Walsh, RN., and Rk Cummim The open-field test A critical review. P s v c h o l o e i c a l Bulletin ' . - 83:482-504, 1976.

: + - . . . . P ' ) r b b ' d B&lb 1976. ~oy81, KO. 3. 482-IM

The Open-Field Test: A Critical Review

Roger N. Walsh Robert A. Cummins Department of Psych+ and ~ e b a o k r a ~ Department of Physiology, Llnioersify .c$

S&nces, Stanford Unioersity Queensland, Brisbane. Australia

Over the w t 4 0 years Lhc open field bar ~ o l v e d a a commonly used tool for the rnmumcnt of animal behavior. Thh m i e w taka a dtical look at the we of I& instrument, u&y with regard to the development of a standard form for iu w. The variou procedura and .their sbortcomingr are 'dkmued, - w i t h partimlar rcfupacc to the seemingly incooseguential detdk which have b.& showb to ioduhte open-field prformancc per w. Depndent paramrerr are considered both vith regard to their reliability and their valid- ity for the measurement of such undul,ing mnstrucb as emotionality.

Since its introduction some 4 0 years ago. the open-field test bas attained the status of one of the most widely used instruments in animal psychology. Its popularity probably stems in large part' from the simplicity of the

apparatus, tbe-easy &d rapid meaurement o f dearly defined behaviors, and a generally accepted interpretation of these behaviors. I n . addition, certain o l tbe measured behaviors are sensitive to a a ide range of genetic! ex- periential, physiological, and pharmacological manipulations and are sufficiently reliable under standardized conditions t'o give repeat- able measures on an enormous range of in- dependent variables. Simplicity, ease of quan: tification, and a ide applicability are therefore probably the prime determinants of its popularily.

However, in spite of i ts 'sbtus as one of the most widely used tests of animal behavior, the open-field test bas survived for 40 years uilh only one major review (Archer, 1973), and this was limited to a consideration of its usefulness for sl imatinn emotionalitr in ro-

diversified enormously, yet in the literature, sweeping generalimtions and conflicting inter- pretations continue to he made on the basis of univariate studies. This paper aims a t de- tailing the many variables involved in testing and their influences on dependent param- eters, as well as methods of analysis and interpretation of results. We thereby hope to gain an overview o i this area of study! to point out shortcomings of some present work, and LO derive implications for future research.

The originator of the open field was Calvin Hall (Hall 1934 ; Hall & Ballechey 1932), who used d e f a t i o n in the open field as an index of timidity. However, time bas seen a proliieration of other dependent variables. Some 30 have been used a t one stage or an- other:. and the number of underlying con- structs they supposedly measure bas multi- plied accordingly. I n niany cases, however. proof of construct validity has been a d l y lacking.

IV'hat then is the open field and what, in general terms: does it really measure? I n -

dents. Apparatus, techniques, subjects, pa- essence, the test consists of the measurement

rameters, analyses, and interpretations .have p i behaviors elicited by placing the subject In a novel open space from which escape is prevented by a surrounding wall.

We would like to exDrps our ~ratitudc to Otto T h e elicitation of these behaviors is de- Budlz-Olsen, Bun ~icdelman, TO& Horvalh, and pendent upon the interaction of the animal Alice LH for their encouragement, advice, and with a variety of test factors such (a)- criticism. Our uperimeotal work and prepration of be manulc,jpt were supporkd from -stimulation z a result of removal from a Australian Rwarch Grants Committee and t h e familnr home environment; (b ) stimulation Foundatton Fund for Rcvarch in Psychiatry. involved in transferring the animal to the

Rcqursu for reprints should be x n t to Robert A. cumrnins, D~~~~~~~~~ of Psycholow, Univerrlt,. of Open field; !c) m ~ o s u r e to the test environ- \Vutero .4ustralia, Nedlands, Perth 6039, ~ust~al ia . ment, consisting of both the open field itself

. OPEN-FIELD TEST 4S3

and its surroundings; and (d ) aU prior ei- perience of the t e t situation. In e&'ect, tbis last factor means that one is measuring,. among other things, habituationand learning in response to the test environment. Where several test s q i o n i are employed, further factors to be' considered indude previous stimulalion involved in removing the animal from. the. testing situation, transfer and re- exposure !o the home environment a t the completion of prior sesions, and modification of h o m e a g e pehavior .and interaction- with other home environment animals.

The magnitude of any particular behavior elicited will therefore be a function of the multiway interaction oi these factors. .4s yet their relative importance is almost completely unknown,.but thereis evidence to suggest that each factor exerts a differential elfect on animals oi varying genetic and experiential backgrounds, and this is examined later.

Effectively 3nv behavioral experimentmea- sures responses to the above fattors. How- ever, in many-studies, such as maze learning;. one hope3 that the subject will habituateto, and hence be minimally influenced by, aspects

. of thetes: siruation other than the specific stimulus c~lmponent being u s e d % the inde- pendent variable. In - the open-field test, on the other hand, the whole test situation (rather than any one specific stimulus com- ponent) is the independent variable, and.by its very nature this independent variable must be multifactorial. These factors are now con- sidered under the headings of Techniques and Apparatcs Characteristics, Testing Environ- ment, and Procedural Details:

The difficulty of finding comparable tech- niques in the use o f the open field has been mentioned previously. Almost every physical characteristic of the apparatus; i t i surround- in&, a n d every s!ep have been widely varied, so that although standardin- Lion may have been established aithin in- dividual laboratories, there is a disturbing

~ o o k i & a t the range of techniques em- ployed. one finds ' tbat for the apparatus, its size, shape, color,. subdivision, wall height, Boor texture, odor, nature and locauon.of the starting arm, and presence or absence and .

nature o i additional inherent stimuli have all been varied; while for the surroundings the nature, intensity and position of light, sound andodor sources, and the visibility and posi- tion of the observer have- suffered similarly. In fact it is hard to think of any facet which has not been modified. This .diMculty of standardization b compounded by the ex- 1rerre:rarity of reports which cite details of more than a smdl proportion .of relevant procedural variables.

Ph?sical .Characteristics o j the Apparatus

The most common shape of the open field has been circular, but square and rectangular designs are also prevalent. The wide range of shapes leads eventually to the quejtion. \i:hen is an open field not an open field? The ex- treme of the rectangular design is usually termed a straightaway or a rur&a.v, yet the behaviors measured may be akin to those employed in the open. f idd (e.g., Poley & Royce, 1970; Zimhardo & Ilontgomer)., 1 9 5 7 ) . Perhaps in view of the paucity of ei i- dence on the comparability of behavior ob- served in these devices and in the open field, they are best regarded separately. As ,yet there would seem to be no data available on the effect o i open-field shape on behavior, hut in view oi the evidence on strain difierences in thigmota<s (or "wall hugging," Valle, 19i0) and the tendency of some animals to remain in corners, standardization would seem a sensible precaution.

