Methods to Identify and Characterize Developmental … · 2017-10-15 · behavioral tests are easily accomplished. To the contrary, as in any area of science, exper-tise and training

Environmental Health Perspectives • VOLUME 109 | SUPPLEMENT 1 | March 2001 79

The normal structure and function of thenervous system may be altered as a result ofexposure to some xenobiotics before or afterbirth. Alterations in nervous system functionmay be identified in laboratory animals usingneurobehavioral methods. Our understand-ing of neurobehavioral methods is derivedprimarily from an extensive history of use infour related disciplines: experimental psychol-ogy, ethology, biopsychology, and behavioralpharmacology. Because of the vast array oftechnologies and experimental models pro-vided by these histories, the neurobehavioralpanel of the working group convened by theInternational Life Sciences Institute RiskScience Institute decided to limit and focusour discussion to those most relevant andpromising for developmental assessment. Wedivided the available assessment methods intosix categories: sensory function, motor func-tion, cognitive function, social behaviors,autonomic/thermoregulatory processes, andbiologic rhythms. Developmental neuro-toxicity (DNT) data available for the firstthree categories far exceed the data availablefor the latter three; therefore, discussion ofthe first three categories was more extensiveand focused on basic principles that form the

basis for proper use and interpretation. Theconsensus of the neurobehavioral panel wasthat the behavioral test methods used inDNT testing are, for the most part, employedcorrectly in hundreds of different laboratoriesaround the world. However, there are numer-ous examples of the misuse of these methodsand misinterpretation of results derived fromthese methods. Therefore, a major focus ofthe discussion that follows outlines the princi-ples for proper use and interpretation of thesemethods. The latter three categories were dis-cussed less extensively and primarily from thestandpoint of potential usefulness for DNT.Although neurophysiologic techniques alsohave an extensive history of use in both neu-roscience and clinical neurology (1), thesehave not been used on a wide scale for DNTstudies and were outside the scope of thecurrent discussion.

Common Issues

Many issues are common to most, if not all,methods of behavioral testing for DNT. Theimportance of most of these issues has beenrecognized, with attempts to ensure thatmethodology addresses them appropriately.Discussions in this section provide additional

guidance on consideration of these issues inthe use of behavioral test methods in DNTtesting.

Behavioral tests vary along many dimen-sions of desirable properties, including theamount of available validation data, speed oftesting, breadth and/or specificity of testresults, availability of equipment and person-nel to conduct the test, and extrapolation ofresults among species. Thus, the most desir-able properties depend upon the experimentalcontext (i.e., what is the question beingasked? What other end points are available toaddress the issue?). The latter question isextremely important and often overlookedbecause most tests are part of a battery. If lit-tle is known about a test substance and theinvestigator is screening for an effect, themost desirable properties of a test battery maybe a wide breadth of function(s) tested, rela-tively short testing period, low cost, and avail-ability of personnel. In contrast, specificityand sensitivity of effect are more desirable testattributes if the chemical is known to produceweakness, for example, and a second tier testis used to characterize the effect and to deter-mine a no-observed adverse effect level, i.e.,distinguish between diminished ability toexert high forces versus ataxia. In the lattercase, cost of the test and wide availability ofpersonnel may be less important.

Animal Model The choice of animal models in developmentalneurotoxicology studies is influenced by a

Alterations in nervous system function after exposure to a developmental neurotoxicant may beidentified and characterized using neurobehavioral methods. A number of methods can evaluatealterations in sensory, motor, and cognitive functions in laboratory animals exposed to toxicantsduring nervous system development. Fundamental issues underlying proper use and interpretationof these methods include a) consideration of the scientific goal in experimental design, b) selectionof an appropriate animal model, c) expertise of the investigator, d ) adequate statistical analysis, ande) proper data interpretation. Strengths and weaknesses of the assessment methods includesensitivity, selectivity, practicality, and variability. Research could improve current behavioralmethods by providing a better understanding of the relationship between alterations in motorfunction and changes in the underlying structure of these systems. Research is also needed todevelop simple and sensitive assays for use in screening assessments of sensory and cognitivefunction. Assessment methods are being developed to examine other nervous system functions,including social behavior, autonomic processes, and biologic rhythms. Social behaviors are modifiedby many classes of developmental neurotoxicants and hormonally active compounds that may acteither through neuroendocrine mechanisms or by directly influencing brain morphology orneurochemistry. Autonomic and thermoregulatory functions have been the province ofphysiologists and neurobiologists rather than toxicologists, but this may change as developmentalneurotoxicology progresses and toxicologists apply techniques developed by other disciplines toexamine changes in function after toxicant exposure. Key words: behavioral testing, cognitivefunction, developmental neurotoxicity, motor activity, sensory function. — Environ Health Perspect109 (suppl 1):79–91 (2001).http://ehpnet1.niehs.nih.gov/docs/2001/suppl-1/79-91cory-slechta/abstract.html

Address correspondence to B. Mileson, ARCADISGeraghty & Miller, 1131 Benfield Blvd., Millersville,MD 21108 USA. Telephone: (410) 987-0032. Fax: (410)987-4392. E-mail: [email protected]

This project was supported by a cooperative agree-ment between the International Life Sciences InstituteRisk Science Institute and the U.S. EnvironmentalProtection Agency Office of Pesticide Programs, bythe National Institute for Environmental HealthSciences, and by the International Life SciencesInstitute Research Foundation.

The opinions expressed herein are those of the indi-vidual authors and do not necessarily reflect the viewsof their respective organizations or the sponsoringinstitutions. Mention of trade names or commercialproducts does not constitute endorsement or recom-mendation for use.

Received 5 September 2000; accepted 13December 2000.

Methods to Identify and Characterize Developmental Neurotoxicity for HumanHealth Risk Assessment. I: Behavioral Effects

Deborah A. Cory-Slechta,1 Kevin M. Crofton,2 Jeffery A. Foran,3 Joseph F. Ross,4 Larry P. Sheets,5 Bernard Weiss,1 andBeth Mileson6

1Department of Environmental Medicine, University of Rochester Medical School, Rochester, New York, USA; 2U.S. Environmental ProtectionAgency, National Health and Environmental Effects Research Laboratory, Research Triangle Park, North Carolina, USA; 3Citizens for a BetterEnvironment, Milwaukee, Wisconsin, USA; 4The Procter & Gamble Company, Ross, Ohio, USA; 5Toxicology Department, Bayer Corporation,Stilwell, Kansas, USA; 6ARCADIS Geraghty & Miller, Millersville, Maryland, USA

109S1.Part 2 02/21/01 2:16 PM Page 79 (Black plate)

Cory-Slechta et al.

80 VOLUME 109 | SUPPLEMENT 1 | March 2001 • Environmental Health Perspectives

number of factors. Each of these factors mayhave more or less weight depending upon theultimate scientific goal. Choice of the animalmodel should be determined primarily by thehypothesis being tested. The species andstrain of animal should be appropriate for thetarget system being tested or modeled. Forexample, albino strains of laboratory animalsare not appropriate when modeling effects ofchemicals on human vision, as albino rodentstypically have extremely poor vision.

In studies conducted for risk assessmentpurposes, extrapolation of animal data tohumans may lead to a different choice. Thechicken is the animal model of choice whentesting for organophosphate-induced delayedneuropathy, primarily because of the predic-tive power of the resulting data (2).Economics may also factor into a decision ofwhich animal model to use. If screeningunknown agents is the major driving need,lower-cost assays may allow one to evaluatemany more compounds. Screening chemicalsin a common rodent species such as themouse or rat is more cost effective than use ofnonhuman primates. If economics are a decid-ing factor, one needs to be convinced that theresulting data will still be meaningful. Age canbe a crucial factor in determining the correctanimal model. For example, testing a hypoth-esis concerning the role of exposure to pesti-cides as a risk factor for Parkinson’s diseasewould necessitate the need to employ lifetimestudies (or other models appropriate for test-ing mechanisms of aging). If enough informa-tion is available on the mechanism of action ofa class of chemicals, transgenic or congenicanimal models may be useful.

Although not a prerequisite for choosingan animal model, adequate background infor-mation on the normal anatomy, physiology,and behavior of the test species can be usefulin interpretation of toxicant-induced changes.Historical control data are also useful in thisregard. Last, the generality of the testmethod, or concurrent validity, should beestablished by determining whether the effectcorrelates with other indices of toxicity. Forexample, do effects observed in a behavioraltest of olfactory function correlate with theunderlying pathologic damage to the olfac-tory epithelium in the nasal cavity and/orolfactory bulbs in the central nervous system?

Age relevance. Design of behavioral exper-iments and hypotheses should consider theage relevance of the animal model. The onsetand maturation of most behaviors are neces-sarily linked to the age of the animal.Although not always critically evaluated, agerelevance of the test procedure can be veryimportant in determining the validity of a testmethod for use in DNT studies. Use of pro-cedures validated in adult models may not beappropriate for young animals.

Resource Demands Regulatory testing requires that many behav-ioral techniques be able to test large numbersof animals (e.g., 10–20 litters per treatmentgroup, with four to five treatment groups) ina relatively rapid fashion. Procedures that donot lend themselves to rapid testing can leadto an inability to test enough subjects at anappropriate age. It is important to minimizepersonnel costs directly related to time andeffort spent testing animals; however, costshould not be an excuse for lack of testing.Instead, excessive costs associated with a spe-cific test method should be a clarion call fordevelopment of more cost-effective tests.

Many companies now make availablecomplete, turn-key, computer-based systemsto assess specific behavioral functions (e.g.,water mazes for measurement of learning/memory, shuttle-box avoidance for memory,locomotor activity chambers, startle testingequipment). In many cases, the software andhardware components of these systems tend tobe simplistic and inflexible. Several issuesshould be considered before such equipmentis acquired. For example, a misperceptionabout the real costs of carrying out tests ofcognitive functions often leads investigators topurchase less expensive equipment in anattempt to economize. The less automated thedevice, the more time required of the investi-gator/staff to conduct the behavioral tests, e.g.,putting animals into start boxes, measuringtimes, errors or other dependent variables, andhousing animals during intertrial intervals.Thus, although single-use equipment maycost less to acquire originally, the economicresources to use the equipment may actuallybe far greater than starting with automated,flexible equipment. In addition, the use ofhardware/software dedicated to measurementof a specific behavioral function does notguarantee that it measures that behavioralfunction selectively, and means that additionalhardware/software will have to be acquired forevery other behavioral function the investiga-tor ultimately wishes to measure. In contrast,operant chambers, which may have a greaterup-front cost, may at the same time providegreater automation and flexibility of use.

Expertise and TrainingThere often appears to be a misperceptionthat implementation and interpretation ofbehavioral tests are easily accomplished. Tothe contrary, as in any area of science, exper-tise and training in behavioral sciences arecritical for both. The absence of such trainingoften results in a lack of understanding of thevariables that may influence a behavior beingmeasured and failure to adequately controlfor the impact of these variables on behavior.Additionally, absence of appropriate trainingand expertise in behavioral methods often

leads to inappropriate interpretation ofoutcome measures in cognitive tests. This hasbecome apparent in the appearance in the lit-erature of studies examining cognitivefunction in genetically engineered mice. Inthis case, state-of-the-art molecular biologicapproaches may be used in conjunction withmisuse and misinterpretation of simple testsof learning or memory. Such occurrencesreflect the lack of expertise and training of theinvestigators in this dimension of their experi-ments and the absence of expertise on theadvisory boards of these journals.

Many people involved in assessing behav-ioral and/or neurologic function in toxicitystudies have some training in experimentalpsychology or psychopharmacology as well asstatistics. This training is needed to ensureproper study design, data collection, and dataanalysis. Behavioral tests of sensory endpoints are sensitive to a wide variety of envi-ronmental variables (e.g., ambient noise, han-dling history, time of day) that may notappear important to untrained investigators.

