaging studies have systemically shown that a wide variety of cognitive tasks can be seen as activating a distributed set of neural areas, each of which can be identified with specific mental operations (Posner & Raichle, 1994, 1998). Perhaps the areas of activation have been more consistent for the study of attention than for any other cognitive system. We can view attention as involving specialized networks to carry out functions such as achieving and maintaining the alert state, orienting to sensory events, and controlling thoughts and feelings.

The field of neuroscience is progressing so rapidly that even expressions such as “by leaps and bounds” fail to capture the pace of its growth. Questions that at one time were thought to be unanswerable – perhaps even unaskable – have now been asked and in some cases answered, and new questions once unthinkable are now asked matter-of-factly. Much of this acceleration is due to the maturing of the field – advances in techniques as well as in theory – fueled by an infusion of research support during the 1990s “Decade of the Brain” effort.

It is impossible to capture fully the sweep of discoveries and advances that emerged from that decade within the covers of a single volume. It is possible, however, to provide a sample of the best of that work, both as recognition of what has been accomplished during that period of time and since, and as a harbinger of what is surely to come as the pace of neuroscience shows no hint of slowing down.

Our goal in the present volume is to provide that sample through carefully chosen topics and even more carefully chosen researchers in those fields. Singling out the four most important problems in neuroscience is probably an unwise goal and is a surefire way to start an argument. That said, however, few would argue that the four featured here are anything less than powerful candidates for that inner circle: higher order perception; language; memory systems; and sensory processes.

The three chapters that follow this introduction all deal with aspects of visual perception related to the processing of scenes and the recognition of objects. There was a time when it was clear that higher order visual perception meant processing that took place in brain areas beyond the primary visual cortex. The primary visual cortex was thought to perform simple computations, each covering a small separate part of the visual world (receptive field) and hard wired in the sense that little could be done by learning or attention to modify them. This view stressed hierarchical processing among visual areas, particularly those from primary visual cortex V1 to the anterior temporal areas. Evidence for the hierarchical view is thoroughly summarized in the chapter by Kastner, De Weerd, and Ungerleider. However, all the three chapters deal in rather different ways with qualification to the hierarchical view of visual areas driven passively from the bottom up, based upon the influence of context, attention, and task demands.

In his chapter, Charles Gilbert describes the research work of his group, which has changed the view of how the primary visual cortex works. The older view gave rise to the hope that studies of primary visual cortex might provide the basic immutable building blocks from which it might be possible to launch an analysis of the remaining functions grouped under the title of higher perception.

Introduction

The primary visual cortex is the first cortical stage at which the visual world is analyzed. It has classically been thought to be a passive filter, only deriving information about local contrast and orientation, and passing that on to later cortical stages for the more complex task of object recognition. But a very different view is now emerging, showing that V1 plays a central role in much more complex processes involving intermediate level vision, integrating contours and parsing the visual world into surfaces belonging to objects and their backgrounds. The higher order properties of cortical neurons are reflected in the dependence of their responses on the context within which features of the visual stimulus are embedded. In addition, the properties of neurons in V1 reflect an ongoing process of experience-dependent modification, known as “perceptual learning.” This process begins early in life, incorporating the structural properties of the visual world into the functional properties of neurons. It continues throughout adulthood, encoding information about different shapes with which individuals become familiarized. Superimposed upon the influence of context and experience is a powerful top-down modification of neuronal function, such that the properties exhibited by neurons change according to attentional state, expectation, and perceptual task.

The receptive field and cortical circuitry: contextual influences

The central functional element of sensory systems is the receptive field, the portion of the sensory surface (retina) or environment (visual field) within which a stimulus will cause a cell to fire.

For all its diversity, one can view neuroscience as being concerned with two central issues – the hard wiring of the brain and the brain's capacity for plasticity. The former refers to how connections develop between cells, how cells function and communicate, and how an organism's inborn functions are organized (e.g., its sleep–wake cycles, hunger and thirst, and the ability to perceive the world). The nervous system has inherited such adaptations through evolution, because these are functions too important to be left to the vagaries of individual experience. In contrast, the capacity for plasticity refers to the fact that nervous systems can adapt or change as the result of experiences that occur during an individual lifetime. Experience can modify the nervous system, and as a result, organisms can learn and remember. Learning is the process by which new information is acquired about the world, and memory is the process by which this information can persist across time.

The scientific study of memory has reached a particularly fruitful stage. Memory is being studied at many levels of analysis – from questions about the cellular and molecular events that underlie synaptic change to questions about the organization of behavioral memory. This chapter considers memory from the perspective of brain systems and behavior and focuses on three topics (for recent reviews, see Squire & Bayley, 2007; Squire et al., 2004).

Introduction

Birdsong, like human speech, is a learned vocal behavior that requires auditory feedback. Both as juveniles, while they learn to sing, and as adults, songbirds use auditory feedback to compare their own vocalizations with an internal model of a memorized target song. Here we describe experiments that explore the properties of the songbird anterior forebrain pathway (AFP), a basal ganglia–forebrain circuit known to be critical for normal song learning and for adult modification of vocal output, but not for normal adult singing. First, neural recordings in anesthetized, juvenile birds show that single AFP neurons become specialized to process the song stimuli that are compared during sensorimotor learning. AFP neurons develop tuning to the bird's own song, and in many cases to the tutor song as well, even when these stimuli are manipulated to be very different from each other. Second, neural recordings from adult, singing birds reveal robust singing-related activity in the AFP, which is present even in deaf birds. This activity is likely to originate from premotor areas, and could represent an efference copy of motor commands for song, predicting the sensory consequences of motor commands. Finally, in vitro studies of the AFP show that recurrent synapses between neurons in the AFP outflow nucleus can undergo activity-dependent and timing-sensitive strengthening that appears to be restricted to young birds.