Difierent s ~ e c i e j have naturdlv been al- located appropriately sized open fields, such

a whole room for human infants (fiein- gold, 1969) and a .5-m square field fo r mice (hlanosevitz. 1970). More importantly, how- ever, open-field size has -commonly been varied for intraspedes studies, and the efiects of such dimensional changes have been shown to exert a significant &ect on some aspects oi behavior. .4s a ~ e n e r a l rule. ambulation -

lack oi conformity in procedure and resulu urould seem to be more susceptible to change within theliterature as a whole. in apparatus a n d environment than would

, . '484 , ' , . ROGER K. WALSH ASD ROBERT A. COMbfINS

1 defecation: Thus, while defecation is largely unaflected by arena size in rats (Broadhurst, 1957, 1958a, 1958b), gerbils (Oldham & Morlock; 1970), but not male WH mice (Nagy & For&, 1970), ambulation increases with increasing field size in rats (Broadhurst. 1957; Montgomery, 1951), mice (Blizard; 1971 ; Krsiak & J d u , 1971 ; Nagy & Forest, 1970); and gerbils (Oldham .& Morlock; 1970). Large field size has been reported to produce a , disproportionately large increae in ambulation under conditions of low illum- ination (Blizard, 1971):

An .example of the dacul t ies .of compari- son where different apparatuses are used is afIorded by ~ l a r i , Gorman, and Vernadakk (1970), wbo tested ambulation of the off- spribg of mothers injected with impbetamine or chlorpromazine. Treatment effects were found on Days 13-21 but not on Days 46, and 60, and the results were interpreted ar

. . being due to differential maturation rat& (an Age x Treatment' interaction hypothesis). However, since two different field k w e r e used for the infant and adolescent animals, an interaction between treatment and field

. size was also possible. Color bas ranged through white (Oldham

&'3Torlock, 1970) to black (Delbarre, Durn=, & duionniere, 1970), and the material com-

- ' posing the field, particularly the Boor, has included wood, metal, cqncrete, rubber, and even glass (Satinder, 1968). There is almost no evidence as yet on the eEect of these vari- ables on behavior, but in view of the evidence for the effects of similarity of test envimn- ment lo rearing environment on exploratory behavior (Nielson, 1970; Wells, Lowe, Shel- don, & .Williams, 1969), some of these vari- ables may be of particular relevance when testing animals with differential rearing histories.

Klost commonly, the open field has been bare but occasionally experimenters have added objects, usually with the aim of obtain- ing a measure of interaction with these addi- tional stimuli. The number of these objects has ranped from 2 (Ehrlich & Burns, 1958) to 15 (Furchtgott, Wechkin, & Dees. 1961) a n d t h e y have usually been small solid ob-'' jec&, such as nuts and balk, door stops, and

mirrors. However, the more exotic have in- cluded caged rats, f ld ing ' l i gh t s and loud- speakers (Ehrlich & Bums, 1958; Fox & Spencer, 1969), food pellets (Jarrard 9. Brunnel, 1968), designs painted on the floor and walls (Wimer & Sterns, 1964); and Rheingold's (1969) study of human infants induded toys and the .subject's mother or another person(basicaUy a social facilitation study). The relevant dependant paramiters are most commonly the number of approaches and time spent invejtigating the objects, and sometimes preference scores for particular objects.

THE TESTING E X ~ O N M E N T

Characteristics not o n l y of the open field itself but also of its surroundings infringe upon the subjects and modify behavior.

The level of illumination has been the variable most closely examined; but in spite of repeatedly demonstrated eElects, . f&er than one third of the articles mention the test lighting used. With the exception of Candland and Nag?r (1969), who reported Lhe reverse, i t bas consistently been found tbat high levels of illumination are associated a i t h diminished locomotor. behavior. Valle (1970) found that ambulation, rearing, and thigmotaxis were all less under higher illumination.

A number of interactions of illumination level a i t h other independent variables bare been reported. Thus, L i v w y and Egger (1970) found tbat ambulation was less under an illuminance of 9.3 lx than 0.005 lx in rats a t 47 or 115 days of age but not a t 18 or 24 days. Testing C57 BL/6J mice, Nagy and Glaser (1970) found no main effect for il- lumination (9.3 k versus 0.09 k) but found si~nificant interactions between level of illumination, age of subjects, and day of test- ing. Under high illumination, there were no reliable activity differences due lo age, while under low levels, IO&day-old subjects were more active than 5&day+lds. The subjects were less active under high than under low illumination on Days 1 and 2 of testing but not thereafter, suggesting that for the C57 BL/6J strain, illumination effects are largely transitory. Interactions with strain have been reported by McClearn (1960) and by Dixon

OPES-FIELD TEST 483

and Defries (1968b), who also found albinos to he more photosensitive than pigmented rats. Interaction witb. prior visual rearing environment (dark, alternating light-dark, or light in black, white, or patterned cages) was noted by Sachett (1967).

Another commonly studied dependent pa; rameter of. illumination has been defecation, and Rith. some exceptions (Blizard, 1971); no significant main.e8ect has been reported (Kagy & Forest, 1970; Nagy & Glaser, 1970; Nag) 8: Holm, 1970). The result seems un- expected in terms of much current thinking which considers high illumination as "stress- ful" and defecation as the prime index of emotionality.

A dijadvantage of the available studies of illumination is that they include only two, or a t the-most three levels, so that the precise relationship with the magnitude01 the ob- served behavior is quite unlmoivn.

Less study has been made of ambient noise and few papers refer to it. Presumably. Lhere- fore, in most cases it comprises merely the' background noise oi the test room, but white noise, most commonly of about SO dB (e.g., Eysenck & Broadhurst, 1964; Ivinskis, 196S), has sometimes been superimposed. Its addi- tion h s usually reduced locomotion (e.g., Bindra & Spinner, : l9jS), but L i v e ) , i n d Egger (1970)' reported that white noise of 90 dB increased both ambulation and de- fecation. Certainly any abrupt loud noise can markedly inhibit locomotion a n d even induce prolonged immobility in a variety of species (Hofer, 1970; Cummins, Walsh: & Budtz- Olsen, Kote 1).

