In composing this final chapter, we were faced with a number of decisions. On the one hand, we thought it useful to point out the shared elements and agreements amongst the chapters, as well as their differences and novel contributions. On the other hand, in spite of our efforts to ensure that the chapters of the book covered all of the important issues in LI research and theory, a number of topics remained either unresolved or unexplored. As a compromise, we have opted for the broadest possible summary, recognizing that it will be at the expense of depth, a debt that can be redeemed only by writing another book.

The present chapter, then, begins by summarizing a few latent inhibition (LI) facts and the major themes that are common to several or more chapters. Following this, methodological issues that have not been resolved, and which, we believe, deserve the future attention of researchers, are addressed. These include the role of masking, the need for new LI procedures, the distinction between two- and three-stage LI procedures, and the dual source of LI effects. The next major section takes up theoretical issues, examining specific behavioral theories of LI, with an attempt to integrate them, while pointing out the relevance for future research. The final section discusses the neuropsychology of LI particularly as it relates to schizophrenia.

Historically, the suggestion that dopaminergic (DAergic) neurons are involved in the latent inhibition (LI) phenomenon is linked to psychopharmacological studies carried out by two laboratories reporting an attenuation of LI responses in animals (rats) treated with the indirect DAergic agonist d-amphetamine, after chronic (Solomon, Crider, Winkelman et al.,1981; Weiner, Lubow & Feldon, 1981, 1984) or acute administration (Weiner, Lubow & Feldon, 1988). Involvement of DAergic neurons in LI was further supported by data showing that the LI attenuation induced by d-amphetamine was reversed by the concomitant administration of the neuroleptic chlorpromazine (Solomon et al., 1981), and by subsequent studies showing a facilitation of LI expression after administration of haloperidol (Weiner & Feldon, 1987; Weiner, Feldon & Katz, 1987), a well-known typical neuroleptic with a potent blockade action on DA receptors. Since these first studies, the reversal of the d-amphetamine-induced LI reduction by DAergic antagonists has been found in different LI paradigms after administration of several atypical neuroleptics, including olanzapine (Gosselin, Oberling & Di Scala, 1996) and clozapine (Trimble, Bell & King, 1998; Russig, Murphy & Feldon, 2002). In other respects, enhancement of LI expression has also been reported, with the atypical antipsychotics displaying a selective blockade action on D2 receptors such as sulpiride (Feldon & Weiner, 1991) or remoxipride (Trimble, Bell & King, 1997), whereas selective D1 antagonists were found to have no effect on LI phenomenon (Trimble, Bell & King, 2002).

Summary

Latent inhibition and habituation are behavioural phenomena that result from the same simple procedure: when an animal receives exposure to a stimulus, the unconditioned responses that were once provoked by that stimulus diminish or habituate, and subsequently the rate at which conditioned responding can be established to the stimulus is reduced. One parsimonious interpretation of these phenomena is that they reflect the operation of the same attentional process: a novel stimulus attracts attention, provokes responding and is rapidly learnt about; exposure to a stimulus produces a decline in attention and concomitant reductions in responding and learning. Here, we present an integrative review of recent research that provides support for a specific account of habituation and latent inhibition that is grounded in a formal analysis of the mnemonic processes involved in associative learning. This analysis also provides one interpretation of the role of the hippocampus, a structure traditionally associated with both mnemonic and attentional processes, in habituation and latent inhibition.

Introduction

Latent inhibition (LI) is a phenomenon in both senses of the word: it is a very simple and ubiquitous behavioural effect, and it has stimulated intense empirical and theoretical interest, and some controversy over a protracted period. After a faltering start (see Lubow, this volume), the observation first reported by Lubow and Moore (1959) has become a central concern for theoretical analyses of associative learning: once its generality had been established, LI became the territory of models of associative learning wherein variations in attention play a key role (e.g., Lubow, 1989; Mackintosh, 1975; Pearce & Hall, 1980).