Many of the more sophisticated and sensi-tive cognitive measures may require taking thesubjects through a series of training programs,in which expertise is again critical for preciselymolding behavior to that required for the finalstage of assessment. Likewise, the conduct ofspecial sensory tests (e.g., reflex audiometry)requires adequate training in sensory psy-chophysics, statistics, and the principles ofbehavioral testing. However, there is no sys-tematic training or certification program com-parable to those available for personnelinvolved in recording and interpreting otherimportant toxicity end points using anatomicor clinical pathologic techniques (3). Thus,there is considerable variability in the expertiseof individuals generating and interpretingbehavioral and/or neurologic data in acade-mic, industrial, or contract laboratories. Theinherent problems associated with the lack ofprofessional standards are compounded by thewide availability of off-the-shelf equipmentthat may appear to obviate the need for exper-tise. It must be stressed that behavioral testingdevices are tools that depend on the expertiseof the user. Proper selection and use of thetools require expertise. Moreover, the design,conduct, and interpretation of studies includ-ing behavioral and/or neurologic end pointsrequire personnel with relevant training andexperience. For example, personnel whoexamine neurologic end points includingreflex and reaction, spontaneous movementabnormalities, and open-field changes in gaitand posture should have adequate training inthe conduct of these tests and use of appropri-ate terminology (4). In addition, effortsshould be made to ensure consistency amongobservers, and interobserver reliability shouldbe reported.


Behavioral testing in developmental neurotoxicity

Statistics A number of statistical issues should beconsidered in developmental neurotoxicologystudies. The first issue concerns how best tocontrol for litter effects. Littermates may beassigned to more than one task, individualanimals may be repeatedly tested on a task, ordifferent littermates may be tested on differ-ent tasks. The statistical implications differfor these testing strategies. Because mostdevelopmental studies use the litter as theunit of measure, repeated sampling from thatlitter (using more than one animal from eachlitter) poses unique statistical problems.Simply put, use of more than one animal perlitter inflates the number of subjects pergroup and can increase type I error (i.e., falsepositives). There are a number of goodreviews of this subject that should be con-sulted prior to designing studies (5–8).

Repeated measures are also a commonfactor in DNT testing. Current U.S.Environmental Protection Agency guidelines(9) recommend motor activity testing in thesame animal at least 3 times during thepreweaning period. When analysis of variance(ANOVA)-based statistics are used, repeated-measures data must be treated as such in themodel. Alternative models such as regressiontechniques also can be used (10). Finally, theassignment of litter members to one or moretasks must be carefully considered. Litter rep-resentatives can be selected and used for alltests, or separate animals can be assigned toeach test. In the former case, behavioral his-tory is a serious confounder. For example, thesame animals should not be used in learningand memory tests at both younger and olderages.

Variability is an inherent property of alltest methods and population samples. Thegreater the variability among animals, thelarger the number required to detect an effect.A variety of techniques can be used to deter-mine, a priori, the power of a test methodbased on historical control data, and thus pre-dict the number of animals needed to detectan effect of a specific size (11,12). Use ofthese statistical procedures is strongly advised.

In the case of negative findings, especiallyin regulatory testing, documentation of histor-ical control data and positive control data isimportant. These data are necessary to ade-quately document the power of the test andthus assure a low incidence of type II error.Historical databases are encouraged to bemade available or published whenever possible[for example, see Crofton et al. (13)].Laboratories employing test methods for thefirst time are encouraged to compare theirdata variability with published reports and/orhistorical control data [for example, see Wiseet al. (14)]. Test methods that result inextremely high variability compared to that in

other laboratories should be inspected todetermine the source of the variability or notbe used. Positive control data, although not asimportant, are needed to establish changes inthe end point related to dose response when-ever possible. Dose–response data (and notsingle-dose studies) should be stressed, as thesedata are needed to adequately document theability to detect different magnitudes of effect.Dose–response data are also needed to deter-mine the linearity or nonlinearity of the effect.

System-Specific Issues

Although some issues are common to most ifnot all DNT behavioral testing methods,many methods and their desirable propertiesare unique to specific functions. Further,some functions or behavioral parameters havenot been the focus of DNT testing; thus,methods are only poorly developed or remainundeveloped. This section presents desirableproperties of methods for DNT testing incommon behavioral functions (e.g., sensory,motor, cognitive). It also presents a discussionof social behavior, and the issues to be consid-ered as DNT testing methods are developedfor this area.

Sensory FunctionSensation plays a crucial role in the ability ofthe organism to interact with its environ-ment. Loss of some aspect of sensory functionis one of the most common occupationalinjuries (15,16), and approximately 44% ofall neurotoxic chemicals are reported to affectsensory function adversely (15,17,18).Although the exact magnitude of the problemmay be debatable, there are two major rea-sons that sensory systems should be evaluatedin the safety assessment of potential neurotox-icants. First, most sensory organs are not pro-tected by the blood–brain barrier and mayexperience greater exposure than other partsof the nervous system. This is especially truefor the olfactory system. Second, properinterpretation of the results from other typesof behavioral function (e.g., cognitive testing)requires consideration of alterations in sen-sory system processing as a confounder.

Assessing sensory function in animals hasbeen a significant research tradition in com-parative psychology, and many of the meth-ods developed in that field have been appliedin neurotoxicology. These methods varyfrom relatively simple and subjective sensoryreflex tests, such as elicitation of the pinnareflex and pupil constriction, to more com-plicated operant-discrimination paradigmsand evoked potential procedures. The majorsensory systems of concern in toxicologyinclude visual, auditory, olfactory, noci-ception (pain and other noxious stimuli),somatosensory, and vestibular. The methodsused to study these different systems, by

necessity, will be somewhat different.However, the characteristics of the propertiesof the test methods that should be underexperimental control will be very similar [forreview, see Maurissen (19)].

Animal model. Species and strain of thetest animal are important in tests of sensorysystems. Selection of the animal model shouldbe determined primarily by the hypothesis orspecific aims. First and foremost should be theuse of a species/strain appropriate for the sen-sory system being tested or modeled. Forexample, use of some strains of mice (e.g.,C57Bl/6J) may be a poor choice for auditorystudies because of the early-onset presbycusisfound in this strain (20). However, use of therat as an animal model of ototoxicity has beenhighly successful in modeling the adverseeffects of xenobiotics in humans (21–23).Stebbins and colleagues, in a series of now-classic studies in this area, used monkeys andguinea pigs to model the dose response andtime course of the ototoxicity of aminoglyco-side antibiotics (24–26). An excellent exampleof the simultaneous assessment of multiplesensory systems comes from the work of Pryoret al. (27). These authors used a conditionedavoidance procedure to assess the effects of awide variety of chemicals on both auditoryand visual system functions in rats.

An important issue in developmentaltoxicity testing is that the animal modelselected must also be age relevant. For exam-ple, stimulus parameters (e.g., nociceptivestimuli) using a conditioned lick-suppressionparadigm for adult animals may be inappro-priate in much younger animals. Procedureswith long acquisition times (e.g., operant)may not be able to target rapidly maturingsensory systems. The work of Merigan andcolleagues (28), which characterized theadverse effects of acrylamide on visual func-tion in monkeys, used operant methods thatwould be unsuitable for testing the ontogenyof visual function in rats because of extensivetraining demands. Lastly, stimuli may not beperceived, or may be perceived as less intense,in animals with immature sensory systems.For example, sonalerts and click stimuli arecommonly used auditory stimuli in adult test-ing. However, these generate low-frequencystimuli that would be inappropriate inpreweanling rats or mice, as sensation of lowfrequencies is the last to develop in most altri-cial rodents (29,30). An appropriate under-standing of the normal ontogeny of thesensory system being assessed is necessary toensure a good match between the age of theanimal and the test procedure.

Any discussion of animal models shouldalso include the issue of sensitivity. There arenumerous sensory techniques that are simpleand inexpensive, but a) may lack precision inestimating psychophysical thresholds, b) may



Cory-Slechta et al.


be subject to experimenter bias, c) are difficultto automate, or d ) are generally result in largeinterlaboratory variability (19). Reflexivemovements (e.g., pinna reflex, corneal reflex)are examples of simple and inexpensive endpoints to test. These tests are fast, easy, andinexpensive. However, they are thought to berelatively insensitive to toxicant-inducedalterations in sensory function (31,32).Conversely, operant and conditioned discrim-ination procedures represent some of themore sensitive and specific methods used forassessing sensory system dysfunctions. Suchtests have been used to characterize the effectsof a wide variety of agents that disrupt varioussensory functions. Examples include auditorydeficits produced by aminoglycoside antibi-otics (25), visual deficits produced by methylmercury (33), ammonia-induced disruptionof olfaction (34), and somatosensory dysfunc-tion produced by acrylamide (35).Performance of these tests generally requiresextensive training of the subject. These testsare time consuming and generally are limitedto small numbers of subjects. Collection ofdata required to characterize the audiogramfor the rat using a conditioned lick-suppression paradigm required over 150 daysof testing (36). Reflex modification audiome-try has been used as a fairly rapid and sensi-tive test of auditory function [for example, seeCrofton et al. (37) and Young and Fetcher(38)]. This procedure also works in humans[for review, see Ison (39)]. However, this pro-cedure requires equipment not available as aturn-key system from commercial suppliers.The trade-off between economy and sensitiv-ity may be difficult to balance for somesensory function testing.

Stimulus parameters. The proper genera-tion and use of stimuli are crucial for testingthe effects of xenobiotics on sensory systemfunction. A number of stimulus properties areshared by all sensory systems, includingintensity, frequency, duration, and locationin space. When using sensory methods in tox-icity testing, it is extremely important thatnormative data be developed for a methodthat demonstrates the ability of the testmethod to detect and characterize the effectsof varying the magnitude of such variables.For example, in tests of visual function,response magnitude should be directly relatedto the intensity of the visual stimulus (40).Auditory testing thresholds should vary acrossthe frequency domain (41). Stimulus fre-quency should also be appropriate to the testspecies. For example, use of a 0.5-kHz tone asthe conditioning stimulus in rats (42) is notoptimal, as rats do not hear this frequencyvery well (36). An 8-kHz stimulus would bemore appropriate for use in rats because thethreshold for this frequency is approximately45 dB lower than a 0.5-kHz stimulus.

Stimulus amplitude may be too high to detectsmall changes in thresholds. For example,routine DNT testing of auditory functionusing startle habituation uses a high-decibelstimulus (e.g., 110 dB sound pressure level).A false-negative finding would result shouldanimals demonstrate only a small increase inthreshold for that stimulus. Recent work withexposure to polychlorinated biphenyls duringdevelopment has found only small changes(i.e., 15–25 dB) in low-frequency thresholds(37). These changes would not be detected intests using high-decibel stimuli.

Statistics. Variability can arise from differ-ent sources in tests of sensory function. It isinherent in any test species (43) and is aproperty of the test method itself or in itsapplication. Any potentially confoundingvariable not controlled will lead to increasedvariance. The rapid development in younganimals should preclude averaging data acrossages, as the response could change drasticallyin a very short time. In DNT testing, averag-ing startle response data in animals testedover a few days of age in adults will not neces-sarily add to the group variance. However, inyoung animals, averaging response data overseveral days could easily increase variance dueto increases in body weight and increases insensitivity of the developing auditory system.

All tests of sensory function should beable to generate data that have adequate sta-tistical power to detect biologically relevantchanges in behavior. For example, a 15-dBchange in auditory thresholds is generallyregarded as adverse (44). Therefore, anymethod used should be capable of detectingstatistical differences between group meansthat differ by 15 dB or more. Historical con-trol data and positive control data should beavailable to document this property ade-quately. Positive control data should establishchanges in the end point related to the doseresponse whenever possible.

Analyses and interpretation. Interpretationof the results of sensory assessment testsshould follow the classic rules of behavioralscience. First, sensory testing never yields adirect measurement of sensation. Instead, oneinfers a change in sensory function based onthe observed change in the motor responsebeing evaluated. For example, increasedlatency to paw lick in the hot-plate test is abehavioral measure that is interpreted, afterruling out other causes, as an increase in anociceptive threshold (45). Motor impairmentdue to muscle fiber degeneration may alsolead to increases in latency that are not causedby sensory mechanisms. Administration of ahigh dose of a sedative, with subsequentdecrease in the amplitude of the acoustic star-tle response, does not mean that the sedativeinduces an auditory dysfunction. Instead,decreases in motor capability are more likely

the culprit for this effect [see Maurissen (19)for a review of the use and misuse ofpsychophysic methods].