Introduction

Much of human social life depends on the notion that agents have control over their actions and are responsible for their choices. In daily life it is commonly assumed that it is fair to punish and reward behavior so long as the person is in control and makes choices knowingly and intentionally. Without the assumptions of agent control and responsibility, human social commerce is hardly conceivable. As members of a social species, we recognize co-operation, loyalty, honesty, and helping as prominent features of the social environment. We react with hostility when group members disappoint certain socially significant expectations. Inflicting disutilities (e.g., shunning, pinching) on the socially erring and rewarding civic virtue help restore the standards.

In other social species too, social unreliability, such as a failure to reciprocate grooming or food-sharing, provokes a reaction likely to cost the erring agent, sooner or later. In social mammals at least, mechanisms for learning and keeping the social order seem to be part of what evolution has bequeathed to our brain circuitry. Given that the stability of the social-expectation baseline is sufficiently important for survival, individuals are prepared to incur some cost in enforcing those expectations. Just as anubis baboons learn that tasty scorpions are to be found under rocks but cannot just be picked up, so they learn that failure to reciprocate grooming when it is duly expected may incur a slap.

Introduction

The human faculty of language is a breathtaking skill. It allows us to communicate observations, thoughts, wishes, intentions, emotions, etc., to another person in the same room (by speaking), to a person in the next room (by shouting), to someone thousands of kilometers away (by speaking on the telephone or sending a fax), and even to future generations (by writing stories, poems, books, or scientific articles). Language is characterized by almost infinite variation and creativity. Every person alive today (with the exception of pre-verbal infants and people with severely impaired language skills) probably utters a number of sentences every day that he or she has never produced before. What other form of behavior could compete with this for degree of novelty and originality?

Language is typically considered to involve a set of interacting, but somewhat separate, domains of ability or knowledge. These include the sound structure of the language (phonology); word meanings (semantics); the ways in which individual morphemes combine to create complex words (morphology); the ways in which morphologically simple or complex words combine to create phrases and sentences (syntax); and finally, at least in the relatively brief time since a substantial proportion of the world's population has become literate, knowledge of how words are written in the speaker's language (orthography).

How and where does the brain represent and process this complex set of abilities? Because language is unique to humans, we can only learn about this topic by studying humans.

The last decade of the twentieth century was unprecedented in its progress toward discoveries linking the anatomical structures and physiological systems of the brain to the human mind. This enterprise is possible now, both because of a large body of behavioral data characterizing the operations and subsystems within different domains of cognitive processing and because of great advances in the methods and techniques available to noninvasively image the structure and the physiology of the functioning human brain. The focus of this part is to consider different perspectives and approaches to the study of the brain systems important in language processing and in the development and differentiation of the language systems of the brain.

The study of language is particularly well poised to benefit from knowledge about underlying neural mechanisms. It has been recognized since the 1950s that the study of language is a model case for understanding the species-specific capacities of human learners and the brain mechanisms in human adults and infants that permit them. Language in humans is an extraordinary ability, showing many properties without parallel in other species; understanding the mechanisms underlying human language will therefore shed special light on human cognition. At the same time the lack of animal models that have made such powerful contributions to the characterization of nonlinguistic cognitive systems underscores the importance of the new noninvasive techniques for imaging the language systems of the human brain.

Neuroethics overview

The reach of neuroscience is unusually broad. As the title of this book indicates, our present interest in neuroscience extends all the way from cells to cognition – from how neurons operate at the microscopic level to how people think, speak, perceive, and remember at the macroscopic level. Given that the brain is the most complex structure in the known universe, this breathtaking breadth may come as no surprise and makes it understandable why neuroscience is of compelling interest to engineers, physicians, computer scientists, and even to musicians.

The reach of our concerns is so great that it engages even philosophers, including those grappling with the most difficult – perhaps intractable – questions of them all, those touching on the question of human choice and free will. In this special chapter with which we lead off this book, the noted philosopher Patricia Churchland takes on the challenge of reconciling the seemingly deterministic neurological system underlying human choice behavior, the common belief in free will, and the question of who, if anyone, is responsible for the behavior of human beings. When people commit crimes or other ethical breeches, may they rightfully claim that “their neurons made them do it”? Or, is there lurking within their nervous systems an accountable agent that must take responsibility for decisions made?

Introduction

The language system is that aspect of mind/brain function that forms the basis for phonological, morphological, syntactic, and semantic computation. The “currencies” (or the ontology) of this central and abstract computational system are representations that are amodal, for example the concepts “feature” (phonology) or “affix” (morphology) or “phrase” (syntax) or “generalized quantifier” (semantics). Representation and computation with such concepts is typically considered independent of sensory modalities. Of course, the linguistic computational system is not isolated but interacts with other cognitive systems and with sensory–motor interface systems.

With regard to the input and output, the system has at least three modality-specific interfaces: an acoustic-articulatory system (speech perception and production), a visuo-motor system (reading/writing and sign), and a somatosensory interface (Braille). Speech and sign are the canonical interfaces and develop naturally; written language and Braille are explicitly taught: barring gross pathology, every child learns to speak or sign (rapidly, early, without explicit instruction, to a high level of proficiency), whereas learning to read/write Braille requires explicit instruction, is not universal, and occurs later in development.

In this chapter we focus on speech perception, specifically with regard to linguistic constraints and cortical organization. We first outline the key linguistic assumptions, including the concept of “distinctive feature,” and then discuss a functional-anatomic model that captures a range of empirical findings.

The linguistic basis of speech perception

The central importance of words for language use and understanding

An essential part of the cognitive ability underlying the linguistic behavior of a competent speaker of a language consists of knowing the words of the language or their constituents (roots).