Keynote Lecture 4: The Development of Behavioral Tests to Assess the Effects of Cognitive Enhancers D. S. O/ton, L. Wenk Neuromnemonics Laboratory, Department of Psychology, lohns Hopkins University, Baltimore, USA
The search for drugs that enhance cognition requires the development ofbehavioral tests for animals. These tests must be able to identify potentially therapeutic drugs and reject inefTective drugs. A reference compound for cognitive enhancement does not exist, and may not be a reasonable goal. Therefore, a coherent conceptual and experimental framework is needed to organize the various research endeavors in this area. This paper proposes a framework to assist in the development and evaluation of specific behavioral tests. It requires explicit identification ofthe goals of the research, the cognitive process, the neural systems involved in the process, the selectivity and sensitivity of tasks that measure the process, the validity of the behavioral tasks that measure the process, and the validity of the behavioral tasks as a model to predict the efTects ofthe drug in humans.
Introduction
Die Entwicklung von VerhaItenstests zur Bewertung der Wirkungen kognitionsverstärkender Verbindungen
Die Suche kognitionsverstärkenden Arzneimitteln erfordert die Entwicklung von Tierverhaltensversuchen. Diese Versuche müssen es ermöglichen, potentiell therapeutische Arzneimittel festzustellen und unwirksame Präparate zu verwerfen. Es gibt kein Vergleichspräparat hinsichtlich einer Kognitionsverstärkung und dürfte auch keine notwendige Zielsetzung sein. Erforderlich wäre ein kohärenter Rahmen, sowohl konzeptmäßig als auch experimentell, zur Organisierung der verschiedenen Forschungsvorstöße auf diesem Gebiet. Der nachstehende Beitrag schlägt einen Rahmen zur Unterstützung der Entwicklung und Auswertung spezifischer Verhaltenstests vor. Zur Durchführung dieser Tests bedarf es der genauen Formulierung des jeweiligen Forschungszieles, des kognitiven Prozesses, der daran beteiligten neuralen Systeme, der Selektivität und Sensibilität der zur Messung verwandten Vorgänge und der Gültigkeit der Verhaltensaufgaben als Modelle zur Vorhersage der wahrscheinlichen Wirkungen der betreffenden Präparate auf den Menschen.
Although enhanced cognitive function is the goal, cognition cannot be measured directly. Rather, cognitive function must be examined by investigating its various behavioral expressions. Cognitive function in animals must be examined by determining the quality of performance in a variety ofbehavioral tasks.
Animal research offers a number of possibilities for the purpose of elucidating cognitive function and for the study of agents that are capable of enhancing cognition and the neural processes that underlie it. These include the ability to manipulate and analyze the specific neural processes ' There are three questions that must be ansthat underlie cognitive funetion. The goal of research directed towards the discovery of drugs that can enhance cognitive wered prior to beginning behavior studies with animals and function requires that these agents somehow improve these agents that enhance cognition. What component of cognition underlying neural processes. is to be enhanced? What behaviors reflect this component of cognition? How is this enhancement to be measured? What does it mean to enhance cognitive function ? In order to answer this question, we must first underBenefits and Risks of Animal Models to stand cognition. This article will briefly discuss issues relevant Develop Cognitive Enhancers for Humans to studies designed to investigate cognition and agents that enhance cognition. Investigative approaches using animal reResearch with animals is aprerequisite for the search will be examined specifically for their ability to test the development of drugs to help improve cognitive processes in effects of compounds on neural processes that underlie cogni- humans. Animal models offer the advantages inherent in every simplified system: the opportunity to make manipulation. tions and conduct analyses that are not feasible in the primary system of interest (lngram, 1985; Kesner, 1985; O/ton, 1983, Pharmacopsychiat. 23 (1990) 65 - 69 (Supplement) 1984, 1985; Squire, 1985). Cl GeorgThieme VerlagStuttgart . New York
Downloaded by: University of British Columbia. Copyrighted material.