Interpretation of any treatment-relatedchange should be done in concert with anunderstanding of the ability of the testmethod to detect changes in the response.Data from positive control studies can yieldmuch information on the sensitivity of a testmethod, as the method is employed in thelaboratory generating the data. Data demon-strating experimental responsiveness tochanges in stimulus parameters are also neces-sary. A method that fails to detect differencesin stimulus strength in control animals isbeing either inappropriately employed orinappropriately interpreted. Historical controldata are also invaluable in this regard.Excessive variability in control group meansover time may be indicative of a lack ofproper experimental control of the behaviorbeing assessed. Alternatively, individual dif-ferences due to a bimodal effect in a popula-tion are possible. Reliability of the testmethod will, of course, depend on a findingof low variability in control values over time,as well as replication in the effects of positivecontrol agents.

Research needs. Included in the researchneeded to advance our ability to determinesensory system toxicity is a better understand-ing of the relationship between the simplesensory system tests currently used in DNTstudies (i.e., simplex reflex tests, startle habit-uation) and any underlying changes in thestructure and/or function of the sensory sys-tem. A commonly held belief is that simplereflex tests are not as sensitive as tests of sen-sory thresholds or sensory signal processing(32). Very few studies have tested thishypothesis (31). In addition, current DNTtest batteries do not routinely assess the age ofonset of sensory function. Without this typeof testing, delays in development of sensorysystems will go undetected routinely. Thequestion of whether these delays lead to long-term adverse effects in the organism is alsocurrently unanswered. There is also a need toidentify and characterize the impact of vari-ous potential confounds on the ontogeny andfunction of sensory systems. What effect dochanges in maternal nutritional status, whichmay be due to chemical exposure, have onmeasurement of function in offspring? Wheredo screening methods fit into a tiered testingscheme? This is an especially vexing problemfor sensory testing, as the sensitivity of themore rapid test currently used in tier 1 testingis unknown. Current sensory function testingusually evaluates treatment-related effectssolely by measuring the behavioral responseto amplitude changes in the stimuli. Manytypes of stimulus processing are not normallyassessed (e.g., frequency, duration, and spatial



information). These stimulus propertiesreflect different degrees of sensitivity to theeffects of xenobiotics. Boyes and colleagues(46), using evoked potential, demonstrateddeficits in spatial contrast sensitivity with noapparent change in the amplitude of flash-evoked potentials. Last, behavioral toxicolo-gists need to keep abreast of recent advancesin neurobiology and genomics that may allowan increased understanding of the physiologicand structural bases for sensory function.This information will be crucial in updatingmethods in the future.

Motor Function In toxicity studies the goal of motor functiontesting is to detect and/or characterize motordysfunction. Behavioral tests of motor dys-function in animals are differentiated intotwo types: those that detect spontaneousmovement disorders such as changes in gait,tremors, and myoclonus; and those thatdetect changes in induced movement such asreflexes, reactions, and movements underoperant control. Some end points arerecorded subjectively, using categorical(present/absent) or ordinal (e.g., absent, min-imal, moderate, severe) scales. For these endpoints, the data are based upon the judgmentof the tester, much as a veterinary neurologistevaluates a patient. Quantitative procedurescan be used to measure qualities such as fore-limb or hindlimb grip strength. In this case, atransducer detects the response of the subject,and a value is recorded.

A review of all the behavioral tests ofmotor function is beyond the scope of thissection, which will be limited to those cate-gories of tests most commonly used to evalu-ate motor function in experimental andregulated DNT studies. These categoriesinclude observation of locomotion, measure-ment of locomotor activity, and tests ofreflexes and reactions.

Observation. Observation of locomotionis used primarily to detect changes in postureand gait (e.g., ataxia, low carriage) and spon-taneous movement abnormalities (e.g., stereo-typy, myoclonus, tremors). Observationaltechniques are fast and inexpensive andenable detection and characterization of awide variety of functional changes (47). Thecorrect application of observational tech-niques requires a level of training and experi-ence sometimes not fully appreciated (3), andappropriate use of terminology is necessary toderive maximum usefulness of this approach(4). The sensitivity of qualitative observationsis not clear but is generally thought to be lessthan that of quantitative procedures.

Motor activity measurement. The termmotor activity refers to a wide variety of teststhat measure different aspects of behavior,for example, exploration, navigation, and

emotionality (48,49). At least three technicalaspects must be considered to appreciate thesimilarities and differences among test sys-tems for measuring locomotor activity: a) thesize (i.e., small or large), shape (e.g., square,round, or figure eight), and illumination(i.e., dark, dim, bright); b) the transducer ordetector device (e.g., photocells or videocamera); and c) the statistical analysis appliedto the raw data. The data analyses includemacroanalysis (e.g., path length, percent ofmovement on interior or exterior of environ-ment), microanalysis [e.g., quantification ofindividual behaviors such as turning, rearing,sniffing; (50)], and characterization of pathshape [e.g., predictable vs unpredictable(51)]. The test animal may be placed in themiddle of a symmetric device and its sponta-neous behavior measured, or the environ-ment may be enhanced asymmetrically withobjects that present distinctive visual or tac-tile stimuli. In the latter case, the orientationof the subject or attention to the stimuli maybe evaluated (52).

In DNT studies one common approach isto measure activity using photocells in a smalland unenhanced chamber. The behavior eval-uated by a particular photocell or combina-tion of photocells may vary by location in thechamber. For example, in some devices alower row measures horizontal movement(i.e., ambulation), and an upper row measuresvertical movements (i.e., rearing). It is essen-tial that the tester understand the relationshipbetween photocell location and behavioralspecificity before trying to interpret the resultsobtained by automated equipment.

Level of activity, reported as number ofphotocell counts, is the end point most com-monly reported, analyzed, and interpreted.The pattern of movements both within ses-sion and between sessions may be character-ized. Patterns observed within a session mayreflect habituation, and patterns observedbetween sessions may reflect evolution ofactivity over time, for example, during days13–21 of postnatal development. Locomotoractivity results sometimes have high variabil-ity, particularly in developmental studieswhen animals mature at different rates. Theinterpretation of motor activity values is gen-erally more controversial than observations ofmovement disorders (53–55). The algorithmfor relating changes in motor activity to neu-ropathologic end points is not as clear as it isfor movement disorders.

Reflexes and postural reactions. Tests ofelicited motor function most often includereflex and reaction tests, which can be meas-ured qualitatively or quantitatively. A reflexis an involuntary and relatively stereotypedresponse to a specific sensory stimulus. Thelocation and amplitude of the responsedepends on both the location and strength of

the stimulus. For spinal reflexes (e.g., flexorreflex), the sensory stimuli arise from recep-tors in muscles, joints, and skin, and theneural circuitry responsible for the motorresponse is contained entirely in the spinalcord. Homologous cranial nerve reflexes arecontained within the brain. Although theneuronal circuits that mediate reflexes aresimple, the brain frequently coordinates theaction of several reflex circuits to generatemore complex behaviors that clinical neurol-ogists term reactions (e.g., placing reaction).Reflex and reaction tests are generally quickand easy to perform and require modesttraining and expertise. Procedures used toquantify some of the reflexes and reactionsinclude forelimb/hindlimb grasp, auditorystartle, and extensor thrust (56,57).

Postural reactions are complex responsesthat maintain the normal, upright position ofan animal under conditions of shifting loads.If the weight of an animal is shifted from oneside to the other, from front to rear, or fromrear to front, the increased load on the sup-porting limb or limbs requires increased tonein the extensor muscles to keep the limb fromcollapsing. Part of the alteration in tone isaccomplished through spinal reflexes, but forthe changes to be smooth and coordinated,the sensory and motor systems of the brainmust be involved.

Abnormalities of complex postural reac-tions (e.g., hopping) do not provide asanatomically precise information about neu-rologic abnormalities as do reflex tests, whichare more limited in scope. However, theintense demands on functional performancerequired by tests of postural reactions mayreveal deficits in neurologic components thatare not detected simply by observing gait.The basis for the deficits may then be clarifiedby further testing of individual reflexes or byelectrodiagnostic testing.

Two popular procedures for evaluatingmotor function—the hindlimb splay androtorod tests—require special discussion.The hindlimb splay test is conducted byholding a rat horizontally several centimetersabove a table surface, then measuring theinterpaw distance of the hind limbs afterdropping the rat to the table surface. Thistest is popular because of its sensitivity indetecting acrylamide neurotoxicity (58).However, despite its popularity and its inclu-sion in the U.S. EPA neurotoxicity testingguidelines (59), little is known about thistest. The anatomic basis for the test isunknown, there is no obvious analog used byveterinary or human neurologists, and theneurologic basis for the test can only behypothesized. The rationale for using thewidth of the response, as in index of func-tion, is not at all clear. Although an increasein width has been reported for animals



Cory-Slechta et al.


treated with acrylamide, interpretation of adecrease in width is less clear.

In the rotorod test a rat is placed on aslowly rotating rod or dowel, and the endpoint is the duration that the rat maintains itsperch on the rod (60). Although commer-cially available the rotorod has achieved onlymodest popularity, primarily because it istime-consuming and cumbersome to con-duct, and some animals jump spontaneouslyfrom the rod. The rotorod test illustrates thepremise that off-the-shelf equipment does notsubstitute for expertise and training (seesection “Expertise and Training”).

Analysis and interpretation. In adultneurotoxicity studies the strength of a reflex orreaction is an important index of function.Because the strength of the response is mostfrequently measured, it is easy to overlook thefact that all reflexes and reactions have a sen-sory component. For example, failure to flexthe leg in response to a toe pinch can reflectloss of sensation or inability to move the leg.Thus, patterns of changes in several reflexesand reactions are generally characterized byneurologic examination to provide inter-pretable data. In developmental studies theevolution of a reflex or reaction over time isparticularly important. Delays in the appear-ance of a reflex or reaction are important indi-cators of an adverse developmental effect. Thereflexes and reactions most commonly evalu-ated in developmental studies include flexorreflex, extensor thrust, rooting, placing, surfaceand air righting, grasp reflex, auditory startle,negative geotropism, and swimming (61,62).

Research needs. A current dilemma inassessing motor function is that tests of reflexesand reactions are relatively quick and easy toperform, but their sensitivity in the context ofDNT testing is uncertain. In contrast, com-plex procedures exist for characterizing evenminor changes in some motor functions [forexample, see Stanford and Fowler (63)], butthe scope of such procedures is limited, andthe procedures too cumbersome to be usedroutinely. Thus, there is a need to develop andvalidate technology and procedures thatmeasure motor function objectively and sensi-tively, yet are sufficiently flexible to be usedwith large numbers of animals.

There is an additional need to developtechnology to measure motor activity.Evaluating the behavior of rats in a novelenvironment has fascinated and frustrated sci-entists in a variety of disciplines, includingethology, experimental psychology, psy-chopharmacology, neuroscience, and neuro-toxicology (48,53–55,64,65). Decades ago,observers of behavior recognized that rats innovel environments engaged in a variety ofbehaviors that suggested this testing environ-ment might be used to evaluate sensory andmotor function, emotions, and/or cognition

(66–68). Indeed, contemporary experimenta-tion has shown that the movements of rats ina novel environment reflect the activity of acoordinated navigational system that dependson allothetic (e.g., visual, olfactory, tactilestimuli) and ideothetic (e.g., proprioceptivestimuli from the animal’s own movement)stimuli (69–70). The pattern of behavior alsoreflects the emotional state of the rat (71,72)as well as motor function (73). Recently,observation of rat behavior in a novel envi-ronment has emerged as a principal tool fordetecting the effects of neurotoxicants on awide range of neurologic functions (3,74).