Introduction

After 50 years of latent inhibition (LI) research, it may come as a surprise that there is, as yet, no generally accepted theoretical account of this apparently simple, robust effect (i.e., a deficit in conditioned responding as a result of prior exposure to the conditioned stimulus). A variety of explanations have been proposed, and many of them, such as conditioned inhibition and habituation, discarded (for a review, see Lubow,1989, pp. 141–189; but see Honey, Iordanova & Good, this volume; Schmajuk, this volume). Currently, one can identify three basic types of LI theories. A-theories, which provided the first accounts of LI (e.g., Lubow, Weiner & Schnur, 1981; Mackintosh, 1975; Pearce & Hall, 1980; Wagner, 1976, 1981; more recently, McLaren & Mackintosh, 2000), accept that passive, inconsequential stimulus preexposures reduce the ability of that stimulus to enter into new associations. In addition, most A-theories explicitly acknowledge the role of attention and attribute the reduction of associability to the decline in stimulus-specific attention or salience.

In contrast to A-theories, R-theories eschew attentional concepts and maintain that stimulus preexposure has no effect on stimulus associability; the stimulus preexposed (PE) and non-preexposed (NPE) groups enter the acquisition stage with the same capability for forming new associations with the unconditioned stimulus. According to R-theories, in the stage-three test, the associations formed during the stage-one preexposure compete for expression with the association formed in the stage-two acquisition (e.g., Bouton, 1993; Miller, Kasprow & Schachtman, 1986; Weiner, 1990).

The complexity of the environment is such that even a snapshot in time from a single modality can exceed the processing capacity of the human brain (Tsotsos,1990). This means that selection of some incoming stimuli for more detailed analysis (while ignoring other stimuli) is essential for efficient cognitive processing. Without such filtering we would be inundated with competing sensory impressions. Thus notions of stimulus salience and relevance have assumed prominence in a number of domains, from visual attention (Fecteau & Munoz, 2006; Li, 2002; Morris, Friston & Dolan, 1997) to animal cognition (Mackintosh, 1975).

But what exactly is meant by the term stimulus salience? There is no straightforward answer to this question, as the apprehension of salience is likely to be determined by a variety of factors. First, the inherent features of the stimulus will contribute to its effective salience, for example its physical intensity (what we will term intrinsic salience). More physically intense stimuli will be learned about more easily. A stimulus's physical intensity cannot be modulated, but the perception of its intensity can be. Second, associative learning theories tell us that past experience with the stimulus can also influence the amount of attention that is paid to it (what we will term acquired salience). This affects what is known as its associability – how easily it may be learned about – prototypically demonstrated in latent inhibition (LI). Thus past experience can also modulate the effective salience of a stimulus.

Latent inhibition (LI) refers to the phenomenon of reduced conditioning seen after nonreinforced stimulus preexposure. Animal studies of LI typically employ a between-subject procedure to demonstrate LI. In such procedures, one group of subjects is preexposed (PE) to the to-be-conditioned stimulus (CS), whereas another group of subjects is not preexposed (NPE). The CS is subsequently paired with an unconditioned stimulus (US). LI is measured by the difference in some index of learning (e.g., conditioned fear) of the CS–US association between the PE and NPE groups, and is manifested as poorer learning in the PE group. As detailed in several chapters in the present volume by Weiner, Lubow, Kumari and Ettinger, Escobar and Miller and others, a number of theories have been put forward to explain LI as a form of learned inattention in which subjects learn to ignore or reduce attention to irrelevant stimuli (Lubow,1989). LI has also been characterized as a form of proactive interference in which the inattentional response (acquired as a consequence of a “CS–no-consequence” association) interferes with, or competes for, the expression or retrieval of the conditioned response (resulting from an effective CS–US association) (Miller & Escobar, 2009; Weiner, 1990, 2003, 2009).

Deficits in attention and information processing are central features in schizophrenia, and these deficits may lead to stimulus overload, cognitive fragmentation and thought disorder common in this disorder (Freedman et al., 1991; Perry et al., 1999; Strauss et al., 1993).

Latent inhibition (LI), a phenomenon that reflects an outcome from the processing of irrelevant stimuli, has been of interest to the research community for five decades. And, if anything, its appeal and influence is growing. To mark the fiftieth anniversary of the publication of the first LI experiment, we asked a number of leading scientists to contribute chapters to a volume that would cover the broadest possible range of recent developments in LI research and theory.