Summary
Pharmacopsychiat. 23 (1990)
For development of a cognitive enhancer, the primary advantage of animal models is the opportunity to study the nervous system. Ultimately, of course any "cognitive" enhancer does not directly enhance cognition. Rather, it affects the nervous system, and the changes in the nervous system are the ones responsible for the altered cognition. Consequently, the ability to measure the state of the nervous system prior to and following treatment, and to manipulate it in systematic ways during the treatment, can obtain infonnation that is critically important in the development of an effective intervention. Animal models, Iike all model systems, also have the risk that the results obtained in the model system do not generalize to the primary system of interest. In such a case, the research might obtain some useful infonnation about the functioning ofthe model, but this infonnation will not be relevant forthe primary goal ofthe endeavor. The absence of a consensus about the drugs can be defined as reference compounds for cognitive enhancement suggests that in spite ofthe advantages of animal models, the risks have not yet been reduced as effectively as they might. Part of the difficulty may arise from the tenn "cognitive", which can be so vague and encompassing that it may include mutually-exclusive endeavors; a drug that is a cognitive enhancer for one investigator may be a control substance for another investigator. If behavioral tasks are to be used to discover potential cognitive enhancers, the cognitive process that is to be enhanced must be c1early identified. A concise definition of "cognitive" has proved elusive, and may be impossible. Consider the logical outcome of c1assifying only some psychological processes as cognitive; other psychological processes must, by definition, be noncognitive. The idea of a noncognitive psychological process doesn't have much intuitive appeal, and if all psychological processes turn out to be cognitive, then the tenn cognitive hasn't provided much infonnation. Indeed, the tenn "cognitive" may produce more problems than benefits. The following discussion introduces an alternative framework that provides greater specificity to identify the process that is to be enhanced. Representations A representation is a fonnal description of a set of elements that can carry out a specific funetion. Every representation has five characteristics that should be specified (Roitblat. 1982, 1987). The discussion here iIIustrates these five characteristics of a representation for recent memory, which is the type of memory that stores newly acquired information (Olton. 1984). The domain of recent memory is the set of daily events or laboratory tasks that require this representation. Different types of problems can require different abilities in order to solve them correctly. Thus, any representation is elicited more by some problems than by others, and the problems that maximally elicit the representation define its domain.
In everyday Iife, recent memory is required to answer questions such as: What did you have for breakfast
D. S. Olton, L. Wenk
today? Who visited you last night? What is the name ofthe person who talked with you a few minutes ago? Where did you park your car today? In the laboratory, recent memory is required to solve delayed conditional discriminations. In this type oftaks, some infonnation is presented at the beginning of each trial, and then removed. Following an interval, two or more alternative responses are made available to the animal, and the response that is correct is conditional upon the infonnation that was presented at the beginning of the trial. In essence, the behavioral taks asks the question at the end of the trial: "What stimulus was presented at the beginning of this trial ?" Because the brain has extensive localization of funetion, some parts are critically involved in recent memory, while others are not. If the goal of the research is to create a compound that enhances recent memory, then that compound should affect the neural systems involved in recent memory, and behavioral tasks to assess the effectiveness of the compound should be those in the domain ofthis representation. The content of the representation is the information from the experience that is maintained in the representation. Most of the infonnation in the environment at any given time has little lasting impact on our behavior, and most experiences can be remembered on the basis ofvery different kinds of infonnation: visual image, verbal label, association with some other event, etc. Because different types ofinfonnation are stored in different parts of the brain, any attempt to create an enhancer for recent memory will be assisted by the content of the representation, which should help identify the part ofthe brain in which function must be enhanced. The code is the relationship ofthe stimuli in the environment and the content of the representation. The code for recent memory is the relationship between stimuli in the environment and the elements that are stored in the representation. The medium of the representation is its physical basis. For the current endeavor, the medium is the entire set of neural changes involving representation. These might be described at many different levels: neuroanatomical structures, neurochemical systems, membrane channels, ion flows. An understanding ofthe neural medium ofthe representation, the ways in which this medium is impaired, and the ways in which it can be improved, will obviously help to identify an appropriate enhancing drug. The dynamics of the representation are the changes it experiences, and the variables that affect those changes. For recent memory, two important dynamics are the complementary processes of forgetting and consolidation. From any given event, some infonnation is lost as the time since the event increases (forgetting), while other infonnation is better preserved (consolidation). Any effort to improve recent memory might focus on either one ofthese dynamics, and seek to inhibit forgetting or improve consolidation. This discussion of representations indicates the type of specificity that is necessary to provide an adequate definition of the type of cognitive process that is to be enhanced. Being specific has obvious advantages. It has defined:
Downloaded by: University of British Columbia. Copyrighted material.
66
the c1inical complaints for which the enhancer is appropriate, the types of behavioral tasks to assess the effectiveness of the enhancer, the forms of information that might be improved, the neural systems that should be influenced, and the specific changes in the memory that might be altered. No longer is the search for an iII-defined substance, but for one with a specific set of attributes. A failure to be specific can lead to unnecessary problems. Consider two different research projects designed to enhance performance in a delayed conditional discrimination, a task that assesses recent memory. One project focuses specifically on a compound that is designed to alterthe dynamics of recent memory that are reflected as an increased rate of forgetting during the delay interval between the presentation ofthe information at the beginning ofthe trial and the opportimity for responding at the end of the trial. Manipulations that specifically affect this rate of forgetting should produce an interaction between the manipulation and the length ofthe delay intervaI. For example, animals in the control group should show some inverse function relating choice accuracy to the length of the delay interval ; the longer the delay interval, the , worse the performance. A drug that specifically affects this rate of forgetting should decrease the slope of this function. The specific form ofthe interaction might vary depending on a number of parameters. For example, the drug might be able to enhance only those memories that have undergone a small amount of forgetting. If such is the case, the interaction should occur only with intermediate delays, and not with the shortest or longest delays. A1tematively, the drug might retard only severe forgetting, in which case the interaction would occur most strongly with the longest delays. Whatever the variations, however, any drug that effects rate of forgetting should produce an interaction with the function relating choice accuracy to the delay interval, and change the slope of this function at some delay intervaI. A second project may focus on some other aspect of the representation, perhaps the content of the information. Changing the elements of the experience that are remembered can have a substantial effect on choice accuracy, but this influence may appear as a main effect rather than an interaction. Thus, choice accuracy may be improved equivalently at all delay intervals, shifting the intercept rather than the slope ofthe function. A failure to be c1ear about the goals ofthese two projects could lead to some unfortunate outcomes. While both might be described as seeking a cognitive enhancer, they differ in the types of behavioral change desired, the underlying neural bases ofthe behavioral change, and the types of drugs c1assified as successful and unsuccessfuI. Unless these differences are explicitly stated and acknowledged, they will inevitably confound any attempt to compare results and evaluate the relative merits ofthese two projects as a means to identify a cognitive enhancer.