Availability of this rich collection ofbehaviors that rats exhibit in a novel environ-ment has prompted both scientists and com-mercial equipment manufacturers to developtechnology to measure behavior in this testsituation (38,73,75,76). The two most com-mon types of systems are based on photocell(75) or video (77) technology. To date, noautomated test system has been capable ofsatisfactorily detecting and quantifying even asmall percentage of the range of normal andabnormal behavioral functions that are avail-able and that can be obtained through carefulobservation. The initial richness of behaviorthat stimulates interest in open-field activityis also the challenge to be overcome indesigning test paradigms. Specifically, a num-ber of diverse normal and abnormal behaviorscan occur, and it is valuable to obtain a senseof both the temporal and spatial distributionof the behavior. More complicated technol-ogy systems with multiple subsystems (e.g.,photocell, video, ultrasound, touch detectors,proximity detectors) or data analysis protocolshave been developed (51,76,78–81), butnone has been sufficiently useful or practicalto become universally popular.

Cognitive Function The nature of the experimental questionshould guide the pursuit of the behavioralbaselines that will be used. For example, sim-ple tests of learning, providing they are wellcontrolled, may provide information aboutwhether a potential deficit in cognitive func-tion occurs (detection of effect), whereasmore sensitive and complex procedures mayprovide information about the behavioralmechanisms by which such effects occur(characterization of effect). Reliance on com-plex approaches to answer questions aboutpotential cognitive deficits could be useful,even in screening, to determine whether apotential learning/memory impairment mightbe attributable to deficits in other areas ofnervous system function.

The current U.S. EPA protocol for DNTtesting (9) requires assessment of cognitivefunction at two ages. Certainly such measuresare critical components of a DNT assessment

to address concerns over potential long-termconsequences of exposures to toxicants duringperiods of brain development. When assess-ing cognitive function, numerous issues mustbe considered before implementing suchmeasures.

Cognitive function is often thought of asencompassing learning, memory, and atten-tion processes. Both learning and memoryfunctions have been extensively studied,defined, and described in the behavioral neu-roscience literature. The issue of attention lagsbehind, as it has not been as systematicallystudied and defined. The term attentionremains a global behavioral construct that mayinclude numerous response classes such as dis-tractibility, impulsivity, sensitivity to delay,activity level, perseveration, sustained atten-tion, and inability to manage delay of reward.

Animal model. The learning task selectedshould be one appropriate to the species andto the developmental age of the subjectsbeing tested. For example, many differentparadigms have been developed to assessaspects of cognitive function, but these havedifferent applicability across species as well asacross developmental stages of life. One strat-egy that may be advantageous to the processof risk assessment is the use of the samebehavioral paradigms across species, includinghumans (82). Tests such as repeated learningor acquisition of response chains and delayedmatching paradigms, respectively, can be usedacross species with appropriate parametricmodifications. These are well-validatedapproaches that have been used extensivelyand thus allow incorporation of a large data-base into the assessment. Using these meth-ods eliminates the need for the assumptionthat dependent variables from neuropsycho-logic and clinical tests in humans measure thesame behavioral processes as the experimentalcognitive testing procedures.

Stimulus parameters. Most tests of cogni-tive function employ measures of accuracy as aprimary dependent variable. One importantaspect of the test is the level of accuracy main-tained under normal conditions. The taskshould maintain levels of accuracy from whicheither increases or decreases as a result of expo-sures can be measured. Tasks that are too easy,as indicated by the ability of subjects toachieve high levels of accuracy, are not suffi-ciently sensitive for detecting toxicant effects.Carson et al. (83) reported, for example, thatlead-exposed sheep exhibited no differencefrom control in learning to discriminate a ver-tical from a horizontal line gradient, whereasthey exhibited lower levels of accuracy in dis-criminating a large from a small circle, withthe latter discrimination requiring signifi-cantly longer to learn in controls. Similarly,Wood et al. (84) demonstrated that toluenedisrupted behavior that was at lower accuracy



levels but did not disrupt behavior maintainedat high accuracy levels in a fixed consecutivenumber schedule of reinforcement. Similarly,behavioral paradigms that are too difficult,i.e., those that maintain very low levels ofaccuracy ultimately, also decrease the proba-bility of detecting a toxicant-induced alter-ation. Other important aspects of stimulusparameters that must be considered includethe saliency of any environmental stimuli usedin the cognitive problem to be evaluated andthe relevance of the dimension to the speciesbeing tested. For example, stimuli that differin color may fail to be sufficiently discrim-inable to rodents that possess no color vision(85). Although odors are particularly relevantto mice and rats, the inability of the investiga-tor to precisely control odor onset and offsetlimits the utility of this measurement andcould inadvertently change the nature of thetask contingencies (86). Stimulus parametersof other aspects of such tests are also critical(e.g., delay values in memory tests, timesbetween trials), and the literature should beconsulted for appropriate values.

Statistics. For most tests of cognitivefunction, a primary measure of interest willbe either accuracy or latency. Ideally, thebehavioral tests used to evaluate learning,memory, and attention should provide base-line data for these dependent measures withminimal variability both between subjectsand within subjects across time, such thateither increases or decreases in these mea-sures can be detected with typical groupsizes as reported in the literature. In most

studies of cognitive function, behavior ofindividual animals is measured repeatedlyacross sessions (Figures 1,2). When multipledata points are derived from a single subject,they are not considered independent replica-tions; clearly, the behavior of an individualanimal is expected to be related to its pastperformance. Therefore, unless all relevantdata are collapsed to a single number, initialstatistical analyses require repeated measuresapproaches that consider this lack of inde-pendence. Thus, the common practice ofusing t-tests or one-factor ANOVAs (seeexample in Figure 3) to analyze multipledata points from single subjects is inappro-priate. Repeated measures analyses establishwhether there are any main effects of thetreatment factor per se as collapsed acrossthe repeated measure. Main effects in theanalysis indicate that the behavioral data ofthe control and treated groups are fit by par-allel lines in the simplest case. In other casesthe data of control versus treated groups may

differ only under some circumstances, e.g.,only during the final five sessions of testing.In such a case a statistical interaction wouldbe expected from an analysis, which wouldindicate that there was not only an effect oftreatment but that it also occurred onlyunder specific conditions of the repeated fac-tor. With this type of statistical interaction,one predicts intersecting functions of the fitsof the control versus treated groups. Only ifit can be established that there is either amain effect of treatment or some type ofinteraction, once the repeated measures havebeen taken into account across the repeatedmeasure, is it permissible to begin tocompare specific data points.

Age relevance. Questions regarding cogni-tive deficits in response to exposures early indevelopment often focus on long-term out-come, i.e., changes in these behavioral func-tions as organisms mature. There may becircumstances, however, in which it is desir-able to determine in juvenile animals whethercognitive functions have been affected byexposure early in development. Assessing cog-nitive function in young animals or childrenrequires tests designed to account for thephysiologic and physical limitations of thestage of development. While such procedureshave been reported in the experimental


Figure 1. Hypothetical scheme depicting latency in sec-onds across trials in a water maze in which the task is tolocate a submerged platform that permits an escaperesponse. Typically, as acquisition or learning occurs,the latency to find the platform declines across trials(control curve). However, acquisition can be affected bychanges in sensory function, motor capabilities, andmotivation, all resulting in nonspecific changes in learn-ing. This would typically be manifest as a parallel curve(nonspecific), where differences from control in latencywere evident even in trial 1. A treatment-related changein learning that might be indicative of a specific effect isshown as well (specific), in which one would expect thesame latency as untreated control in initial trials (indica-tive of equivalent motor, sensory, and motivational lev-els), with gradually diverging latencies that do notdecline as rapidly as control.

Figure 2. Data from control (0 ppm) and lead-treated (50ppm, 250 ppm) rats working on a multiple schedule ofrepeated learning (A) and performance (B). The repeatedlearning component, which required rats to learn a newsequence of responses during each session, alternatedduring each session with the performance component,which required only repetition of an already-learnedsequence. Accuracy levels during the performance com-ponent were substantially higher than in the repeatedlearning component, as would be expected since it wasan already-acquired sequence. Control rats showed agradual increase in accuracy in the repeated learningcomponent across sessions, indicative of the develop-ment of a strategy for solving the correct sequence forthe session. Lead-treated rats showed no such increaseand basically remained just above chance levels of accu-racy. In contrast, lead-treated rats showed no difficultyin performing an already-learned sequence, thus demon-strating a selective effect of a treatment on learning.Data modified from Cohn et al. (97).

▲

▲

▲▲

▲▲

▲ ▲ ▲ ▲

●

●

●

●●

●●

●

●●

■

■

■■

■■

■■

■

●

■

▲

ControlSpecificNonspecific

70

60

50

40

30

20

10

01 2 3 4 5 6 7 8 9 10

Late

ncy

(sec

)

Trial

■

■ ■■■

100

80

60

40

20

01 2 3 4

●

●●

●

● ●●●

●

●

■

0 ppm50 ppm250 ppm

A

Accu

racy

Session block

100

80

60

40

20

01 2 3 4

●■●●●■■ ■●● ●●

B

Session block

Accu

racy

Figure 3. The delayed match-to-sample procedureimposes a delay between a sample stimulus and thesubsequent presentation of two stimuli, one of whichmatches that sample. The selection of the matchingstimulus is rewarded. This figure depicts a hypotheticalscheme relating changes in acccuracy to the length ofthe delay (in seconds). Typically, accuracy declines asthe length of the delay is increased (control). A specificchange in memory is indicated by the curve labeled“specific,” where accuracy levels of the treated groupare equivalent to controls at the 0-sec delay, where nodelay is imposed and no remembering required. Whenaccuracy levels are lower even in the 0-sec delay (non-specific), it indicates that other changes in behavior(e.g., motor, sensory, motivational) are contributing tothe deficit. Asterisk (*) and bracket ( ] ) indicate a signifi-cant difference in accuracy between “specific” and“control” at the 12-sec delay point only, based on theinappropriate use of a t-test rather than the appropriaterepeated measures approach.

●

■

●

■

▲

ControlSpecificNonspecific

●

●

●

●

●▲

▲

▲

▲

▲

▲

■

■

■

■

■

*

100

90

80

70

60

50

400 4 8 12 16 20 24

Accu

racy

Delay (sec)


Cory-Slechta et al.


animal literature (87–90), they have yet to befully incorporated into standard test batteries,although the DNT battery requires testing ataround 24 days of age and again at about60–70 days of age. When animal assignmentsare made in a DNT study for cognitive tests,behavioral histories of the the test subjectsshould be considered. In most cases, animalstested at a young age should not be testedusing the same method at the laterage. Previous learned behaviors may carryover into the second testing, confounding anyability to assess learning at the later age. Careshould be taken to fully understand the rami-fications of this. Further, some of the operanttouch-screen technologies used in studieswith adult humans are being used increas-ingly in studies with children as young as 4–5years of age (91–93). This is encouraging, butfurther development and validation of suchtests are urgently needed.

Analysis and interpretation. In many sim-pler tests of learning and memory, it is possi-ble for changes in sensory function, motorbehavior, and/or motivation to indirectlyinfluence the dependent measures that areused and therefore change behavior. Simpleparadigms generally fail to include controlprocedures for assessing these possibilities.Although the water maze is a popular methodto measure learning and has proven useful inmany contexts (94–96), it also providesnumerous examples of the difficulty of inter-preting learning impairments. For example,deficits in motor behavior such as strength,endurance, or coordination might result inincreased swimming times required to reachan escape platform in a water maze. Asdecreases in latencies are considered an indexof learning in this paradigm, the longer laten-cies could be misinterpreted as a learningimpairment. In a water maze this might resultin the type of hypothetical data presented inFigure 1 (nonspecific difference), where thecontrol and nonspecific groups exhibit paral-lel decreases in latency over the course of tri-als in such a task. The notable difference inlatency, even in the first trial, would suggestthat noncognitive influences were producedby treatment and contribute to the differ-ences between the curves. A function moreconsistent with an interpretation of specificchanges in cognitive function would insteadbe manifest as intersecting lines, with noapparent differences initially in latencies butwith a slower rate of decline in latency orerrors over time (Figure 1).