Amongst other things, we were interested in showing how a simple behavioral experiment conducted so very many years ago on sheep and goats has led to a burgeoning research enterprise that has enlisted many neuroscience disciplines, including those in physiology, chemistry, pharmacology, and genetics, and has branched out from academic concerns with learning theory to theoretical interests and applications related to schizophrenia. Unfortunately, many people working in research-specific areas find it difficult to keep abreast of the broad cross-disciplinary advances in LI, often directly relevant to their own interests. As an example, there is considerable research on the pharmacological, molecular, and cellular mechanisms underlying LI, and LI is a popular paradigm for studying the neurobiological basis of schizophrenia. However, many of the neuroscientists in this field are unaware of the cognitive/information processing theories underlying the LI effect. The opposite is also true; behavioral/cognitive theorists are often uninformed about advances in the neurophysiology of LI. The present volume provides these researchers with a comprehensive survey of current LI research and theory, from genetics to behavior, thereby strengthening the particularist approach to research as well as fostering an interdisciplinary methodology.

Selective attention is a process which requires that an individual devotes attention to a task-relevant stimulus and processes it, while ignoring a distracting-irrelevant stimulus. The ability to selectively attend and respond effectively to relevant stimuli is essential for adaptive behavior and normal cognitive functioning, since, according to capacity approaches (e.g., Kahneman, 1973; Navon & Gopher, 1979), attention exists in limited amounts and might be allocated to irrelevant as well as relevant information (e.g., Gopher, 1992).

That normal selective attention is essential for adaptive behavior is supported by the literature which shows a relationship between psychopathology and the impairment of selective attentional processes. The most prominent example of such a relationship is the attentional deficit of schizophrenics and “normal” schizotypals. Studies have shown that, as compared to normal individuals, schizophrenics and schizotypals are distracted by irrelevant stimuli (see below). In addition, there is evidence that individuals characterized by high levels of anxiety also are distracted by irrelevant stimuli, although most studies suggest that anxiety leads to an attentional bias only for irrelevant threat-related stimuli.

Surprisingly, despite the similarity of selective attention impairments in schizophrenia/schizotypy and anxiety, there are hardly any studies that have tested whether the anxiety that characterizes schizophrenia/schizotypy accounts for (even only partially) the difficulties schizophrenics/schizotypals have in ignoring irrelevant information.

This chapter will present data about selective attentional dysfunction in groups rated low and high on self-report anxiety and schizotypy scales and in situations that elicit anxiety (stress).

An introduction: nicotine, schizophrenia, and tobacco use

The purpose of this chapter is to review the role of nicotine in the modulation of attention, with particular application to the latent inhibition model of schizophrenia. Nicotine is the primary psychoactive chemical in tobacco. It is addictive and promotes continued tobacco use. According to the Centers for Disease Control and Prevention, the adverse health effects from cigarette smoking account for an estimated 438,000 deaths, or nearly 1 of every 5 deaths, each year in the United States – more than all deaths from human immunodeficiency virus (HIV), illegal drug use, alcohol use, motor vehicle injuries, suicides, and murders combined. The current estimate of cigarette smoking among adults in the United States is 20.8%; however, people with mental illness are much more likely to smoke, and consume a disproportionately large number of cigarettes (Goff, Sullivan, McEvoy et al., 2005; Grant, Hasin, Chou et al., 2004; Lasser, Boyd, Woolhander et al., 2000).

Schizophrenia is a cognitive disorder characterized by hallucinations and disturbances in memory, attention, and executive function. It is also characterized by an extremely high prevalence of smoking. Reliable estimates indicate that approximately 85–90% of schizophrenics smoke cigarettes (Hughes, Hatsukami, Mitchell, & Dahlgren, 1986). The co-morbidity between schizophrenia and nicotine addiction is striking and suggests that there may be common mechanisms in the pathways that lead to these diseases.