Validity The object of a model system is to predict the outcome from procedure used in the primary system of interest. Several types of validity are relevant to evaluate animal
Pharmacopsychiat. 23 (1990)
67
models for the development of cognitive enhancers. Again, the specific examples will refer to recent memory (Olton. 1987). Psychological (cognitive) validity refers to the characteristics of representations that are used to describe the cognitive process being modeled. In recent memory interference (proactive and retroactive) and serial order effects (primary and recency) are typically present, for both humans and rats (Kesner, 1985; Olton, 1982). Operational validity refers to the manipulations and measurements that are made in the experiment. For recent memory, the amount of information to be remembered and the length of time that it is to be remembered are important variables influencing choice accuracy, for both humans and rats. Ethological validity refers to the similarity between the types of demands imposed by laboratory tasks and those normally experienced by animals in their natural habitat. Numerous natural situations impose a demand on recent memory when animals search for food, and considerations of optimal foraging point out many situations in the natural environment that may be formally described as a delayed conditional discrimination (Kamil and Sargent, 1981; Sherry and Schacter, 1987). Neuroanatomical validity refers to the neural systems that underlie performance in the task. In both humans and rats, the hippocampus is a critical element in the neural system necessary for recent memory (Olton, 1984). Neurotransmitter validity refers to the neurotransmitter systems that underlie performance in the task. The cholinergic system is c1early involved in recent memory (Bartus, 1985; Davies, 1985). Cholinergic antagonists impair recent memory, cholinergic agonists may improve it in some circumstances. Developmental validity refers to the changes that take place with early development and with aging. Recent memory takes some time to appear in both young humans and young rats, and it is one ofthe types of memory that is impaired withaging. The more valid a given behavioral task the more Iikely it is to predict the effects of a drug in humans. Thus, for any given task, the validity along each ofthese dimensions should be specified. Sensitivity and Selectivity Every task can be described in terms ofits sensitivity and selectivity to a given manipulation. Sensitivity describes the relationship between performance in the task and the manipulation that alters that performance. In a highly sensitive task, a small difference in the independent variable can produce a large difference in behavior so that the task acts as a magnifier. Given the appropriate experimental parameters so that large differences in performance can be measured accurately, highly sensitive tasks are always desirable.
Downloaded by: University of British Columbia. Copyrighted material.
Keynote Lecture 4: The Developmem 0/Behavioral Tests
D. S. OIton, L. Wenk
Pharmacopsychiat. 23 (1990)
Seleetivity describes the relationship between performance in the task and other manipulations. In a highly seleetive task, large difTerences in other independent variables produce little change in behavior so that the task aets as a filter. The amount of seleetivity desired for a task depends on the goals of the experiment and the payofT matrix for the various outcomes : correet identification of an efTeetive drug (hit), correet identification of an inefTeetive drug (correet rejeetion), incorreet identification of an efTeetive drug (false alarm), incorreet rejeetion of an efTeetive drug (miss).
Stein, 1982), refleeting the plasticity of the nervous system. If so, any changes in neural aetivity induced by the intended cognition enhancer might aetually produce a decrement in performance.
Given the difficulty of identifying efTeetive compounds for cognitive enhancers, and the potential uses of these compounds once identified, the rewards for correet identification of an efTeetive drug are very high, so that relatively nonseleetive tests may be desirable, especially as an initial screen. Subsequently, when deciding whether or not a compound should go to the expense and risks of a clinical trial, more seleetive tests are useful because use of an inefTeetive drug may produce serious harm.
A potential answer to this question can be gleaned by examining the history of drug use in our culture. Initially, drugs were given to people to treat a disease, eradicate an infeetious agent, and produce good health. Later, drugs were introduced to treat impaired mental funetion. These compounds, such as the anti-psychotics, revolutionized the treatment of psychosis. The rationale for the use of these compounds was similar to that of the antibiotics; a sick organ, the brain, was treated pharmacologically in order to produce good health. Next, a new class of drugs were introduced for birth control. These compounds were given to healthy women for the major purpose of preventing pregnancy. For the first time, a powerful chemical was being prescribed for a healthy person simply for social convenience. The introduetion of these agents to the pharmaceutical market was the beginning of a movement that ultimately influenced the acceptance of compounds whose sole purpose is to enhance mood.