One mechanism to separate learningeffects from nonspecific behavioral influencesrelies on a paradigm such as the multipleschedule of repeated learning and perfor-mance (97). This comprises two differentbehavioral components that alternate over thecourse of a behavioral test session, with each

component associated with a different envi-ronmental stimulus. The active environmen-tal stimulus provides information to thesubject about which component is currentlyoperative. In the repeated-learning compo-nent the subject is required to learn asequence of responses, and this sequencechanges with each test session in an unpre-dictable way. This allows the generation of alearning curve during each session. The per-formance component requires the executionof a sequence of responses of the same lengthas that in the repeated-learning component,but which has already been learned andremains the same over the course of theexperiment. It also requires the same motor,sensory, and motivational capabilities as doesbehavior in the repeated-learning componentbut does not require learning per se as long asthe task can be learned initially by the subjecttreated during development. Thus, a truedeficit in learning under this schedule wouldbe manifest as a decrease in accuracy in thelearning but not in the performance compo-nent of the schedule. Concurrent decreases inaccuracy in the performance componentwould be indicative of nonspecific changes inbehavior, whether sensory, motor, or motiva-tional, that indirectly contributed to anydecreases in the learning component. Figure 2presents an example of a selective effect onlearning following chronic low-level post-weaning lead exposure of rats (97), as indi-cated by decreases in accuracy in therepeated-learning component and the absenceof any such changes in the performance com-ponent. Validation of this paradigm in thelaboratory requires that the investigator beable to demonstrate that acquisition doesindeed occur in the repeated-learning compo-nent, i.e., that an increase in accuracy overthe course of this component from chancelevels can be shown. This approach also hasapplicability across species ranging from themouse to the human (94,97–99).

Like the water maze used to measurelearning (or short-term memory), simpleapproaches to measurement of memory suchas the frequently employed passive avoidanceparadigm also present difficulties of interpre-tation. This technique relies on the ability ofsubjects to remember in which compartmentof a two-compartment chamber they had pre-viously received shock; the longer it takes forthem to re-enter that compartment, thegreater the attributed memory. However, dif-ficulties in sensory processing may render theenvironmental stimuli that dissociate theshocked from the nonshocked compartmentsas less distinct, thus causing premature re-entries. In cases where the shock trainingoccurs after experimental treatments inbetween-groups designs, the treatment itselfmay produce differences in shock sensitivity

that are not apparent in any way but that caninfluence the subsequent avoidance of theshocked compartment. This can also bechecked with a shock titration curve (100).

Paradigms that explicitly control for suchalternative explanations include delayedmatching-to-sample, which can also be usedacross species. Memory paradigms typicallymeasure accuracy of remembering followingvarious delay intervals. Increasing delays areassociated with increasing difficulty in remem-bering and thus increases in errors (decreasesin accuracy), resulting in a typical delay func-tion (Figure 3, control). To determine theextent to which any alteration in the delayfunction in response to a treatment is causedby memory impairment rather than changesin other behavioral processes, it is critical toinclude a no-delay condition (0-sec delay). Inthis trial no delay is imposed before the sub-ject is asked to match two stimuli, and thus noremembering is required. If deficits in accu-racy are observed under these conditions (seenonspecific effect curve in Figure 3), they can-not be ascribed to memory impairments andwould suggest that treatment-related decreasesin accuracy are non-mnemonic resulting fromnonspecific behavioral influences. A truedeficit in memory would be reflected insteadin a curve in which there were no impair-ments of accuracy at the 0-sec delay, andincreasing delay values would be associatedwith an increasing decline in accuracy relativeto control (Figure 3, “specific” curve). Figure4 shows a delay function for children 10–11years of age using the same behavioral testadministered from a computerized touch-screen apparatus (101). Such paradigmsrequire the incorporation of delay values thatultimately result in chance levels of accuracyfor the species being tested. Using delay valuesthat are too short, and thus do not produceany substantive decline in accuracy, will

Figure 4. A delay function relating changes in accuracyto delay value (seconds) derived from a sample of 10normal children 10–12 years of age using the Cantabversion of the delayed match to sample procedure (101).

100

90

80

70

60

50

404 12

Accu

racy

Delay (sec)

0Stimulus

●

●

●

●



render the paradigm potentially insensitive tothe detection of a treatment-related decline.

Research needs. Several research needsmerit particular mention. First is the need foradditional paradigms for testing cognitivefunction that can be used earlier in develop-ment. The question of long-term adverseconsequences usually results in testing ofexperimental animals in adulthood, but inhuman populations, tests are used earlier indevelopment to evaluate the ontogeny of sucheffects. An additional need is for simple assaysof learning and memory with adequate sensi-tivity but that could be used in the context ofscreening assessments and thus trained morerapidly than more sophisticated proceduressuch as the multiple schedule of repeatedlearning and performance and delayed match-to-sample. Finally, a more systematic andrefined understanding of attention and itsvarious components will be required tounderstand its component parts and theirunderlying anatomic and neurochemical sub-strates, and how these aspects of attentionmay be differentially affected by exposures tovarious toxicants.

Social BehaviorLarge portions of the behavioral repertoire ofmost species are devoted to relations withconspecifics. Aggressive, affiliative, mating,play, and parental behaviors are examples ofthis category and are among the phenomenamost studied by ethologists, biopsychologists,and other life scientists. These behaviors tendto receive less attention from toxicology thanassessments of individual behaviors, in partbecause in the typical laboratory environmentrodents—the predominant test species—arenot given many opportunities for social inter-actions. Additionally, measurements of socialbehaviors are less standardized than, forexample, motor activity, and are not as easilyincorporated into batteries of screening tests.In addition, because most social behaviorsmust first be interpreted to be quantified bycounts or ratings of defined actions and maysometimes be difficult to automate, theyoften require trained observers. They mayalso require the observer to record theresponses of two or more animals concur-rently, which is another complicating factor.

Social behaviors may be destined toattract more attention from neurobehavioraltoxicology because of the types of questionsrecently aroused by endocrine disrupters.Conspecific behaviors in adults such as mat-ing and aggression are linked directly to pre-vailing hormonal mechanisms and states, asare behaviors of somewhat greater subtletysuch as birdsong patterns and the ordering ofdominance hierarchies.

Social behaviors, moreover, are not theexclusive province of hormonally active

agents. They are also modified by many otherclasses of developmental neurotoxicants thatmay act either through neuroendocrinemechanisms or by directly influencing brainmorphology or neurochemistry. Prenatalalcohol exposure, for example, can impaircopulatory behavior in male rats. Prenatallead exposure intensifies aggressive behaviorsin hamsters, as measured by the response tointruders (102). Lead is also a recognizedreproductive toxicant. Does this effect repre-sent actions on neuroendocrine status? Manytherapeutic agents administered prenatally,such as the benzodiazepine oxazepam, canmodify subsequent social behaviors of the off-spring such as maternal care (103). Aggressiveand defensive behaviors are accompanied bylarge changes in selected brain dopamine,serotonin, and γ-aminobutyric acid systems(104). Maternal behaviors, aggressive behav-iors, and sexual behaviors are among the mostpromising candidates for social behavior mea-sures in developmental neurotoxicology.

Maternal behavior. Altered endocrine sta-tus during fetal development can modifymany postnatal behaviors, but little informa-tion is available on how prenatal treatmentaffects maternal behavior in female offspring.If prenatal exposures interfere with endocrinesystem development, the consequences couldappear as abnormalities in maternal behav-iors, which are synchronized with a series ofhormonal changes that act on the reproduc-tive tract, the mammary gland, and the cen-tral nervous system. The immediatehormonal events for maternal behavior occurduring pregnancy and also around parturitionand lactation, when maternal behavior is fullyinitiated. Precursors of the full repertoire ofmaternal behaviors begin during gestation. Ifpregnant rats are tested for maternal behaviorby presenting them with test pups, they showa gradual increase in such behaviors as partu-rition approaches (nursing posture, lickingand retrieving, nest building). After parturi-tion, maternal behavior is maintained essen-tially by stimulation from the young, such assuckling. A variety of behaviors may bescored in studies of maternal behavior(105,106). These include retrieval of dis-placed pups, nest building, nursing and lick-ing, and attacks against intruders (107).Instrumental techniques have been reportedby Vernotica et al. (108) and Lee at al. (109)that could serve as more automated methods.

Aggressive or attack behaviors. Aggressionis a label applied to common responses inmany species to invasions of territory, in con-testing for mates, in exercising dominance,and even in play behavior. It is among themost frequent social behaviors displayed byanimals, including common laboratoryspecies (110). Aggressive behaviors, whichconsist of several components, including

attack, defensive, and submissive responses,can be modified by many drugs and havebeen linked to specific neurotransmitter sys-tems (104). In rats, for example, an intrudertypically responds to threat postures orattacks by the resident by adopting defensivepostures, while the resident may follow threatpostures by leaping and biting. Normally,aggressive behaviors in laboratory and housemice are both organized and maintained bytestosterone. For the full expression of suchbehaviors to occur, androgens must be pre-sent both during brain development and sub-sequently. A scoring system has been used torecord these and other behaviors on the partof the resident and the intruder (111).Examples of prenatal chemical exposurelinked to adult aggressive behaviors are givenin Palanza et al. (112) and Fiore et al. (113).

Mating behaviors. Copulatory mechanicsare only a minor feature of male sexual behav-ior, which is driven and organized by the cen-tral nervous system. Female sexual behavior isalso predominantly dependent on the centralnervous system. For these reasons, any plan tostudy mating behaviors should include situa-tions designed to reveal their behavioral andespecially motivational complexities.

Various measures of receptivity thatdescribe female motivation and the reinforc-ing potency of sex and that simulate the con-ditions of sexual behavior in natural settingshave been devised. For example, a two-compartment test apparatus has been devel-oped in which only the female is able tomove from one compartment to the otherbecause of her smaller size (114). A similarapproach makes use of a bilevel chamber(115). Such chambers consist of two levelsconnected by a set of ramps. Because femalescan run from level to level, the males areforced to follow to attain copulation. In thestandard assessment of copulatory function,males are generally provided with a primedfemale, most often one that has been ovariec-tomized and then acutely treated with a com-bination of estradiol and progesterone toinduce receptivity. Observers typically recordmeasures of copulatory performance.Copulatory performance in male rats pro-vided an index of interference with gonadaldevelopment produced by gestational expo-sure to 2,3,7,8-tetrachlorodibenzo-p-dioxinin a study by Mably et al. (116).

Motivational and incentive measuresmight, in fact, prove more sensitive to devel-opmental toxicants than scores based simplyon the isolated sex act itself, because experi-mental data show the breadth and complexityof anatomic, neurochemical, and neuroen-docrine influences governing sexual motiva-tion (117). Amstislavsky and co-workers (118)relied on a simple technique with mice. Theytreated pregnant mice with methoxychlor,



Cory-Slechta et al.


then assessed sexual motivation in maleoffspring by using a cage with a plasticpartition behind which they had placed anestrous female. In addition, the bilevel cham-ber described above has also been used toexplore sexual motivation in males (119).

Research needs. Social or conspecificbehaviors have received relatively little atten-tion from neurotoxicologists. Naturalisticstudies of behavior, the discipline of ethology,have not proven a popular area of neurotoxic-ity research (120). Now, with the expandinginterest in endocrine disruption as an index oftoxicity, the appearance of reports linkinglead exposure to aggression (121), and publicconcern that environmental chemicals may beresponsible for some antisocial behaviors, thesituation is primed for a new look.

If the types of social behaviors describedin this review are to be integrated into screen-ing batteries, they must display the attributescommon to most of the tests now incorpo-rated into such protocols. Discussed beloware three interconnected issues would have tobe resolved.

Reliability among observers. Extensivetraining is required to ensure that differentobservers in the same laboratory agree onscoring. Attaining agreement amongobservers demands attention to precise defini-tions and practice, even for simpler functionalobservation batteries. Few reports includemeasures of interobserver reliability or train-ing procedures. Agreement among laborato-ries must be achieved if social behaviors are toserve as useful end points.