Environmental odors have largely been ig- nored. although several experimenters have washed thei r apparatus. with a vaiiety of agents in order to obviate possible biasing efiects of odor trials le f t by previous subjects. \\%ittier and McReynolds (1965) found this to be asigni&ant effect for the test behavior of mice. However, Satinder (1969) found that the time spent by a rat on the side of the open field which had been occupied by i t s predecessor was significant only for males preceded hy females. Interestingly; urination or defecation by the,predeceswr was without effect.

McCall (1969) has demonstrated tbe im- portance of odor for what he termed the "caretaker efiect." I t was found that rats tended to spend more time on the side of the field nearest to the caretaker, who had reared them, than on the side opposite where there waj another experimenter, who had no pre- vious experience with them. This effect was. shown to b e dependent on olfactory rather than %%ual cues, The visibility and behavior 01 the experimenter during testing have rarely been mentioned. Presumably, therefore, in most test situations he is visible to the sub- ject except where obsenration is frombehind a screen (e.g., Nielson, 1970) or a one-way mirror (e.g., Fox 8: Spencer, 1969). This hiding would seem a wise precaution, since although no precise studies of the effects n i experimenter visibility are available, it i i quite obvious that any visible abrupt move- ment may cause inhibition of ongoin. ac: =;.. ~ i v i t y and in some u s e s prolonged immobtl~ty (Cummins e t a]., Note I ) .

PkioCEDurw~ DETAILS

As is obvious from the foregoing discussion, careful measurement can detect changesin behabior due to very small changes in the apparatus. I t is therefore regrettable that there has beenalmost no study of the efiects of variation ' in testing procedure. Testing normally consists of removing the animal from its home environment, carrying it to the test area, and placing it in the open field. The carrying can be done either by hand or via a start box. Testing itself may vary on a num-

.ber of temporal factors, such as the duration of exposure. to the open field and t h e inter- trial interval. I t would be surprising if changes in these procedures did not measur- ably aRect open-field behavior.

Details o i the animal's removal from the home cage and transfer to the open field are seldom reported, so that presumably they are usually transported by hand. However, some workers aiming to reduce handling stress have employed transport compartments, such as a box (Nielson: 1971: Tighe, 1965), tube ( D i ~ o n & Defries, 1968a), or transparent cup (Hofer. 1970): In some a e s , the subjects have been habituated to a transport compart-

456 ROGER x. \VALSH AKD ROBERT A: C U ~ ~ ~ I I X S

I.'B&>~o;

k Whole m mnjor body mmcmorl 1. Type of movement

a. Distance covered per unit time b. Time spent in ambulntioo c. Wesring frequency d: Escape attempts e. Laleocy (uwdly time taken lo l a v e start area) i. Time spentaithout movement

2. Loc1tions a. Field area viGtnl (inner or peripheral a r m .

mmm, elr.) ' 4. .AMliation (distance from p n n e r subject)

c. Stimulusintetxtion (e.g., dirtan& irom stimulus abject)

B. Par1 body nromnmr ' . a. \lanipulatius of objects b. S S m g c. Smtchinc d: Digging - e. Teeth chattering I. Grooming g. vocalimtion h. Visual erplomtion

11. Autonomic nervnus %stem 'a. Defecation b. Digestive transit time

. c. Urination. d. Hurt tale a d rhythm

-e. X s p i r s t ~ r ? rstc

1U. Adrend activity ' . a, Adrenal ascokbic sdd

b. Serum corticosteroidr

W. E3m~0pbysiolo~

3. Hippocampal theta utivity b. Elccvomyogmm activie

ment and carrying procedure prior to testing (e.g., Nielson, 1970). Such familiarization may increase the levels of ambulation a t the start o f an open.-field trial (Abel, 1971).

Once ' traEported, the animals may be placed either in the center of the open field (e.g., Clark e t al., 1970) or against the re- taining wall (6ing, 1970). I n this context i t is interesting 1o'note.Satinder's (1969) find- 'ing that rats tehd to remain on that side of the field on which they are originalls placed.

Jndividual trial durations are usually short (Krsiak & Janku,-19i l ) , but trials of 1 hour

have been employed by Block and Essman (1965). Many workers measure only total behavior lor e3cb trial (intratrial time unit equals trial length), thus relinquishing any possibility of studying intratrial temporal patterns of behavior. Some go a staee further, mesuring only total behavior over all trials. These practices sacrifice a considerable amount of available information, and their disadvanta,nes are detailed i n the d i s m ' o n on temporal analysis of behavior.. Many a- perimenters employ only one trial: a pro- cedure of doubtful value in view of the dif- ficulties of interpretation indicated by Wlini- bey and Denenberg (1967), who found a chanye in factor loading of ambulation from Day 1 to Day Z..3.fultiple trials avoid this criticism and afford greater reliability and the opportunity of temporal analysis. Those who give several trials seldom use more than 4, but up to 6 0 h a v e been used (Bronstein, 1972). \\'here multiple trials. are given, the question of the optimal intertrial internal arises; but most commonly, 24 hours has been chosen, partly from convenience hut a h since this obviates complications of diurnal rhythm.

Almost no study seems to have been per- lornied on .theefiects of variing these tem- poral factors. However, Battig (1969) found that patterns of activity were undected by exposure to the apparatus 2 hours previous to the experiment, but that intertrial intervals o i 3 and 12 minules produced a mild inhibi- tion of ambulation on subsequent testing.

DEPENDEXT ' P A R A ~ ~ R S

The number of dependent variables used to aswy open-field behavior has grown sub- stantially since the early work of Hall (1934). until now over 50 are listed. This list (see Table ' 1 ) is mostly compiised of behavioral measures, although recently some physiolog-

'ical parameters, such as heart and respiration rates! electrom~ogram recordings, and plasma steroid levels,. have also been included. .k can be seen from Table 1 most param-

eters have been varieties of motor behavior, and of these, measures of ambulation have been most favored. This is presumably be- cause of the ease of quantification and the evident face validity for interpretive con-

0PES.FIELD TEST 487

s t rucb of exploration and arousal. Amount of ambulation has usually been scored spati- ally by the number of subdivisions entered, but has also been estimated temporally by the proportion of time spent inambulation. I t has also been quantified according to cbaracteris- tics 01 its apparent aim, locality, and latency. Thus, ambulatiob directed toward escape has been measured by Ehrlich and Burns (195S), who scored an escape attempt i f the subject got its head and forefeet over the wall 01 the field.

Ambulation has also been scored according to its locality, the most frequently .used regional divisions being either inner or periph- eral areas of the apparatus (Ebrlich & Burns, 1953; Ivinskis, 196s). Occupancy of the peripheral a r m , either in corners ( h l o r r k n 8. Thatcher, 1969) or near walls (Valle, 1970);has been used as ai index of timidity.