Comparison of Different Tasks
Many forms of tests of recent memory are appropriate for the search for cognitive enhancers. The representations for the tasks can be precisely specified and they have substantial validity as outlined above. A detailed review of each type of task that could be potentially used to develop a cognitive enhancer is clearly beyond the scope of this paper. Furthermore, specific comparisons are not meaningful until the dimensions and criteria for judging are explicitly defined. Every game has rules, and until the rules are clear to every player, the results of any given match are uninterpretable. Thus, we encourage interested readers to use this discussion for two endeavors. First, the criteria should be evaluated to determine if they are appropriate. Second, once the appropriate set of criteria are identified, specific tasks should be judged against them. Practical and Etbical Issues Research with animals ofTers us the possibility to manipulate specific neural processes and to investigate the ability of compounds to enhance these specific neural processes. However, the agents that are investigated are intended for use in humans, not animals; the validity of the animal model must always be considered. An aged human under the influence of a neurodegenerative disease is not the same as a normally aged rat or a rat with a specific lesion. These difTerences need to be considered when comparing the efTeets of cognitive enhancers in animal studies and clinical studies on humans. If the drug is to be used to enhance cognitive funetion, then the question arises as to whether it is even possible to enhance cognition. The neural processes that underlie the various aspects of cognitive funetion may already funetion at an optimal level, and this level may not be able to be increased further. An inverted U-shaped dose-response curve probably charaeterizes the efTects of these agents on performance, and the normal individual may funetion near the top of this dose-response curve. Even the brain-damaged individual may undergo considerable recovery of funetion (Finger and
Finally, there are numerous ethical issues that need to be considered when using cognition-enhancing agents. If cognitive performance can be enhanced, should it be enhanced? Under what circumstances is it appropriate to enhance the funetion ofthe human brain?
Cognition-enhancing agents may fit into a number of these categories. They can be given to impaired individuals to alleviate their impairment, and to normal individuals to produce superior performance, or to make ordinary performance easier and more pleasant. Deciding the potential use of the compound is necessary to determine the best type of animal model. Should we use animal models to investigate age-related degenerative diseases, or should we design studies to investigate the efTects of these drugs on normal young healthy animals? The choice of the animal should refleet the intended use of the particular agent, and the associated moral and ethical issues oftheir use in humans should be addressed while the research is being condueted, not after the compounds have been investigated and clinically tested. Acknowledgements We thank A. Du"for typing the manuscript.
References Banus, R. T., R. L. Dean, M. J. Pontecorvo, C. f7icker:The cholinergic hypothesis: A historical overview, current perspeetive, and future direetions. An Integration of Animal and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Gamzu, S. Corkin (Eds.) 444 (1985) 332-358 Davies, P.: A critical review of the role of the cholinergic system in human memory and cognition. An Integration of Animal and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Gamzu, S. Corkin (Eds.) 444 (1985) 212 - 217 Finger. S., D. G. Stein: Brain Damage and Recovery. 1982 Ingram, D. K.: Analysis of age-related impairments in leaming and memory in rodent models. Memory Dysfunetions: An Integration of Animal and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Gamzu, S. Corkin (Eds.) 444 (1985) 312-331
Downloaded by: University of British Columbia. Copyrighted material.
68
Keynote Lecture 4: The Development 0/Behavioral Tests
69
Prof D. S. Olton Neuromnemonics Laboratory Department of Psychology Johns Hopkins University Baltimore, MD 21218 USA
Downloaded by: University of British Columbia. Copyrighted material.
Kamil. A. c.. T. D. Sargent (Eds.): Foraging Behavior: Ecological, Ethological, and Psychological Approaches. Garland STPM Press, NewYork 1981 Kesner. R. P.: Correspondence between humans and animals in coding oftemporal attributes: Role ofhippocampus and prefrontal cortex. An Integration of Anima) and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Gamzu, S. Corkin (Eds.) 444(1985) 122-136 Olton. D. S.:Charaeteristics ofspatial memory. In: Hulse, S. H. et al. (Eds.) Cognitive Aspeets of Anima! Behavior. Lawrence Erlbaum Associates, Hillsdale, NJ(l978)341-373 OIton. D. S.:The use of animal models to evaluate the effects ofneurotoxins on cognitive processes. Neuro. Tox. Tera. 98 (6) 1983 635-640 Olton. D. S.: Animal models of human amnesia. In the Neuropsychology of Memory. Squire, L. R., N. Butters (Eds.) Guilford Press, New York(1984) 367-373 Olton. D. S.: Strategies for the development of animal models of human memory impairments. An Integration of Animal and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Garnzu, S. Corkin (Eds.) 444 (1985) 113-121 Roitblat. H. L.: Introduetion to Comparative Cognition. 1982 Sherry. D. F.• D. L. Schacter:The Evolution ofMultiple Memory Systems.94(1987)439-454 Squire. L. R .• S. Zola-Morgan:Theneuropsychologyofmemory: New links between humans and experimental anima1s. An Integration of Anima! and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Gamzu, S. Corkin (Eds.) 444 (1985) 137-149
Pharmacopsychiat. 23 (1990)
The search for drugs that enhance cognition requires the development ofbehavioral tests for animals. These tests must be able to identify potentially therapeutic drugs and reject inefTective drugs. A reference compound for cognitive enhancement does not exist, and may not be a reasonable goal. Therefore, a coherent conceptual and experimental framework is needed to organize the various research endeavors in this area. This paper proposes a framework to assist in the development and evaluation of specific behavioral tests. It requires explicit identification ofthe goals of the research, the cognitive process, the neural systems involved in the process, the selectivity and sensitivity of tasks that measure the process, the validity of the behavioral tasks that measure the process, and the validity of the behavioral tasks as a model to predict the efTects ofthe drug in humans.