Expertise. Most of the research conductedon social behaviors originates in academic set-tings, where investigators strive for originalityin technique. It would be rare for a group ofexperts in maternal behavior, for example, toagree on common definitions and approachesso that data from different laboratories can becompared.

Time commitments. It would be difficultto include end points requiring a large invest-ment of investigator or even technician timein screening batteries, even for tier II assess-ments. Automation is not yet a common fea-ture of social behavior research, althoughsome of the methods described here eitherhave been adapted for it or can be convertedwithout extensive modification.

A number of methods suitable for use asthe basis for creation of more efficient tech-niques for measuring social behaviors havebeen described in this report. Little standard-ization has been accomplished, comparedwith accepted techniques such as functionalobservation batteries, schedule-controlledoperant behavior, and motor activity.Neurobehavioral toxicologists should be inthe vanguard of an effort to devise newtechniques and to perfect older ones.

Autonomic and ThermoregulatoryFunctionThe autonomic nervous system (ANS)controls the function of a wide variety of organsystems, including the respiratory, cardiovascu-lar, and genitourinary systems. Physiologistsand pharmacologists have developed manysophisticated methods to measure the functionof these organ systems. Measurement of ANSfunction has not been a priority for eitherdevelopmental neurotoxicologists or DNTtesting guidelines. In part, this lack of atten-tion may reflect the relative paucity of chemi-cals that damage the ANS in rats (122). Inaddition, it is well recognized that the ANScontrols vital functions, so that damage to theANS should significantly compromise generalhealth and/or reproductive capacity. Thus,children with inherited or acquired dysauto-nomia evince multiorgan disturbances in criti-cal body functions and do not thrive (123). Ingeneral, the signs of ANS toxicity are quiteobvious. For example, clinical conditions thatdisrupt the innervation of the bowel areexpressed as hyper- or hypomotility states andare reflected by colic, abdominal distension,constipation, or diarrhea. Body weight loss is afrequent correlate (124).

Thermoregulation is accomplishedthrough a network of peripheral and centralthermoreceptors and effectors that includesomatic (e.g., moving to a warmer or colderlocation), endocrine, and autonomic (e.g.,peripheral vasodilation or vasoconstruction)components (125,126). The function ofmany components of the thermoregulatorysystem can be measured in rats using a varietyof established test methods. Like autonomicfunction (see above), thermoregulation hasbeen the purview of physiologists, pharmacol-ogists, and neuroscientists and has not been apriority for either developmental neurotoxi-cologists or DNT testing guidelines. Thisinattention may be unfortunate, because ther-moregulatory responses to neurotoxicants arean important component of the reaction ofadult rats to neurotoxicants (125–127). Forexample, hyperthermia is an important partof the pathophysiology of the neurotoxiceffect of methamphetamine on dopamine-containing nerve terminals in the corpusstriatum of the rat (128). Moreover, Gordonand colleagues have shown that perinatalexposure to dioxin can produce long-lastingchanges in autonomic and behavioralthermoregulation (129,130).

Biologic RhythmsClasses of behavior that exhibit biologicrhythms include feeding, drinking, sleeping,motor activity, and mating (131). The cycleassociated with each of these behaviors repre-sents a potential tool that could be examinedfor potential effects of chemical treatment.

For example, chemicals may disrupt or alterthe diurnal pattern of locomotor activity(132,133), and there are chemicals that eliciteither a more pronounced or diminisheddiurnal pattern of locomotor activity in rats.In addition, diurnal patterns of ingestionexhibit changes not evident by measures ofthe total amount ingested (134). Triethyltin,for example, alters the diurnal pattern ofwater ingestion but not total daily consump-tion (132), whereas trimethyltin increasestotal water consumption while the diurnalpattern is largely preserved. Although thevalue of such approaches is illustrated bythese examples, this approach to neurotoxic-ity assessment remains relatively unexplored.

Considered more relevant to the presentdiscussion is how biologic rhythms or cyclescan affect behavioral test results by introduc-ing additional variability and complicatingthe interpretation of behavioral test results.For example, the level of activity exhibited byanimals over the course of a day is one of themost apparent and well-established behavioralmanifestations of circadian rhythms. Withinan 8-hr workday, levels of horizontal and ver-tical activity vary by as much as 20–30%(135). If not adequately controlled, this cancontribute substantially to variability in mea-sures of motor activity. Hormonal cycles mayalso contribute to variability. Levels of activityin the running wheel are 3–10 times higherfor female rats in estrous than levels duringdiestrous (136). Diurnal factors have alsobeen shown to affect the ability of tests todetect the effects of certain chemical treat-ments (137,138). Thus, circadian rhythmsrepresent a significant source of variability forbehavioral test results. If left unmanaged, sta-tistical power will be reduced, increasing theprobability of a type II error. One approachused to compensate for this is to increase thenumber of animals in each dose group.However, the potential gain associated withthis approach may not be realized if appropri-ate precautions are not taken, as additionaltime will be required to test those animals.Thus, to the extent possible, appropriate mea-sures should be incorporated into the studydesign, such as including representatives fromeach dose group in each set of animals testedat one time, and testing those animals overmore days rather than extending hours on agiven test day.

Conclusions

Careful consideration of a number of experi-mental design issues is the key to the successof a study using behavioral methods to exam-ine DNT. Identifying clear study goals andobjectives is paramount in designing a strongstudy. Study goals and objectives are a guidein selection of the methods used, the appro-priate animal model, and the equipment



needed. The study goals can be used toidentify the behavioral test methods by guid-ing the evaluation of the sensitivity versusselectivity required to answer the scientificquestion at hand. Method selection also restson consideration of available resources,including equipment, funds, and personnel.An understanding of the inherent variabilityin the methods selected can and should beused to determine the number of animalsrequired to detect an effect of concern.Variability can arise from a number of differ-ent sources in tests of sensory function, andthis is particularly true in studying effects ondeveloping animals. Averaging response dataacross ages in developing animals couldincrease variability unnecessarily, as theresponse may change drastically in a veryshort time because of increases in bodyweight and increased sensitivity of the devel-oping sensory system. Appropriate statisticalanalyses are vital for defensible interpretationof behavioral data. Repeated measures areoften used in DNT testing and must betreated as such in the statistical analysis.When using behavioral methods in toxicitytesting, it is important that normative data bedeveloped for a method demonstrating thatthe test method as used can detect and char-acterize the effects of varying the magnitudeof these properties. Positive and negative con-trol data are needed to support the validationof the method and aid in data interpretation.

Proper expertise is required to design,conduct, and interpret a study of develop-mental neurotoxicity using behavioral meth-ods. Training in experimental psychology orpsychopharmacology and in statistics pro-vides a background important for design andinterpretation of these studies. Those whoconduct the tests should be trained in properperformance of each behavioral test andshould have a good understanding of poten-tial confounders.

Behavioral methods are used to detect andcharacterize developmental neurotoxic effectson sensory cognitive and motor system func-tions. The major sensory systems of concern intoxicology include visual, auditory, olfactory,nociceptive (pain and other noxious stimuli),somatosensory, and vestibular. There are anumber of stimulus properties shared by allsensory systems, including intensity, fre-quency, duration, and location in space. Onetenant of sensory function testing is that itnever yields a direct measurement of sensation;instead, a change in sensory function isinferred based on the observed change in themotor response evaluated. A better under-standing of the relationship between the sim-ple sensory system tests currently used in DNTstudies and any underlying changes in thestructure and/or function of the sensory systemwill advance our ability to identify sensory

system toxicity. Behavioral toxicologists cancontinue to learn from recent advances inneurobiology and genomics that may allow anincreased understanding of the physiologic andstructural bases for sensory function.

Behavioral tests of motor dysfunction inanimals include those used to detect sponta-neous movement disorders such as changes ingait, tremors, and myoclonus, and those usedto detect changes in induced movement suchas reflexes, reactions, and movements underoperant control. Tests of motor functioninclude observation of locomotion, measure-ment of locomotor activity, and tests ofreflexes and reactions. Patterns of changes inseveral reflexes and reactions are generallycharacterized by neurologic examination.There is a need to develop and validate tech-nology and procedures that measure motorfunction objectively and sensitively, yet areflexible enough to be used with large numbersof animals. Interpretation of motor functiontests must take into account the limitations ofthe test equipment as well as any potentialbiologic confounders.

Cognitive function is thought to encom-pass learning, memory, and attentionprocesses. Assessment of cognitive function isa critical component of a DNT assessment toaddress concerns over potential long-termconsequences of exposures to toxicants duringbrain development. In many simpler tests oflearning and memory, changes in sensoryfunction, motor behavior, and/or motivationmay indirectly influence the dependent mea-sures used, and therefore change behavior.Simple paradigms generally do not includecontrol procedures for assessing these possi-bilities. Reliance on relatively complexapproaches to answer questions about poten-tial cognitive deficits could be useful, even inscreening, to determine whether a potentiallearning/memory impairment might beattributable to deficits in other areas ofnervous system function.

Social behaviors, such as aggressive, affilia-tive, mating, play, and parental behaviors tendto receive less attention from toxicologiststhan individual behaviors. Social behaviorsmay be modified by developmental neurotoxi-cants, including hormonally active agents thatmay act through neuroendocrine mechanismsor by directly influencing brain morphologyor neurochemistry. Techniques used to mea-sure social behaviors are less standardized thanindividual behaviors. Because most socialbehaviors have to be interpreted to be quanti-fied, the tests are difficult to automate andtrained observers are often required to per-form the tests. If the types of social behaviorsdescribed in this review are to be integratedinto screening batteries, they will have to dis-play the attributes common to most of thetests now incorporated into such protocols.

Measurement of ANS function has notbeen a priority for developmental neurotoxi-cologists, though a number of sophisticatedtests developed by physiologists, pharmacolo-gists, and neuroscientists could be adopted.This inattention may be unfortunate becausethermoregulatory responses to neurotoxicantsare an important component of the reactionof mammals to neurotoxicants. Classes ofbehavior that exhibit biologic rhythmsinclude feeding, drinking, sleeping, motoractivity, and mating. The cycle associatedwith each of these behaviors represents a toolthat could be examined to identify effects ofchemical treatment.

Behavioral testing methods to measuresensory, motor, and cognitive function arewell developed, but there is room for improve-ment in study design, conduct, analysis, andinterpretation. Tests to characterize effects ofdevelopmental neurotoxicants on socialbehavior, the ANS, thermoregulation, andcircadian rhythms are presently underutilized.

REFERENCES AND NOTES

1. Aminoff MJ, Ross A. Electrodiagnosis. In: Clinical Neurology.New York:Churchill Livingstone, 1999.

2. Lotti M. Organophosphorus compounds. In: Experimental andClinical Neurotoxicology, 2nd ed (Spencer PS, Schaumburg HH,eds). New York:Oxford University Press, 2000;897–925.

3. Ross JF, Mattsson JL, Fix AS. Expanded clinical observations intoxicity studies: historical perspectives and contemporaryissues. Regul Toxicol Pharmacol 28:17–26 (1998).

4. Moser VC, Ross JF. Training Video and Reference Manual for aFunctional Observational Battery. Research Triangle Park,NC:U.S. Environmental Protection Agency, 1996.

5. Abbey H, Howard E. Statistical procedures in developmentalstudies on a species with multiple offspring. Dev Psychobiol6:329–335 (1973).

6. Cox C. Statistical issues for animal studies in developmentalneurotoxicity. In: Neurobehavioral Toxicity: AnalysisInterpretation (Weiss B, O’Donoghue J, eds). New York:RavenPress, 1994;93–101.

7. Holson RR, Pearce B. Principles and pitfalls in the analysis ofprenatal treatment effects in multiparous species. NeurotoxicolTeratol 14:221–228 (1992).

8. Reily EP, Meyer LJ. Consideration for the design, implementa-tion and interpretation of animal models of fetal alcoholeffects. Neurotoxicol Teratol 6:97–101 (1984).

9. U.S. EPA. Developmental Neurotoxicity Study. OPPTS 870.6300.EPA 712-C98-239. Washington, DC:U.S. EnvironmentalProtection Agency, 1998.