This chapter summarizes my work with Jeffrey Gray (1934–2004) on latent inhibition (LI), i.e., the retardation in the generation of the conditioned response (CR) during the pairing of the conditioned stimulus (CS) and the unconditioned stimulus (US) following CS preexposure. From 1996 to 2004, we published a series of articles on (a) a neural network model that describes LI and other classical conditioning paradigms (Schmajuk, Lam, & Gray,1996), (b) the mapping of variables in the model onto brain structures and neurotransmitters (Schmajuk, Cox, & Gray, 2001; Schmajuk, Gray, & Larrauri, 2005), (c) the application of the model to the description of the effect of brain lesions and drug administration on LI and the characterization of the neural activity and neurotransmitter release in different brain areas during LI (Buhusi, Gray, & Schmajuk, 1998; Schmajuk, Buhusi, & Gray, 1998), and (d) how the absence of LI, regarded as a positive symptom of schizophrenia, is caused by hippocampal dysfunction and ameliorated by DA blockers (Schmajuk, Christiansen, & Cox, 2000; Schmajuk, 2001; Schmajuk, 2002).

After 2004, still guided by Jeffrey's suggestions, Jose Larrauri and I (Schmajuk & Larrauri, 2006) successfully applied the model to data that seem to challenge the power of existing models of classical conditioning. Namely, recovery from blocking (the reduced responding to one CS of a conditioned compound when the other CS was separately conditioned first), backward blocking (the reduced responding to one CS of a conditioned compound when the other CS was separately conditioned following compound conditioning), the counteraction between LI and overshadowing (the reduced responding to one or both CSs of a conditioned compound compared with their individual conditioning), and super LI (decreased responding to a preexposed CS after a delay following conditioning,).

Introduction

Schizophrenia is a severe neuropsychiatric disorder of unknown aetiology. The condition is clinically heterogeneous and affects virtually all areas of life, often resulting in disabling cognitive, perceptual, and emotional symptoms. The symptoms of schizophrenia are generally classified as positive (e.g. hallucinations and delusions), negative (e.g. anhedonia, alogia, thought paucity) and cognitive (e.g. thought disorder, bizarre thinking). The disease course is often chronic and the financial cost on the health-care system and society is tremendous, in addition to the personal consequences of the illness for friends and family of sufferers (Mangalore & Knapp,2007; McEvoy, 2007). Tragically, a substantial number of patients commit suicide.

The most successful treatments for schizophrenia that are currently prescribed are pharmacological, and most of these involve blockade of striatal dopamine receptors (Kapur & Remington, 2001). These treatments are successful in reducing the acute symptoms of the condition but provide no cure; accordingly, the need for a better understanding of the aetiology and pathophysiology of schizophrenia and the development of novel treatments is considerable. Drug development draws on many experimental strategies as well as on serendipity (Carpenter & Koenig, 2008; Javitt et al., 2008); translational models such as latent inhibition (LI) play an important role in the strategic effort to develop new treatments (Weiner, 2003). Indeed, a wealth of animal data exist that support the utility of the LI paradigm in studies of antipsychotic treatments (e.g., Weiner, this volume).

The first latent inhibition (LI) paper was published 50 years ago (Lubow & Moore,1959), and the present book marks that anniversary. As such, it offers a convenient time for providing a historical perspective to a phenomenon that was born by accident, barely survived the first several post-parturitional years, and yet developed into a flourishing research enterprise, with activities cutting across such diverse fields as learning theory, schizophrenia, and even creativity. Indeed, in the weekly episodes of “Prison Break” on American TV, the concept of LI has even reached prime-time television.

In spite of the relatively widespread use of the LI paradigm in the laboratory, and, in particular, because of its adoption in research areas that are far removed from its origins, the present editors felt that there was a need to acquaint the larger audience with both the history and recent advances in LI research and theory.

Before describing the serendipitous discovery of LI, this apparently simple, yet ubiquitous, phenomenon requires a definition and a description of its adaptive function. Specifically, LI is a name for the decrease in the ability to acquire or express a new association to a stimulus that has previously received passive, non-reinforced preexposures, as compared to a stimulus that is either novel (not preexposed) or one that has been reinforced or attended. Importantly, LI is not a process. It is an effect that, as will be seen, is in search of a process that generates it.