Locality of ambulation has also been used as a measure of interaction with either in- animate objects or other subjects. Fox and Spencer (1969) measured object interaction by the frequency and duration of activity occurring within the same square as the ob- ject. In addition to these measures, McCall, Lester, and Dolan, (1969) noted any qualita- tive difierences in the animals' responses to the objects. The-w included the approach be- havior t o the object and the exploration oi the object itieli. Where conspecifics have been used instead of inanimate objech, mea- sures of "affiliation" (Latank, Cappell, & Joy, 1970) hare been obtained by scoring the average distance between the animals and the time spent in direct contact with one another.

Activity can also be scored by it; absence. Here the two major parameters are idenrified by.the terms latency and freezing. Laten? is measured by the time taken from the start of a trial to the occurrence of a certain tvoe of <.

behavior. Theoretially, h e latency of an arbitrary amount of any dependent param- eter might be used, but open-field studies have mercifull!, confined themselves to a rela- tively m a l l number, usually involving am- bulation. Most commonly, latency to leave the start area has been used (I~insb-is, 1363; Poley & Royce, 1970). However, laten? to

position (King, 1970) and laten? to reach the periphery from the center (Tobach, 1966) have also been measured.

The other open-field parameter measuring the absence of activity is freezing, defined as the absence.of movement. This is a widely used parameter usually taken as indicative of a high-stres state. Certainly it is a behavioral response which occurs across the phylogenetic scale in the iace of perceived danger; elements such as abrupt change,. environmental dis- location, physical restraint, and presence of predators are patticularly potent eliciting fac- tors. In the open field? freezing in response to stimuli such as the above shows a marked species specificity and has been recorded as sometimes lasting for over 1 hour. I t has also been found to be associated with marked tachpnoea, bradycardia, and vagotonic cardiac arrhytbrnis (Hofer, 1970).

Another cause of prolonged immobility is of course sleep, and the not infrequent occur- rence of apparent sleep has been noted b y Fox and Spencer (1969) and Cummins e t al. (Note I ) . Fox and Spencer observed that dogs liable to apparent sleep were extremely inactive and that they would often defecate. urinate,. and shiver when placed in the start box. Once apparently asleep they were easily roused by a sudden-noise but rrould then ap- pear to promptly fall as!eep again, and the authors suggested that this might possibly represent displacement sleep as a result of intense arousal. Similar apparent sleep has been noted i n r a t s by Cummins e t al. (No:e I ) , sometimes associated withapparent shiv- ering, which might actually be a result of the marked tacbypnoea.observed in subjects dur- ing immobility (Hoier, 1970). Against the suggestion of this apparent sleep being the direct result of h>perarousal is the fact that i t s frequency increases with increasing time elapsed within and between trials. This time course is more in accordance with that o f habituation, and in point of fact, it has fre- quently been observed that habituation of an arousal reaction may be accompanied by the rapid onset of drowsinw and sleep (see Lynn, 1966, for a review). Paradoxically, a general- ized orientation reaction may be easily elicited during this drowsy phase. Subjec-

reach the center from a peripheral starting tively? it seems possible to differentiate the

immobility followingsudden threat from that of apparent s leep by such features as the abruptness of onset, wide-open eyes, and muscular rigidity in the former case, as op- posed .to the gradual development, closed eyes, and relaxed posture accompanying a p pareni sleep. Perhaps the two behaviors representexamples of the startle reaction and habituation of the arousal reaction, re- spectively.

One test of this hypothesis might be the use of electroencephalograph (EEG) recordiog? since the orientation reaction is accompanied by bippocampal theta rhythms,. whereas hip- pocampal desynchronization accompanies the startle reaction-an interesting example of electroph~siological data differentiating two behaCiorally similar responses. I n summary then, immobility i n . the open field has been regarded as indicative of high stress hut may actually confound two distinct behaviors.

Another widely used measure of activity has been rearing and bas proved a reliable (Itinskis, 1968) and valuable measure. Com- bined with ambulation it has proved to reflect a stab!e individual trait, "nonspecific excit- ability level," which has been significantly correlated with hippocampal slow. wave ac- tivity. a well as with a wriety of other be- haviors. This.measure.aljo displays systematic variations with individual differences in;so- matic functions such .as growth rate, body temperature, caloric intake per surface area, qualitative food preference, and endocrine functions ( L i t -& Gollov6-Hemon. 1969; Martinek & Lit, 1969). However, attempts other than. the above to. systematically link rearing a i t h other variables have been rare, and it is most frequently taken at face value as an index of activity.

.4s -rriight be predicted, sitling has been found to be more frequent during conditions associated with low levels of activity and to be negatively correlated with ambulation, rearing, and sniffing (Prescott, 1970). A surn- mary of the influences exerted by different es- perimental-procedures on whole body move- ment and defecation a n be seen in Table 2.

:With the exception of defecation, param- eters other than whole body movement have been less closely studied and i n many cases have been cited in very few reports. These

include scratching, digging, teeth ,chattering, respiration, electromyograph (E3,lG); diges- tive tra&t time, cardiac rhythm, and EEG measures. Of these part body movement mea- sures, snihing has been relatively common and bas generally been taken as an index of exploration. It has akw proved a . useful co- index for nonspecific excitability, although a less adequate one than either ambulation or rearing. The terms scratching and digging probably refer' to wential ly the same be- havior evinced under conditions 01 hard a n d loose flwring material, respectively. Prescott (1970) found that greater amounts of digging in the sawdust floor of the open field were associaled a i th other evidence of high- activity. states. Teeth grinding was used by Hughes (1969) as a possible indicator of anxiety on the rationale that i t mayrepresent a' '"tooth sharpening defenske behavior." Respiration and E M C were used by Hofer (1970) in her studies of prolonged immo- bility and have been mentioned previously. Grooming and wushing are terms describing tbe.same be5avior and have been found to be of relatively low reliability (.Ivin!kis? ,1968) and to be negatively related to indexes of high-activity states (Prescott, 1970). Vocal- izo!ion bas been taken as another indicttion of distress, and the vocalization emitted on placing an animal in the open field has some- times been termed distress calling (e.g., Cand- land & Kagy. 1969). Canaland and Kagy have suggested distress calling a s the major index of emotionality in species for whom the normal indexes of defecation and am- bulation are not reliable, such as he cat and domestic fowl, although alartinek and L i t (1965) found hat it was not a reliable mea- sure in dogs.