10. Levin ED, Simon BB, Schmechel DE, Glasgow HB Jr, Deamer-Melia NJ, Burkholder JM, Moser VC, Jensen K, Harry GJ.Pfiesteria toxin and learning performance. Neurotoxicol Teratol21(3):215–221 (1999).

11. Cohen J. Statistical Power Analysis for the BehavioralSciences. Hillsdale, NJ:Lawrence Erlbaum, 1987.

12. Muller KE, Benignus VA. Increasing scientific power with statis-tical power. Neurotoxicol Teratol 14:211–219 (1992).

13. Crofton KM, Howard JL, Moser VC, Gill MW, Reiter LW, TilsonH, MacPhail RC. Interlaboratory comparison of motor activityexperiments: implications for neurotoxicological assessments.Neurotoxicol Teratol 13:599–609 (1991).

14. Wise LD, Allen HL, Hoe CMVDR, Gerson RJ. Developmentalneurotoxicity evaluation of the avermectin pesticide,emamectin benzoate, in Sprague-Dawley rats. NeurotoxicolTeratol 19(4):315–326 (1997).

15. Fox DA. Sensory system alterations following occupationalexposure to chemicals. In: Occupational Neurotoxicology (CostaLG, Manzo L, eds). Boca Raton, FL:CRC Press, 1998;169–184.

16. Morata TC, Dunn DD. Occuptional Medicine: State of ArtReviews. Philadelphia:Henley and Belfus, 1995.

17. Anger WK, Johnson BL. Chemical affecting behavior. NeurotoxInd Comm Chem 1:51–148 (1985).



Cory-Slechta et al.


18. Crofton KM, Sheets LP. Evaluation of sensory system functionusing reflex modification of the startle response. J Am CollToxicol 8:199–211 (1989).

19. Maurissen JPJ. Neurobehavioral methods for the evaluation ofsensory functions. In: Neurotoxicology: Approaches andMethods (Chang L, Slikker W, eds). New York:Academic Press,1995;239–264.

20. Willott JF. Anatomic and physiologic aging: a behavioral neuro-science perspective. J Am Acad Audiol 7:141–151 (1996).

21. Henley CM, Rybak LP. Developmental ototoxicity. OtolaryngolClin North Am 26:857–871 (1993).

22. Henley CM, Rybak LP. Ototoxicity in developing mammals. BrainRes Rev 20:68–90 (1995).

23. Mattsson JL. Ototoxicity: an argument for evaluation of thecochlea in safety testing in animals. Toxicol Pathol 28:137–141(2000).

24. Moody DB, Stebbins WC. Detection of the effect of toxic sub-stances on the auditory system by behavioral methods. In:Neurotoxicology (Mitchell CL, ed). New York:Raven Press,1982;109–132.

25. Stebbins WC. Concerning the need for more sophisticated ani-mal models in sensory behavioral toxicology. Environ HealthPerspect 44:77–85 (1982).

26. Stebbins WC, ed. Animal Psychophysics: The Design andConduct of Sensory Experiments. New York:Appleton Century-Crofts, 1970.

27. Pryor GT, Bingham LR, Dickinson J, Rebert CS, Howd RA.Importance of schedule of exposure to hexane in causing neuro-toxicity. Neurobehav Toxicol Teratol 4:71–78 (1982).

28. Merigan WH, Barkdoll E, Maurissen JP. Acrylamide-inducedvisual impairment in primates. Toxicol Appl Pharmacol62:342–345 (2000).

29. Muller M. Developmental changes of frequency representationin the rat cochlea. Hear Res 56:1–7 (1991).

30. Rubel EW. Strategies and problems for future studies of audi-tory development. Acta Otolaryngol Suppl 421:114–128 (1985).

31. Fukumura M, Sithi N, Vorhees CV. Effects of the neurotoxin3,3’-iminodipropionitrile on acoustic startle and locomotoractivity in rats: a comparison of functional observational andautomated startle assessments. Neurotoxicol Teratol20:203–211 (1998).

32. Lochry EA. Concurrent use of behavioral/functional testing inexisting reproductive and developmental toxicity screens: prac-tical considerations. J Am Coll Toxicol 6:433–439 (1987).

33. Merigan WH. Effects of toxicants on visual systems.Neurobehav Toxicol 1(suppl 1):15–23 (1979).

34. Wood RW. Behavioral evaluation of sensory irritation evoked byammonia. Appl Pharmacol 50:157–162 (1979).

35. Maurissen JPJ, Weiss B, Davis HT. Somatosensory thresholdsin monkeys exposed to acrylamide. Toxicol Appl Pharmacol71:266–279 (1983).

36. Kelly JB, Masterton B. Auditory sensitivity of the albino rat. JComp Physiol Psychol 91:930–936 (1977).

37. Crofton KM, Ding D, Padich R, Taylor M, Henderson D. Hearingloss following developmental exposure to polychlorinatedbiphenyls: a cochlear site of action. Hear Res 144:196–204 (2000).

38. Young JS, Fechter LD. Reflex inhibition procedures for animalaudiometry: a technique for assessing ototoxicity. J Acoust SocAm 73:1686–1693 (1983).

39. Ison JR. Reflex modification as an objective test for sensoryprocessing following toxicant exposure. Neurobehav ToxicolTeratol 6:437–445 (1984).

40. Evans HL. Assessment of vision in behavioral toxicology. In:Nervous System Toxicology (Mitchell CL, ed). New York:RavenPress, 1982;81–108.

41. Fay RR. Hearing in Vertebrates: A Psychophysics Databook.Winnetka, IL:Hill-Fay Associates, 1988.

42. Gonda Z, Miklosi A, Lehotzky K. The effect of social learning ona conditioned avoidance response of rats treated prenatallywith aluminum lactate. Neurotoxicol Teratol 19:59–63 (1997).

43. Glowa JR, Hansen CT. Differences in response to an acousticstartle stimulus among forty-six rat strains. Behav Genet24:79–84 (1994).

44. Sataloff RT, Sataloff J. Hearing Loss. New York:Marcel Dekker,1993.

45. Tilson HA, Burne TA. Effects of triethyl tin on pain reactivity andneuromotor function of rats. J Toxicol Environ Health 8:317–324(1981).

46. Boyes WK, Jenkins DE, Dyer RS. Chlordimeform produces con-trast-dependent changes in visual evoked potentials of hoodedrats. Exp Neurol 89:391–407 (1985).

47. Ross JF. Tier 1 neurological assessment in regulated animalsafety studies. In: Neurotoxicology Handbook, Vol 3 (MassaroEJ, ed). Totowa, NJ:Humana Press, in press.

48. Gallistel CR. Animal cognition: the representation of space,time and number. Annu Rev Psychol 40:155–189 (1989).

49. Tchernichovski O, Golani I. A phase plane representation of ratexploratory behavior. J Neurosci Methods 62(1–2):21–27 (1995).

50. Mullenix PJ. The computer pattern recognition system for studyof spontaneous behavior of rats. In: Motor Activity andMovement Disorders: Research Issues and Applications(Sanberg PR, Ossenkopp KP, Kavaliers, eds). Totowa,NJ:Humana Press, 1996;141–174.

51. Paulus MP, Geyer MA, Gold LH, Mandell AJ. Application ofentropy measures derived form the ergodic theory of dynamicalsystems to rat locomotor behavior. Proc Natl Acad Sci USA87:723–727 (1990).

52. Dai H, Carey RJ. A new method to quanitify behavioral atten-tion to a stimulus object in a modified open-field. J NeurosciMethods 53:29–34 (1994).

53. File S. Pharmacological manipulations of responses to noveltyand their habituation. In: Theory in Psychopharmacology(Cooper SJ, ed). New York:Academic Press, 1981;197–232.

54. MacPhail RC, Peele DB, Crofton KM. Motor activity and screen-ing for neurotoxicity. J Am Coll Toxicol 8:117–125 (1989).

55. Maurissen JPJ, Mattsson JL. Critical assessment of motoractivity as a screen for neurotoxicity. Toxicol Ind Health5:195–202 (1989).

56. Meyer OA, Tilson HA, Byrd WC, Riley MT. A method for the rou-tine assessment of fore- and hindlimb grip strength of rats andmice. Neurobehav Toxicol 1:233–236 (1979).

57. Ross JF, Handley DE, Fix AS, Lawhorn GT, Carr GJ.Quantification of the hindlimb extensor thrust response in rats.Neurotoxicol Teratol 19:405–411 (1997).

58. Edwards PM, Parker VH. A simple, sensitive, and objectivemethod for early assessment of acrylamide neuropathy in rats.Toxicol Appl Pharmacol 40:589–591 (1977).

59. U.S. EPA. U.S. Environmental Protection Agency Health EffectsGuidelines, Neurotoxicity Screening Battery. OPPTS 870.6200.Washington, DC:U.S. Environmental Protection Agency, 1998.

60. Cartmell J, Monn JA, Schoepp DD. The metabotropic glutamate2/3 receptor agonists LY354740 and LY379268 selectivelyattenuate phencyclidine versus d-amphetamine motor behav-iors in rats. J Pharmacol Exp Ther 291:161–170 (1999).

61. Altman J, Sudarshan K. Postnatal development of locomotion inthe laboratory rat. Anim Behav 23:896–920 (1975).

62. Fox WM. Reflex-ontogeny and behavior development of themouse. Anim Behav 13:234–241 (1965).

63. Stanford JA, Fowler SC. Subchronic effects of clozapine andhaloperidol on rats’ forelimb force and duration during press-while-licking-task. Psychopharmacology (Berl) 130(3):249–253(1997).

64. Eilam D, Golani I. Home base behavior of rats (Rattus norvegi-cus ) exploring a novel environment. Behav Brain Res34(3):199–211 (1989).

65. Tchernichovski O, Golani I. A phase plane representation of ratexploratory behavior. J Neurosci Methods 62:21–27 (1995).

66. Archer J. Tests for emotionality in rats and mice: a review.Anim Behav 21:205–235 (1973).

67. Robbins TW. A critique of the methods available for the mea-surement of spontaneous motor activity. In: Principles ofBehavioral Pharmacology. New York:Plenum Press, 1977.

68. Walsh RN, Cummins RA. The open-field test: a critical review.Psychol Bull 83:482–504 (1976).

69. Etienne AS, Maurer R, Seguinot V. Path integration in mammalsand its interaction with visual landmarks. J Exp Biol 199(Pt1):201–209 (1996).

70. Whishaw IQ, Brooks BL. Calibrating space: exploration is impor-tant for allothetic and idiothetic navigation. Hippocampus9:659–667 (1999).

71. Hine B. Differential open field reactivity in HAS and LAS rats.Physiol Behav 57(2):301–306 (1995).

72. Treit D, Fundytus M. Thigmotaxis as a test for anxiolytic activityin rats. Pharmacol Biochem Behav 31(4):959–962 (1988).

73. Schwarting RKW, Fornaguera J, Hutson JP. Automated video-image analysis of behavioral asymmetries. In: Motor Activityand Movement Disorders: Research Issues and Applications.Totowa, NJ:Humana Press, 1996.

74. Ross JF. ECOs, FOBs and UFOs. Making sense of observationaldata. Toxicol Pathol 28:132–136 (2000).

75. Sanberg PR, Zoloty SA, Willis R, Ticarich CD, Rhoads K, NagyRP, Mitchell SG, Laforest AR, Jenks JA, Harkabus LJ, et al.Digiscan activity: automated measurement of thigmotactic andstereotypic behavior in rats. Pharmacol Biochem Behav 27(3)569–572 (1987).

76. Geyer MA, Russo PV, Masten VL. Multivariate assessment oflocomotor behavior: pharmacological and behavioral analyses.Pharmacol Biochem Behav 25:277–288 (1986).

77. Kernan WJ Jr, Mullenix PJ, Hopper DL. Time structure analysisof behavioral acts using a computer pattern recognition system.Pharmacol Biochem Behav 34(4):863–869 (1989).