Introduction
Die Entwicklung von VerhaItenstests zur Bewertung der Wirkungen kognitionsverstärkender Verbindungen
Die Suche kognitionsverstärkenden Arzneimitteln erfordert die Entwicklung von Tierverhaltensversuchen. Diese Versuche müssen es ermöglichen, potentiell therapeutische Arzneimittel festzustellen und unwirksame Präparate zu verwerfen. Es gibt kein Vergleichspräparat hinsichtlich einer Kognitionsverstärkung und dürfte auch keine notwendige Zielsetzung sein. Erforderlich wäre ein kohärenter Rahmen, sowohl konzeptmäßig als auch experimentell, zur Organisierung der verschiedenen Forschungsvorstöße auf diesem Gebiet. Der nachstehende Beitrag schlägt einen Rahmen zur Unterstützung der Entwicklung und Auswertung spezifischer Verhaltenstests vor. Zur Durchführung dieser Tests bedarf es der genauen Formulierung des jeweiligen Forschungszieles, des kognitiven Prozesses, der daran beteiligten neuralen Systeme, der Selektivität und Sensibilität der zur Messung verwandten Vorgänge und der Gültigkeit der Verhaltensaufgaben als Modelle zur Vorhersage der wahrscheinlichen Wirkungen der betreffenden Präparate auf den Menschen.
Although enhanced cognitive function is the goal, cognition cannot be measured directly. Rather, cognitive function must be examined by investigating its various behavioral expressions. Cognitive function in animals must be examined by determining the quality of performance in a variety ofbehavioral tasks.
Animal research offers a number of possibilities for the purpose of elucidating cognitive function and for the study of agents that are capable of enhancing cognition and the neural processes that underlie it. These include the ability to manipulate and analyze the specific neural processes ' There are three questions that must be ansthat underlie cognitive funetion. The goal of research directed towards the discovery of drugs that can enhance cognitive wered prior to beginning behavior studies with animals and function requires that these agents somehow improve these agents that enhance cognition. What component of cognition underlying neural processes. is to be enhanced? What behaviors reflect this component of cognition? How is this enhancement to be measured? What does it mean to enhance cognitive function ? In order to answer this question, we must first underBenefits and Risks of Animal Models to stand cognition. This article will briefly discuss issues relevant Develop Cognitive Enhancers for Humans to studies designed to investigate cognition and agents that enhance cognition. Investigative approaches using animal reResearch with animals is aprerequisite for the search will be examined specifically for their ability to test the development of drugs to help improve cognitive processes in effects of compounds on neural processes that underlie cogni- humans. Animal models offer the advantages inherent in every simplified system: the opportunity to make manipulation. tions and conduct analyses that are not feasible in the primary system of interest (lngram, 1985; Kesner, 1985; O/ton, 1983, Pharmacopsychiat. 23 (1990) 65 - 69 (Supplement) 1984, 1985; Squire, 1985). Cl GeorgThieme VerlagStuttgart . New York
Downloaded by: University of British Columbia. Copyrighted material.
Summary
Pharmacopsychiat. 23 (1990)
For development of a cognitive enhancer, the primary advantage of animal models is the opportunity to study the nervous system. Ultimately, of course any "cognitive" enhancer does not directly enhance cognition. Rather, it affects the nervous system, and the changes in the nervous system are the ones responsible for the altered cognition. Consequently, the ability to measure the state of the nervous system prior to and following treatment, and to manipulate it in systematic ways during the treatment, can obtain infonnation that is critically important in the development of an effective intervention. Animal models, Iike all model systems, also have the risk that the results obtained in the model system do not generalize to the primary system of interest. In such a case, the research might obtain some useful infonnation about the functioning ofthe model, but this infonnation will not be relevant forthe primary goal ofthe endeavor. The absence of a consensus about the drugs can be defined as reference compounds for cognitive enhancement suggests that in spite ofthe advantages of animal models, the risks have not yet been reduced as effectively as they might. Part of the difficulty may arise from the tenn "cognitive", which can be so vague and encompassing that it may include mutually-exclusive endeavors; a drug that is a cognitive enhancer for one investigator may be a control substance for another investigator. If behavioral tasks are to be used to discover potential cognitive enhancers, the cognitive process that is to be enhanced must be c1early identified. A concise definition of "cognitive" has proved elusive, and may be impossible. Consider the logical outcome of c1assifying only some psychological processes as cognitive; other psychological processes must, by definition, be noncognitive. The idea of a noncognitive psychological process doesn't have much intuitive appeal, and if all psychological processes turn out to be cognitive, then the tenn cognitive hasn't provided much infonnation. Indeed, the tenn "cognitive" may produce more problems than benefits. The following discussion introduces an alternative framework that provides greater specificity to identify the process that is to be enhanced. Representations A representation is a fonnal description of a set of elements that can carry out a specific funetion. Every representation has five characteristics that should be specified (Roitblat. 1982, 1987). The discussion here iIIustrates these five characteristics of a representation for recent memory, which is the type of memory that stores newly acquired information (Olton. 1984). The domain of recent memory is the set of daily events or laboratory tasks that require this representation. Different types of problems can require different abilities in order to solve them correctly. Thus, any representation is elicited more by some problems than by others, and the problems that maximally elicit the representation define its domain.