F'osihle p2ramete:s o f autononic nenrous svstem function aie legion, but in practice they have been limited to the more easily measurable ones oi excretory and cardiac iunction. Defecation, the parameter for which the oppn field was designed, remains as one. if not the most widely used, measure and remains as the prime index of emotionality, a role whose validity has been confirmed by factor-analytic studies (\\limbe? & Denen- berg, 1967). A related, but uncommon, mea- sure is digestive transit time, defined as the

OPEN-FIELD 'TEST 489

timeinterval between consumption and first appearance of a fecal staining ageni such as chromic acid. hrot surprisingly, digestive transit time and amount of defecation would appear to be negatively correlated. These two parameters have been used by Tobach (1966) to demonstrate that open-field testing elicits a significantly greater effect than the manipu-

'lative procedures involved prior to actual testing. The third excretory measure, urini- tion, has been scored by frequency, presence or absence, or amount and has occasionally been combined with defecation to yield a composite elimination -score (Tobach, 1966), but in any form has failed to yield reliable resullr in a variety of species (Ivinskis. 1968; hlanosevitz, 1970; Martinek 8: Lat, 1969).

Cardiac measures have been used litge ey- cept by Hofer (1970), whose study has pre- viously been mentioned, and by Candland and N a p (1969) in an atttempt to validate the use of the two most commonly used indexes of emotionality, irAbulationand defe-' cation. The CFE strain of rats were exposed to the open field for 21 minutes for 30 days and compared with home caged animals. Cardiac rare and defecation were greater in the open field, but whereas defecation de- clined over trials, there was no similar adap- tation of cardiac rate, although the rate did fall within irials. Furthermore, there was no significant correlation between heart rate and defecation, so that although both changed in stable fashions and were both considered as indexes of emotionality, they changed' in largely independent ways.

Recent years have b e y n to see the appli- cation of physiological techniques to the open field, a'movement which would seem to have immense potential for expanding the =ope of studies of reactions to novel environments, for behavioral-physiological correlation! and for validity testing of parameters and behav; ioral constructs. The examples of respiration and cardiac rate have prevjously been dis- cussed, which leaves biochemical and electro- physiological parameters.

The number of biochemi~al.~arameters of potential interest is boundless, but studies' have a s yet been largely restricted to assays of adrenal function. This interest stems from the long-held view of theinterrelatedness of

adrenal gland function and emotional behav- ior. However, studies have shown little rela- tionship between open-field behavior and adrenal weight or stemid output, and as yet adrenalectomy bas yielded inconsistent or negative effects in t h e r a t , mouse, and ham- ster (Fuller, Chambers, & Fuller, 1956; . Moyer, 1958; Paul 8: Ravlena, 1962). Pare and Cullen (1965) were unable to detect consistent relationships between adrenal ascor- bic acid levels and open-held behavior, and similar negative finding5 were reported by Ader (1969) for plasma and adrenal corticos- terone in hooded and albino rats. Further- more, although there was a significant effect. across trials for mbulation, defecation, and latency, there was no such change in plasma corticosterone levels smplpd 1 5 ininutes alter each test. Ader concluded that Lhe da tap ro - ~ i d e d no evidence lor a relationship between open-field behavior and adrenal function.

In contradistinctinn to this is the report by Levine, Haltmeyer, liaras, and Denenberg (1067) of greater plasma corticosteroid levels in nonhandled than-in handled.subjec& I5 minutes after testing. There was no change. however, in these corticosteroid levels across the 4 days of testing, and the immediate post- t e t steroid' levels did not differentiate Lhe two treatment p u p s . This difierence in abil- ity of the steroid levels to.differentiate the handled and nonhandled subjects a t 0 and 15 minutes after testing can probably be ex- plained in terms of the differences which Levine has consistently noted in the time course of adrenal response in these two poups. However, plasma corticosteroid levels also failed to differentiate between enriched and isolated C57BL/6J mice or to correlate

w i t h defecation or adrenal weight changes in a study by Denenberg, Wehmer, Werhoff, and f i r r o w (1969). I t would therefore seem that plasmasteroid estimates represent a po- tentially useful tool but that their relation- ship to other open-field measures is as yet unsettled. Future studies will probably re- quire = temporal analysis of the adrenal' response after testing, rather than the simpler but less informative method of sampling only immediate or maximum (appmldmately 15 minutes after testing) levels.

lndclendrnl rarlsblc Study

Increased npen- Broadburst (1957, 195Sa, 1958b) ficld size Oldham 8; hlorlock (1970)

hlontgomcry (1951) Elmadhunt (3957) h'ao 8; F o r a t (19i0) Ulizord (1971) R ~ i a k 8- Janku (1971) Oldham 8; hlcrlork (19i0)

Increared Naby & Forest (19i0) illumination Nagy & G l m r (1970)

h'ngy & Holm (19i0) Rlimrd (1971) Vnlle (1970) Livrsey 8: Eggcr (1YiO) hlcClearn (IYW) Diran 8; Eldries (1968b) Cnndland 8: Nagy (1969) NOW 8: Clarer (1970)

Environment noise \\'bite noirc Ilindm (I. Spinner (10%)

Livesev 8: Emcr (1970)

Rnt tlelecation ' 0 Gerbil Defeatinn 0 Rat hmbulation 1

i >louse hmbulation i

7 i

Gerbil :tmbulitinn f hlouse Defecation 0

0 0 1

Rot Ambulation 1 - 1 only alter 24 days 01 age

>louSe An~bulatian I albiios more than pigmented 1 albinos more Lban pignlcnted

Rnt . .

hl,rupt noiw Hofcr 119i0)- Rnt Cumrnins, \Val&, & Budtr4lsen (Sote I)

Odor \\Inittier & >lcReynoldr (1965) 3louse

!%tinder ( 1 x 0 ) ' Rot

I 1 only on Dnys 1 and 2 in 1W- but not jOdny+lds ?