78. Paulus MP, Geyer MA. A temporal and spatial scaling hypothe-sis for the behavioral effects of psychostimulants.Psychopharmacology (Berl) 104(1):6–16 (1991).

79. Paulus MP, Callaway CW, Geyer MA. Quantitative assessmentof the microstructure of rat behavior. II: Distinctive effects ofdopamine releasers and uptake inhibitors. Psychopharmacology(Berl) 113:187–198 (1993).

80. Paulus MP, Geyer MA. Quantitative assessment of themicrostructure of rat behavior. I: The extension of the scalinghypothesis. Psychopharmacology (Berl) 113:177–186 (1993).

81. Young MS, Young CW, Li YC. A combined system for measuringanimal motion activities. J Neurosci Methods 95:55–63 (2000).

82. Cory-Slechta DA. Bridging experimental animal and humanbehavioral toxicology studies. In: Behavioral Measures ofNeurotoxicity (Russel RW, Flattau PE, Pope AM, eds).Washington, DC:National Academy Press, 1990;137–158.

83. Carson TL, VanGelder GA, Karas GC, Buck WB. Slowed learningin lambs prenatally exposed to lead. Arch Environ Health29:154–156 (1974).

84. Wood R, Rees D, Laties V. Behavioral effects of toluene aremodulated by stimulus control. Toxicol Appl Pharmacol68:462–472 (1983).

85. Neitz J, Jacobs G. Reexamination of spectral mechanisms inthe rat (Rattus norvegicus). J Comp Psychol 100(1):21–29(1986).

86. Hastings L. Sensory neurotoxicology: use of the olfactory sys-tem in the assessment of toxicity. Neurotoxicol Teratol12:455–459 (1990).

87. Chen JS, Amsel A. Retention under changed-reward conditionsof persistence learned by infant rats. Dev Psychobiol 13:469-480 (1980).

88. Dominguez HD, Bocco G, Chotro MG, Spear NE, Molina JC.Operant responding controlled by milk or milk contaminatedwith alcohol as positive reinforcers in infant rats. PharmacolBiochem Behav 44:403–409 (1993).

89. Lobaugh NJ, Greene PL, Grant M, Nick T, Amsel A. Patterned(single) alternation in infant rats after combined or separatelesions of hippocampus and amygdala. Behav Neurosci103:1159–1167 (1989).

90. Moye TB, Vanderryn J. Physostigmine accelerates the develop-ment of associative memory processes in the infant rat.Psychopharmacology 95:401–406 (1988).

91. Hughes C, Plumet MH, Leboyer M. Towards a cognitive pheno-type for autism: increased prevalence of executive dysfunctionand superior spatial span amongst siblings of children withautism. J Child Psychol Psychiatry 40:705–718 (1999).

92. Kempton S, Vance A, Maruff P, Luk E, Costin J, Pantelis C.Executive function and attention deficit hyperactivity disorder:stimulant medication and better executive function perfor-mance in children. Psychol Med 29:527–538 (1999).

93. Luciana M, Nelson CA. The functional emergence of pre-frontally-guided working memory systems in four-to-eight yearold children. Neuropsychologia 36:273–293 (1998).

94. Brooks AI, Cory-Slechta DA, Federoff HJ. Repeated acquisitionand performance chamber for mice: a method to assess spatiallearning and memory. Neurobiol Learn Mem 74(3):241–258(2000).

95. Kim C, Nakai K, Kasanuma Y, Satoh H. Comparison of neuro-behavioral changes in the three inbred strains of mice prena-tally exposed to methylmercury. Neurotoxicol Teratol22(3):397–403 (2000).

96. Kuhlmann AC, McGlothan JL, Guilarte TR. Developmental leadexposure causes spatial learning deficits in adult rats. NeurosciLett 233:101–104 (1997).

97. Cohn J, Cox C, Cory-Slechta D. The effects of lead exposure onlearning in a multiple repeated acquisition and performanceschedule. Neurotoxicology 14:329–346 (1993).

98. Kelly TH, Foltin RW, Fischman MW. The effects of repeatedamphetamine exposure on multiple measures of human behav-ior. Pharmacol Biochem Behav 38:417–426 (1991).

99. Moerschbaecher JM, Thompson DM, Winsauer PJ. Effects ofopioids and phencyclidine in combination with naltrexone onthe acquisition and performance of response sequences in mon-keys. Pharmacol Biochem Behav 22:1061–1069 (1985).

100. Peele DB, Allison SD, Crofton KM. Learning and memorydeficits in rats following exposure to 3,3´-imidodipropionitrile.Toxicol Appl Pharmacol 105:321–332 (1990).

101. Cory-Slechta DA. Unpublished observations (2000).102. Delville Y. Exposure to lead during development alters aggres-

sive behavior in golden hamsters. Neurotoxicol Teratol21:445–449 (1999).



103. Petruzzi S, Chiarotti F, Alleva E, Laviola G. Limited changes ofmouse maternal care after prenatal oxazepam: dissociationfrom pup-related stimulus perception. Psychopharmacology(Berl) 122(1):58–65 (1995).

104. Miczek KA, Weerts E, Haney M, Tidey J. Neurobiological mech-anisms controlling aggression: preclinical developments forpharmacotherapeutic interventions. Neurosci Biobehav Rev18:97–110 (1994).

105. Lonstein JS, Wagner CK, De Vries GJ. Comparison of the “nurs-ing” and other parental behaviors of nulliparous and lactatingfemale rats. Horm Behav 36:242–251 (1999).

106. Stern JM. Maternal behavior: sensory, hormonal and neuraldeterminants. In: Psychoneuroendocrinology (Brush FR, LevineS, eds). New York:Academic Press, 1989;105–226.

107. Lonstein JS, Stern JM. Site and behavioral specificity of peri-aqueductal gray lesions on postpartum sexual, maternal, andaggressive behaviors in rats. Brain Res 804(1):21–35 (1998).

108. Vernotica EM, Rosenblatt JS, Morrell JI. Microinfusion ofcocaine into the medial preoptic area or nucleus accumbenstransiently impairs maternal behavior in the rat. Behav Neurosci113:377–390 (1999).

109. Lee A, Li M, Watchus J, Fleming AS. Neuroanatomical basis ofmaternal memory in postpartum rats: selective role for thenucleus accumbens. Behav Neurosci 113(3):523–538 (2000).

110. Scott JP. Animal Behavior. Chicago:University of Chicago Press,1972.

111. Miczek KA, Winslow JT, DeBold JF. Heightened aggressivebehavior by animals interacting with alcohol-treated con-specifics: studies with mice, rats, and squirrel monkeys.Pharmacol Biochem Behav 20:349–353 (1984).

112. Palanza P, Parmigiani S, Liu H, vom Saal FS. Prenatal exposureto low doses of the estrogenic chemicals diethylstilbestrol ando,p’-DDT alters aggressive behavior of male and female housemice. Pharmacol Biochem Behav 64(4):665–672 (1999).

113. Fiore M, Petruzzi S, Dell’Omo G, Alleva E. Prenatal sulfur diox-ide exposure induces changes in the behavior of adult maleagonistic encounters. Neurotoxicol Teratol 20:543–548 (1998).

114. Paredes RG, Vazquez B. What do female rats like about sex?Paced mating. Behav Brain Res 105:117–127 (1999).

115. Pfaus JG, Smith WJ, Coopersmith CB. Appetitive and consum-matory sexual behaviors of female rats in bilevel chambers. I: Acorrelational and factor analysis and the effects of ovarian hor-mones. Horm Behav 35:224–240 (1999).

116. Mably TA, Moore RW, Goy RW, Peterson RE. In utero and lacta-tional exposure of male rats to 2,3,7,8-tetrachlorodibenzo-p-dioxin. 2. Effects on sexual behavior and the regulation ofluteinizing hormone secretion in adulthood. Toxicol ApplPharmacol 114:108–117 (1992).

117. Hull EM, Lorrain DS, Du J, Matuszewich L, Lumley LA.Hormone-neurotransmitter interactions in the control of sexualbehavior. Behav Brain Res 105:105–106 (1999).

118. Amstislavsky SY, Amstislavskaya TG, Eroschenko VP.Methoxychlor given in the periimplantation period blocks sexualarousal in male mice. Reprod Toxicol 13:405–411 (1999).

119. Mendelson SD, Pfaus JG. Level searching: a new assay of sex-ual motivation in the male rat. Physiol Behav 45:337–341 (1989).

120. Cohn J, MacPhail RC. Ethological and experimental approachesto behavior analysis: implications for ecotoxicology. EnvironHealth Perspect 104(2):299–305 (1996).

121. Needleman HL, Reiss JA, Tobin MJ, Biesecker GE, GreenhouseJB. Bone lead levels and delinquent behavior. JAMA 275:363-369 (1996).

122. Spencer PS, Schaumburg HH, eds. Experimental and ClinicalNeurotoxicology. New York:Oxford University Press, 2000.

123. Smith SA. Disorders of the autonomic nervous system. In: ThePractice of Pediatric Neurology (Swaiman KF, Wright FS, eds).St. Louis, MO:C.V. Mosby, 1975;799–828.

124. Oribe E, Appenzeller O. Autonomic nervous system. In: Guide toClinical Neurology (Mohr JP, Gautier JC, eds). NewYork:Churchill Livingstone, 1995;521–540.

125. Gordon CJ, Heath JE. Integration and central processing in tem-perature regulation. Annu Rev Physiol 48:595–612 (1986).

126. Gordon CJ. Thermal biology of the laboratory rat. Physiol Behav47:963–991 (1990).

127. Watkinson WP, Gordon CJ. Caveats regarding the use of thelaboratory rat as a model for acute toxicological studies: modu-lation of the toxic response via physiological and behavioralmechanisms. Toxicology 81:15–31 (1993).

128. Bowyer JF, Davies DL, Schmued L, Broening HW, Newport GD,Slikker WJ, Holson RR. Further studies of the role of hyperther-mia in methamphetamine neurotoxicity. J Pharmacol Exp Ther268:1571–1580 (1994).

129. Gordon CJ, Yang Y, Gray LE Jr. Autonomic and behavioral ther-moregulation in golden hamsters exposed perinatally to dioxin.Toxicol Appl Pharmacol 137:120–125 (1996).

130. Gordon CJ, Miller DB. Thermoregulation in rats exposed peri-natally to dioxin: core temperature stability to altered ambienttemperature, behavioral thermoregulation, and febrileresponse to lipopolysaccharide. J Toxicol Environ Health54:647–662 (1998).

131. Rusak B, Zucker I. Biological rhythms and animal behavior.Annu Rev Psychol 27:137–171 (1975).

132. Evans HL. Behaviors in the home cage reveal toxicity: recentfindings and proposals for the future. J Am Coll Toxicol 8:35–52(1989).

133. MacPhail RC, Crofton KM, Reiter LW. Use of environmentalchallenges in behavioral toxicology. Fed Proc 42:3196–3200(1983).

134. Evans HL. Neurotoxicity expressed in naturally occurring behav-ior. In: Neurobehaviorial Toxicity: Analysis and Interpretation(Weiss B, O’Donoghue J, eds). New York:Raven Press,1994;111–135.

135. Crofton KM, MacPhail RC. Reliability of motor activity assess-ments. In: Measuring Movement and Locomotion: FromInvertebrates to Humans (Ossenkopp KP, Kavaliers M, SanbergRG, eds). Austin, TX:Landes, 1996;227–252.

136. Wang GH. The relation between “spontaneous” activity andoestrous cycle in the white rat. Comp Psychol Monogr 6:2(1923).

137. Norton S, Culver B, Mullenix P. Measurement of the effects ofdrugs on activity of permanent groups of rats. PsychopharmacolCommun 1(2):131–138 (1975).

138. Wolfe G, Bousquet W, Schnell R. Circadian variations inresponse to amphetamine and chlorpromazine in the rat.Commun Pharmacol 1:29–37 (1977).



Documents

Methods to Identify and Characterize Developmental … · 2017-10-15 · behavioral tests are easily accomplished. To the contrary, as in any area of science, exper-tise and training