In everyday Iife, recent memory is required to answer questions such as: What did you have for breakfast
D. S. Olton, L. Wenk
today? Who visited you last night? What is the name ofthe person who talked with you a few minutes ago? Where did you park your car today? In the laboratory, recent memory is required to solve delayed conditional discriminations. In this type oftaks, some infonnation is presented at the beginning of each trial, and then removed. Following an interval, two or more alternative responses are made available to the animal, and the response that is correct is conditional upon the infonnation that was presented at the beginning of the trial. In essence, the behavioral taks asks the question at the end of the trial: "What stimulus was presented at the beginning of this trial ?" Because the brain has extensive localization of funetion, some parts are critically involved in recent memory, while others are not. If the goal of the research is to create a compound that enhances recent memory, then that compound should affect the neural systems involved in recent memory, and behavioral tasks to assess the effectiveness of the compound should be those in the domain ofthis representation. The content of the representation is the information from the experience that is maintained in the representation. Most of the infonnation in the environment at any given time has little lasting impact on our behavior, and most experiences can be remembered on the basis ofvery different kinds of infonnation: visual image, verbal label, association with some other event, etc. Because different types ofinfonnation are stored in different parts of the brain, any attempt to create an enhancer for recent memory will be assisted by the content of the representation, which should help identify the part ofthe brain in which function must be enhanced. The code is the relationship ofthe stimuli in the environment and the content of the representation. The code for recent memory is the relationship between stimuli in the environment and the elements that are stored in the representation. The medium of the representation is its physical basis. For the current endeavor, the medium is the entire set of neural changes involving representation. These might be described at many different levels: neuroanatomical structures, neurochemical systems, membrane channels, ion flows. An understanding ofthe neural medium ofthe representation, the ways in which this medium is impaired, and the ways in which it can be improved, will obviously help to identify an appropriate enhancing drug. The dynamics of the representation are the changes it experiences, and the variables that affect those changes. For recent memory, two important dynamics are the complementary processes of forgetting and consolidation. From any given event, some infonnation is lost as the time since the event increases (forgetting), while other infonnation is better preserved (consolidation). Any effort to improve recent memory might focus on either one ofthese dynamics, and seek to inhibit forgetting or improve consolidation. This discussion of representations indicates the type of specificity that is necessary to provide an adequate definition of the type of cognitive process that is to be enhanced. Being specific has obvious advantages. It has defined:
Downloaded by: University of British Columbia. Copyrighted material.
66
the c1inical complaints for which the enhancer is appropriate, the types of behavioral tasks to assess the effectiveness of the enhancer, the forms of information that might be improved, the neural systems that should be influenced, and the specific changes in the memory that might be altered. No longer is the search for an iII-defined substance, but for one with a specific set of attributes. A failure to be specific can lead to unnecessary problems. Consider two different research projects designed to enhance performance in a delayed conditional discrimination, a task that assesses recent memory. One project focuses specifically on a compound that is designed to alterthe dynamics of recent memory that are reflected as an increased rate of forgetting during the delay interval between the presentation ofthe information at the beginning ofthe trial and the opportimity for responding at the end of the trial. Manipulations that specifically affect this rate of forgetting should produce an interaction between the manipulation and the length ofthe delay intervaI. For example, animals in the control group should show some inverse function relating choice accuracy to the length of the delay interval ; the longer the delay interval, the , worse the performance. A drug that specifically affects this rate of forgetting should decrease the slope of this function. The specific form ofthe interaction might vary depending on a number of parameters. For example, the drug might be able to enhance only those memories that have undergone a small amount of forgetting. If such is the case, the interaction should occur only with intermediate delays, and not with the shortest or longest delays. A1tematively, the drug might retard only severe forgetting, in which case the interaction would occur most strongly with the longest delays. Whatever the variations, however, any drug that effects rate of forgetting should produce an interaction with the function relating choice accuracy to the delay interval, and change the slope of this function at some delay intervaI. A second project may focus on some other aspect of the representation, perhaps the content of the information. Changing the elements of the experience that are remembered can have a substantial effect on choice accuracy, but this influence may appear as a main effect rather than an interaction. Thus, choice accuracy may be improved equivalently at all delay intervals, shifting the intercept rather than the slope ofthe function. A failure to be c1ear about the goals ofthese two projects could lead to some unfortunate outcomes. While both might be described as seeking a cognitive enhancer, they differ in the types of behavioral change desired, the underlying neural bases ofthe behavioral change, and the types of drugs c1assified as successful and unsuccessfuI. Unless these differences are explicitly stated and acknowledged, they will inevitably confound any attempt to compare results and evaluate the relative merits ofthese two projects as a means to identify a cognitive enhancer.