(freezing) i u1

h s t i o n of f ambulatiun in n r w in which prior subjects had b m ~ u - ambulstion lnlrd

:k s b v e i u t ~ignificant only lor maln preceded by femnles

"Caretaker ellec1"-time spcnl on side of o l rn field nearest to P familiar caretaker

Invatrial tmparil ' Broadhunt B; Eywnck (1964) nnalgsis C u m i n s , \\:a&, & Budk-Olsen (Sate 2)

\Vwds, Ruck&haur, & Bowliog (19fLI) Oldhnm gi hlorlock (19iO) Cmdland B; Nagy (1969)

Collins (1966) Nogy b- C l w r (1970) Branstein (1972) Broadburst (1969) Cummi-, Wnlsh, & Budk-Olsen (Note I) Furchtgou, \Vcchkin, & Dees (1961) Ader (1969) Ehrlicb & Burns (i953) hlanoscuitz (1YiO) h R a r t a & \Y&e (1971) Diron 8; Defries (1%) Nngy & Holm (1970) h'agy & Clarer (19i0) Adcr (19C9) Brmdhunt & Eysendt (1964) Cumnlinr. \Val&. & Budtr-Olwn &ole I)

S u b j c ~ t P ~ r ~ m e l c r

Rat ' r\mbulntion

Gerbil Cs t

nlouse Def-lion

Rat

Pcrret hjourc Ambulnlion

Cat Domestic fowl

Ellecf

I - q u a d n l i c !rend I ~ u a d r n t i c trend I-liner trend !-linear Vend 0 change--nmbulnlion described as nancxplaratory

padng f a c m sir trials, in\.ertd U over 10 days

I 1

I I 1 from nu). I to Dny 2 but then f I

1 across triall I ~ C T O I S trials ! across trials 1 acres trials' 1 across trids I linmr trend I q u a h t i c trend 1 quadratic trend 0 ! Lath studies shot\. reliable ! individually specific mler af dccrure across trials 0 1

- .

. 492 . ' ROGER X. \ V U H LVD ROBERT A. CUlnfNS -

Possibly el~troencephalographic measures odd-even correlation. Scores were summed represent some of the most uselul and fasci- across Days 1 and 3 and across Days 2 and 4 nating parameters of the future. .As yet they and correlated for both strains separately. a r e a novelty, but the succes of the Czecho- Coefficients lor ambulation with both deie- slovaL5an workers L i t and Gollovi-Hemon cation and rearing scores were .68 or above (1969) in relating bippocampal theta activity for both strains, wbile frequency and dura- to behavioral open-Aeld measures augurs well tion of wahing .were reliable for albino but for this and other electroencephalography. not for hooded subjects. All other variables

failed to display acceptable reliability. RErrmnl rv OP OPEN-FIELD VARIABLES Somewhat similar findings bave been re-

ported by Martin& and Ut (1969) in their As yet there have been veq' few study of the long-term stability of individual

of the reliability of the many variables in differences in evloratory and other behabior use. This represen& a very disturbing state ,d Lhe rate of habituation in dogs. ne sub. of affairs, since unless a parameter can be j,& were given of ,,,inUte.' dura- shorn to possess a satisfactory degree of reli- tiqn each spread over days and were then ability and hence to be a stable characteristic retested in a smiliar manner approldmately of individual subjects or situations, its use- . 300 days later, ~ ~ b ~ l ~ ~ i ~ ~ satisfac- fubess is almost nil. Fortunately, the most torily smb]e, a i t h the of ambu- widely u . 4 variables, ambulation m d deleca- lation, raing, and snifhng gi\ing the tion, have been found to po- acceptable reliability even though snifing alone proved reliability in several species. However, ambu- reliable, Vdimtion, del&- lation yields a complex pattern of test-retest tion, urination, and body (smding, correlations because it is a factorially complex slthng, . . lying) all proved inadequate. Further- parameter, loading on both exploration and more, the rate of habituation of the combina. emotionalit?, yet changing the direction of tion of ambulation, reuing, and miffing, gi\.en its loading on-emotionality from positive on Gther by the regresion coefident by Day 1 to. negative thereafter (.\\'himbey 6: relative between and last Denenberg, 1967). \Vhile correlationsof Day 1 on a day, was also found to be hi^ ambulation 6 t h scores of subsequent days - studyof the of habituation has been -are satisfactoq' (about .j), they are lower further expanded by it and Gollo\+-Hemon than the intercorrelations between other days. ( 1969), who reported that habituation cumes

Ivinskij (1968) studied a, variety of vari- Show dear intrinsic oscillations whose size able5 mwured in several ways. Albino n d a stable individual characteristic..

. . hooded rats were tested ove'r four daily tnals, The sum of these oscillations has been and albinos were reteste'd in the same way a lobilily rcorc and a correla"on coeficient '

63 and 217 days thereafter. In mamining , 7 7 (!\, = 20) was found when retesting test-retest reliability of variables for the ,,, performed after months in a dinlerent albino group, scores were averaged over 4 situation, days, and i t was found that ambulation and ~h, ,~ , there would seem to be for the frequency of both defecation and washing the not only oi the total magnilude yield& product-moment correlations greater but in son,e also for the rate and in- Ihan .4 ($ < .O1) Over the 217 days. trinsic o ~ i l l a " ~ ~ ~ of habituation of certain frequent). of rearing, latenc?l to leave the . beha\iors. ~ ~ b ~ l ~ t i ~ ~ , rearing, and deieca- center d r d e when initially placed there, and tion would seem to adequak duration of washing all yielded coefficients while urination, snihg, body position, and of .34 (p < .Oj) or above. This left all three .

~nner circle ambulation would seem to be measures of urination (frequency, presence or absence, and amount in ,,,illigrams),, duration definitely inadequate; and a number of 01 rearing, and inner circle ambu~a~ioo nith . others, such as washing and latency to leave very low and quiie nonsignjti-t correla- the central circle where initially placed, must tions. A further test dFas provided by an be viewed as only marginally reliable. H o a -

, I

OPES-FIE CLD TEST 493

ever, these few variables represent less than a Lhird of those reviewed in this paper, so that i t is obvious that most of the remainder must ar y e t also be viewed with suspicion. I t is interesting to note the apparent strain differences in the reliability of the washing measures. (I\inskis, 1968) ;which suggest that due care must he exercised in extrapolating the reported reliabilities to subjects of other genelic and experimental backgrounds. Also interesting is Itinskis' finding that different

-. , . m w u r e s of the same bebavior.gave compa- 1 rable coefficients. I t is disturbing to note the

unacceptably low reliabilities of many of the parameters which have been tested and to reflect on the probable unacceptability of the reports employing them. These low reliabil- ities may partially account for the plethora 01 inconsistent findings which fill the literature.