Background

The analysis of latent inhibition to be developed in this chapter had its origins in the attempt to study a (seemingly) quite different phenomenon – the acquired distinctiveness of cues. James (1890) had suggested that similar cues could be rendered more discriminable by training in which each became linked with a very different associate. The idea was taken up by Lawrence (1949) in his classic experiments with rats, and the 1950s and 1960s saw a rush of experiments (reviewed by Hall,1991) exploring the phenomenon with human subjects. A theoretical account of the effect (based on that of Sutherland & Mackintosh, 1971) was put forward by Mackintosh (1975). He proposed that the representation of any cue has a given level of associability (represented, usually, by the symbol α) but that the value of the parameter α can be changed by experience. When a cue predicts a consequence reliably (or, at least does so more reliably than other cues that might be present) its α value increases; when it predicts a consequence no better than other cues its α value declines. Initial training in which two cues are reliably followed by different consequences will enhance the α value of each of them and subsequent discrimination tasks involving these cues will be facilitated – the acquired distinctiveness effect.

Empirical evidence of the acquired distinctiveness effect came largely from studies of discrimination learning, but Mackintosh's (1975) account implies that the effect should be found in simple conditioning too.

Introduction

To say that latent inhibition (LI) is a well-established phenomenon of animal conditioning is a serious understatement. Since its first demonstration 50 years ago, the finding of a retardation in conditioning following nonreinforced preexposure to a stimulus has been reported countless times across species from sheep to snails, and from goats to goldfish.

Given the ubiquity of LI in animal conditioning, and the suggestion that, under some circumstances at least, a common associative mechanism might also underlie human contingency learning (the acquisition of knowledge about the predictive relationship between a cue and an outcome; Dickinson, Shanks, & Evenden,1984), we might expect a search of the contingency learning literature to reveal many demonstrations of LI in humans. The conclusions to be drawn from such a search depend largely on whether an empirical or theoretical approach to LI is taken.

An empirical approach to LI

If we define LI empirically, as the experimental observation of a retardation in the development of conditioned responding to a stimulus resulting from nonreinforced preexposure to that stimulus, then we must conclude that LI has been demonstrated many times in human contingency learning (e.g. De la Casa, Ruiz, & Lubow, 1993b; Ginton, Urca, & Lubow, 1975; Gray, N.S., et al., 2001; Lubow, Ingberg-Sachs, Zalstein-Orda, & Gewirtz, 1992). For example, during Ginton et al.'s preexposure phase, participants were required to report the number of repetitions of a list of nonsense-syllables (a masking task; see below). The to-be-conditioned stimulus (CS), a burst of noise, was interspersed throughout presentation of this list.

Scores of experiments in the field of animal learning have demonstrated that conditioning to a stimulus depends not merely on its current relationship with a reinforcer, but is affected by animal's past experience with that stimulus. Latent inhibition (LI) is one case of such a biasing effect of past experience: it reflects the proactive interference of nonreinforced stimulus preexposure on the subsequent performance of a learning task involving this stimulus (Lubow,1973, 1989; Lubow, Weiner, & Schnur, 1981).

LI can be demonstrated in a variety of classical and instrumental conditioning procedures, and in many mammalian species, including humans (Lubow, 1973, 1989; Lubow et al., 1981). While a variety of behavioral tasks are used to demonstrate LI in rodents, all of them share a basic procedure. In the first stage, preexposure, animals from each of two groups are placed in an environment that will later serve as the conditioning–test apparatus. Subjects in the “stimulus preexposed” (PE) group are repeatedly exposed to a stimulus (e.g., tone) which is not followed by a significant consequence. Subjects in the “nonpreexposed” (NPE) group spend an equivalent amount of time in the apparatus without receiving the stimulus. When the preexposure stage is completed, either immediately, or a certain time later, all of the subjects enter the conditioning stage, in which the preexposed stimulus is paired with a reinforcer over a number of trials. Performance is assessed by examining some behavioral index of conditioned responding, either during the conditioning stage or in a third, test stage.