Validity The object of a model system is to predict the outcome from procedure used in the primary system of interest. Several types of validity are relevant to evaluate animal
Pharmacopsychiat. 23 (1990)
67
models for the development of cognitive enhancers. Again, the specific examples will refer to recent memory (Olton. 1987). Psychological (cognitive) validity refers to the characteristics of representations that are used to describe the cognitive process being modeled. In recent memory interference (proactive and retroactive) and serial order effects (primary and recency) are typically present, for both humans and rats (Kesner, 1985; Olton, 1982). Operational validity refers to the manipulations and measurements that are made in the experiment. For recent memory, the amount of information to be remembered and the length of time that it is to be remembered are important variables influencing choice accuracy, for both humans and rats. Ethological validity refers to the similarity between the types of demands imposed by laboratory tasks and those normally experienced by animals in their natural habitat. Numerous natural situations impose a demand on recent memory when animals search for food, and considerations of optimal foraging point out many situations in the natural environment that may be formally described as a delayed conditional discrimination (Kamil and Sargent, 1981; Sherry and Schacter, 1987). Neuroanatomical validity refers to the neural systems that underlie performance in the task. In both humans and rats, the hippocampus is a critical element in the neural system necessary for recent memory (Olton, 1984). Neurotransmitter validity refers to the neurotransmitter systems that underlie performance in the task. The cholinergic system is c1early involved in recent memory (Bartus, 1985; Davies, 1985). Cholinergic antagonists impair recent memory, cholinergic agonists may improve it in some circumstances. Developmental validity refers to the changes that take place with early development and with aging. Recent memory takes some time to appear in both young humans and young rats, and it is one ofthe types of memory that is impaired withaging. The more valid a given behavioral task the more Iikely it is to predict the effects of a drug in humans. Thus, for any given task, the validity along each ofthese dimensions should be specified. Sensitivity and Selectivity Every task can be described in terms ofits sensitivity and selectivity to a given manipulation. Sensitivity describes the relationship between performance in the task and the manipulation that alters that performance. In a highly sensitive task, a small difference in the independent variable can produce a large difference in behavior so that the task acts as a magnifier. Given the appropriate experimental parameters so that large differences in performance can be measured accurately, highly sensitive tasks are always desirable.
Downloaded by: University of British Columbia. Copyrighted material.
Keynote Lecture 4: The Developmem 0/Behavioral Tests
D. S. OIton, L. Wenk
Pharmacopsychiat. 23 (1990)
Seleetivity describes the relationship between performance in the task and other manipulations. In a highly seleetive task, large difTerences in other independent variables produce little change in behavior so that the task aets as a filter. The amount of seleetivity desired for a task depends on the goals of the experiment and the payofT matrix for the various outcomes : correet identification of an efTeetive drug (hit), correet identification of an inefTeetive drug (correet rejeetion), incorreet identification of an efTeetive drug (false alarm), incorreet rejeetion of an efTeetive drug (miss).
Stein, 1982), refleeting the plasticity of the nervous system. If so, any changes in neural aetivity induced by the intended cognition enhancer might aetually produce a decrement in performance.
Given the difficulty of identifying efTeetive compounds for cognitive enhancers, and the potential uses of these compounds once identified, the rewards for correet identification of an efTeetive drug are very high, so that relatively nonseleetive tests may be desirable, especially as an initial screen. Subsequently, when deciding whether or not a compound should go to the expense and risks of a clinical trial, more seleetive tests are useful because use of an inefTeetive drug may produce serious harm.
A potential answer to this question can be gleaned by examining the history of drug use in our culture. Initially, drugs were given to people to treat a disease, eradicate an infeetious agent, and produce good health. Later, drugs were introduced to treat impaired mental funetion. These compounds, such as the anti-psychotics, revolutionized the treatment of psychosis. The rationale for the use of these compounds was similar to that of the antibiotics; a sick organ, the brain, was treated pharmacologically in order to produce good health. Next, a new class of drugs were introduced for birth control. These compounds were given to healthy women for the major purpose of preventing pregnancy. For the first time, a powerful chemical was being prescribed for a healthy person simply for social convenience. The introduetion of these agents to the pharmaceutical market was the beginning of a movement that ultimately influenced the acceptance of compounds whose sole purpose is to enhance mood.