T E ~ O I U L ANALYSIS OF OPEN-FIELD B ~ m t ~ o n

I t was mentioned in the discusion on tech- niques that testing may consist of more than one trial and that these trials may be sub- dibjded into intervals, thus alloving temparal analysis and effectively adding another di- mension to the behavioral study. Many workers measure only total behavior per trial or even total behavior over all testing. This failure to perfom1 temporal analyses within and between trials efiectiuely coll3pses results across two dimeasionsand hence the worker tends to view behavior as static rather than as constantly changing, as an event rather than as a process, and partially ignores tbe efiect of prior exposure to the apparatus. 1n so doing, the analyses must result in a di- minished sensitivity, since the subjeclrc may difier in their behavior only a t certain phases of testing. The failure to periorm temporal analyses also results in the dijcarding of an

. immense amount oi information includine the possibility of studying the na:ure, rate, and lability of habituation-characteristics aha%

P individual stability and imporlance hare been clearly demonstrated ( L i t 8. Gollovi-Hemon: 1969; 3Iartineli gi f i t , 1969j. Occasionally where an experiment calls for the manipula- tion of multiple independent 1.ariables: the diiwrding of temporal analy~es may wnier

some advantage in obviating the difiiculties of interpreting higher order interactions, but against this must be weighed the disadvao- t a g s listed above.

Intratrial temporal analyses have b e e n l w frequent h n those across trials and have largely been..limited to ambulation. The - general pattern for a variety of species seems to be a decreasing frequency of ambulation across i n t e n d s mithin trials. This decreased rate may either remain approximately steady (linear trend) or diminish exponentially from an initially relatively high value (quadratic. trend). The latter trend has been reported in both male and female r a h of the hIaudsley Reactive and Nonreactive strains by Broad- h u n t and Eysenck (1964) and in young \\'bite Legborn (GoUus domeslicrr~). The former linear trend has been noted by Oldham and Xforlock (1970) in Mongolian eerbils and in Sprague-Dawley rats (Woods, Ruckelshaus, 8. Bowling, 19.60).

Studying \ c u l i u t i o n in the cat, Candland and Nagy (1969) found differences in intra-, trial trends to be a function of age. Kittens showed a marked intraswional decline over 10 minutes'but did not reach the consider- ably lower levels of the adult cats,. whose vocalization diminished only slightly. In v i e t ~ of the absence of defecation and the appar- ently none.~lora ton. nature of ambulation (subjects paced backwards and forwards near .the unit door without any apparent exploratory intent), the authors suaes ted that vocalization might yield a better index of emotionality in this species than the traditional measures.

S tud ie~ across trials hatre been more exten- sive than those witbin trials. The varieties of satistical analysis are of course limited by the number o i triali. The simplest and most common analysis has been determination of the sienificance level of intertrial diiferences. .9 more inlormative metbod is trend analysis, and 'where several trials are used, i t is possible to examine not only the general trend but also the nature. magnitude, and periodicity of trend Euctuations.

Once again ambulation has most often been studied and a variety oi patterns has been noted. Consistency has been greatest in the

, ,' . . 49.1 ROGER N. WALSH AXm ROBERT A. CUMMISS > -

mouse where the majority of workers have ble change--increase, decrease, or none-has reported decreasing ambulation across trials. been reported. This has. been reported as a significant, efiect Other spmes have been l e s closely studied. for subjects reared under enriched and stan- In the dog! Nartinek and Lit 's (1969) find- dard colony conditions (Manosevitz: 1970), - ins of individually reliable d e c r e a in am- in small or large litters .(LaBarba &! iV%ite, bulation across trials has already. been dis- 1 9 i l ) , and for B.ALB/c.J and CS7BL/6.J cussed, but Fox and Spencer (1969) found inbred mice and their reciprocal F1 hyhrids that exposure to the field some weeks previ- a t sevcral aTes (Dixon & Deiries, 1968a). ously resulted in modest increases. Xeither Ambulation decreased across 4 days in kittens nor adult cats displayed any sig-

. . handled and unhandled C3H .mice tested nificant change across 10 days, while. the under high or low illumination a t 50, 70, or domestic iowl showed a dramatic i n c r w e in 100 days of age ( N a p & Holm, 1970). This spiie of consistent decre& within trials report furnishes an example o i the added (Candland &! N a p , 1969). sensitivity and information afforded by inter- Latency to leave the starting .unit wass trial analysis, since although the main effect found to increase across 3 days in rats of for handling was not significant,. a number of various ages (Furchtgott e t al., 1961). A its interactions with trials were. Thus, 50- more complex picture aas noted by Ader day-handled subjects were more active than (1969) when latency fell markedly from controls.on the first day, but their ambula- Day I t o 2 but then increased slightly over tion decreased more rapidly such that i t was the' next ? days. Latency declined-across less than Lhe controlj on Daxs 3 and 4, and 5 test days in ferrets (Ehrlich k Burns: this pattern was different a t the three ages. 1958) and there was a similar trend for both

. . Similar examples are available in the study rearing and scratching, while vocalization de- of Nag? and Claser. (19i0) , who demon- clined over 6 days in idul t cats but not in strated a significant (apparently asymptotic) kittens (Candland 9. Kagy? 1969). On the decline across 10 days in C57BL/6.J mice. other band, snifing by rats increased across

Less consistent results have been obtained 3 days, Lhough this was dependent on the age with the rat, although a decrease across days of the subject (Furchtgott e t al., 1961).

. has been most often reported. This has some- Cross-trial studies of defecation have been

- times been linear (Ader, 1969) or apparently almost entirely limited to the mouse and the quadratic (Broadhurst 8: Eysenck, 1964) rat, and once again only three or four trials where a decrease was reported f r o m Day I has been the norm. However, within these to 2 which was partially recouped on Days limitations there would seem to be consistent 3 and 4. Broadhurst $nd Eysenck 'hypothe- evidence of species difference3 in trend, io r sized that inhibition may account for the n,hile the rat shows a lewning in deiecation initial intertrial reduction as well as the intra- across days (a pattern that has beenaccepted

. trial decline already discussed, while reduc- as the norm and the basis for hyptheses tion of fear m a i account for the i n c r m e building both within and betueen species), in ambulation across Days 3 and 4. The rate the mouse displays the opposite trend. This of decrexse washeightened by prior rearing decrease in the rat has been widely reported,. in a complex environment, a finding attrih- thought not without exceptions (Ader, 1969), uted to a more rapid habituation of the and has been confirmed over as many as 60 arousal response to novel stimuli (\\'alsh 6- trials (Bronslein? 1972). Furthermore, this Cummins, in press; Cummins et a]., Note decrease has been s h o r n by Broadhurst 1). Barrett and Ray (1970) and VaUe (1969) to have a dea r genetic-component (1971) found that young ra ts become more with a high average level of dominance, with active across trials, but other workers have the s u ~ e s t i o n that natural selection will tend found no significant intertrial chanze (e.g., to favor low-scoring strains resistant to Furchtgott e t al., 1961 ). Thus, every possi- change across trials.