Comparison of Different Tasks
Many forms of tests of recent memory are appropriate for the search for cognitive enhancers. The representations for the tasks can be precisely specified and they have substantial validity as outlined above. A detailed review of each type of task that could be potentially used to develop a cognitive enhancer is clearly beyond the scope of this paper. Furthermore, specific comparisons are not meaningful until the dimensions and criteria for judging are explicitly defined. Every game has rules, and until the rules are clear to every player, the results of any given match are uninterpretable. Thus, we encourage interested readers to use this discussion for two endeavors. First, the criteria should be evaluated to determine if they are appropriate. Second, once the appropriate set of criteria are identified, specific tasks should be judged against them. Practical and Etbical Issues Research with animals ofTers us the possibility to manipulate specific neural processes and to investigate the ability of compounds to enhance these specific neural processes. However, the agents that are investigated are intended for use in humans, not animals; the validity of the animal model must always be considered. An aged human under the influence of a neurodegenerative disease is not the same as a normally aged rat or a rat with a specific lesion. These difTerences need to be considered when comparing the efTeets of cognitive enhancers in animal studies and clinical studies on humans. If the drug is to be used to enhance cognitive funetion, then the question arises as to whether it is even possible to enhance cognition. The neural processes that underlie the various aspects of cognitive funetion may already funetion at an optimal level, and this level may not be able to be increased further. An inverted U-shaped dose-response curve probably charaeterizes the efTects of these agents on performance, and the normal individual may funetion near the top of this dose-response curve. Even the brain-damaged individual may undergo considerable recovery of funetion (Finger and
Finally, there are numerous ethical issues that need to be considered when using cognition-enhancing agents. If cognitive performance can be enhanced, should it be enhanced? Under what circumstances is it appropriate to enhance the funetion ofthe human brain?
Cognition-enhancing agents may fit into a number of these categories. They can be given to impaired individuals to alleviate their impairment, and to normal individuals to produce superior performance, or to make ordinary performance easier and more pleasant. Deciding the potential use of the compound is necessary to determine the best type of animal model. Should we use animal models to investigate age-related degenerative diseases, or should we design studies to investigate the efTects of these drugs on normal young healthy animals? The choice of the animal should refleet the intended use of the particular agent, and the associated moral and ethical issues oftheir use in humans should be addressed while the research is being condueted, not after the compounds have been investigated and clinically tested. Acknowledgements We thank A. Du"for typing the manuscript.
References Banus, R. T., R. L. Dean, M. J. Pontecorvo, C. f7icker:The cholinergic hypothesis: A historical overview, current perspeetive, and future direetions. An Integration of Animal and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Gamzu, S. Corkin (Eds.) 444 (1985) 332-358 Davies, P.: A critical review of the role of the cholinergic system in human memory and cognition. An Integration of Animal and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Gamzu, S. Corkin (Eds.) 444 (1985) 212 - 217 Finger. S., D. G. Stein: Brain Damage and Recovery. 1982 Ingram, D. K.: Analysis of age-related impairments in leaming and memory in rodent models. Memory Dysfunetions: An Integration of Animal and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Gamzu, S. Corkin (Eds.) 444 (1985) 312-331
Downloaded by: University of British Columbia. Copyrighted material.
68
Keynote Lecture 4: The Development 0/Behavioral Tests
69
Prof D. S. Olton Neuromnemonics Laboratory Department of Psychology Johns Hopkins University Baltimore, MD 21218 USA
Downloaded by: University of British Columbia. Copyrighted material.
Kamil. A. c.. T. D. Sargent (Eds.): Foraging Behavior: Ecological, Ethological, and Psychological Approaches. Garland STPM Press, NewYork 1981 Kesner. R. P.: Correspondence between humans and animals in coding oftemporal attributes: Role ofhippocampus and prefrontal cortex. An Integration of Anima) and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Gamzu, S. Corkin (Eds.) 444(1985) 122-136 Olton. D. S.:Charaeteristics ofspatial memory. In: Hulse, S. H. et al. (Eds.) Cognitive Aspeets of Anima! Behavior. Lawrence Erlbaum Associates, Hillsdale, NJ(l978)341-373 OIton. D. S.:The use of animal models to evaluate the effects ofneurotoxins on cognitive processes. Neuro. Tox. Tera. 98 (6) 1983 635-640 Olton. D. S.: Animal models of human amnesia. In the Neuropsychology of Memory. Squire, L. R., N. Butters (Eds.) Guilford Press, New York(1984) 367-373 Olton. D. S.: Strategies for the development of animal models of human memory impairments. An Integration of Animal and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Garnzu, S. Corkin (Eds.) 444 (1985) 113-121 Roitblat. H. L.: Introduetion to Comparative Cognition. 1982 Sherry. D. F.• D. L. Schacter:The Evolution ofMultiple Memory Systems.94(1987)439-454 Squire. L. R .• S. Zola-Morgan:Theneuropsychologyofmemory: New links between humans and experimental anima1s. An Integration of Anima! and Human Research From Preclinical and Clinical Perspeetives. Olton, D. S., E. Gamzu, S. Corkin (Eds.) 444 (1985) 137-149
Pharmacopsychiat. 23 (1990)