OPEX-FIELD TEST 495

. 1 For the mouse on the other hand (practi- cally. the only other species for which evi- dence is available: indicating the reliability of the defecation measure), there has been a high degee of consistency btween reports on intertrial trends. Thus, defecation has been found to increase across as many as six trials (e.g., Collins, 1966). These studies afford an interesting example oi the oversimplification which may result from taking a relatively small number of levels of any factor under investigation, for when the number of trials KU increased to. 10 (Kagy S. Glaser, 1970), i t was found that the trend could now be identified as an inverted U with the number of boluses increasing through to Day 6 fol- lowtd by a decrease over the remaining 4 days . The marked consistency existing be- tween studies of intertrial changes in defeca- tion has recently become explicable in light of a quantitative genetic analysis of open- field behavior in mice by Deiries, HeLpan, Ross, and Howard (1969). I t was found that change in defecation displayed as high a de- gree of heritability as did the daily and total scores:In spite of a similarly hish consistency between reports on change in mouse ambula- tion, genetic influence on change in this mea- sure was found to be considerably less than for eith-r daily or t o h l scores or ior any of the defecation scores.

Interpretations of temporal changes in open- field behavior have varied with the paiameter under study and the meaning attached to it by the experimenter.. Thus, for example, the cro%-trial' decrease in ambulation and deie- a t i o n in the rat has been taken as indicating a diminution in emotionality, anxiety, fear, w p e attempts, and territorial marking, or a t !he more general level: of the development of inhibition or habitualion. However, in conjunction with an!, interpretation must b: an awareness that the placement of a subiect in an originally novel environment must elicit an orientation reaction which subsequently hnl~ituates. One might conceive of this reac- tion as a high+rder factor in a psychophysio- logical construct hierarchy, modulating multi- determined lower order constructs such as fear, anxiety. territoriality, and the like. a,hicl~ would in turn modulate the even more

overdelemined spedric behaviors observed in the open field. The precise nature and maggi- tude of the changes in these bzhaviors will of course be determined by the genetic- experiential background of the individual, but any attempt to esplain temporal behavioral changes must certainly incorporate the con- cepts of orienhtion reaction and habituation.

I t has long been known that many depen- dent variables in the open field correlate significantly with one another. This is all the more to be noted since some positively cor- related behaviors are mutually e~clusi\.e (ambulation and rearing) and hence their .degree of relatedness is presumably under- estimated by simple correlation.

A basic deduction f rom these correlation studiff has been that many parameters tend . to measure t h e same psychophysiological -state and as such have provided several useful implications. First, such correlations allow for validity testing, since the patterns oi behatior serving as a s u y s for altered psycho- logical states should be predictable under di- verse conditions. Second, they may afiord an increased sensitivity and reliability of mea- surement. At the same time, they may also allow an economy of measurement because where parzmeters are highly correlated, one may well dispense with some of them. Thus, L i t and Cmllovi-Hemon (1969) were able to dispense with measures of snifing and dean- ing in the knowledge thzt a weighted sum of 3mbulxion and rearing would still yield an adequate correlation a i t h hippocampal them activity.

Not surprisingly the correlation between ambulation and defecation scores has most often been studied. Since Hall's (1936) re- port there has been a high degree of consist- ency in the literature affirming a negative relationship between these measures in the rat. In the mouse a similar trend has most often been noted! and this in spite of the previously mentioned increase in defecation across davs. More detailed s t u d i s in this. species, however, have begun to reveal some of the complexity of the relationship, and

496 ROGER N.WALSH .Ah73 ROBERT A. CUMhfISS ,

in a diallel study, Bruell (1967) found a positive significant correlation ( r = ,124) in males, a i t h a negative and equally significant correlation ( r = -.I 59) in femal6.s.

The relationship has been less frequently studied in other species ,and .has almost al- ways yielded nonsignificant correlations, espe- dally for $ecies in which defecation bas been found to be an unreliable measure (e.g., Candland & Sagy , 1969).

I n the vast majority of cases, correlation has been performed between scores totaled over aU trials. However, considerably more light h a been afforded by recent studies which have examined the relationships be- tween scores obtained on individual trials. Nagy and Glaser !1970) divided C576L/6.J mice into high and low ambulators and studied the cross-trial patterns o i defecation under a variety of ages and illuminations. High and lorn ambulators did not differ in defecation on Day 1, but thereafter the low subjects defecated significantly more; and their defecation continued to rise until declin- ing from Day 6, whereas high ambulators defecated man'mally on n a y 2. Meanwhile, ambulation decreased steadily in both groups, suggesting that the correlation is not asimple one and afiordina another example of the 1 0 s of information contingent on failure to perform temporal analyses.

As might be anticipated, the various mea- sures of motor activity tend to be positively correlated, even in cases in which the behav- iors tend to be mutually exclusive, ris with amhulation and rearing. These two param- eters have been found to be highly (up to 3 1 ) correlated in a variety of rat strains (e.g., lvinskis, 1968; Ray Fc Hxkhauser, 1969), although Delbarre et al..(1970) have noted a dose;dependent, pharmacologically induced dissociation between ambulation and rearing.

Since the definition oi latency usually in- volves the ahsence of ambulation, i t is not surprising that these two measures have been found to be negatively correlated. This has been widely reported for both the rat and- mouse, and both parameten have been found to load on common factors hut in opposite directions--namely, exploratory behavior in

the rat (Denenberg & Wlimbey, 1968) and on a factor which was unnamed and uninter- preted by the authors in the mouse, but by analogy with studies on the rat would seem to represent a factor of emotionality (Poley 8. Royce, 1970).

Correlational studies of ambulation with oLher variables have been less frequent. Ader's (1969) use of corticosteroids has al- ready been mentioned. A negative correlation with urination, just significant a t t h e .O5 level, was reported b y Satinder (196S), a doubtful finding in view of the demonstrated l o r reliability of urination measures; Simi- larly, ambulation displayed a sipificant posi- tive correlation with inner circle actitity in spite of the latter's demonstrated unreliabil- ity (Ivinskis, 196s). Finally, ambulation and vocalization showed a positive nonsignificant trend in dogs (Fox & Spencer, 1969) and the reverse in human infants (Rheingold, 1969).

.As the most widely studied parameter, defecation has frequently been used as a co;- relate. The negative correlation a i l h ambu- lation has already been detailed. I n view of the ncat ive correlation of ambulation and latency, i t might be anticipated that defeca- tion and latency would he positively related, but in the main 'only nonsi~nificant trend< have been found (e.g.. Ader. 1969: Porter Fc Wehmer, 1969). The lack of correlation uith changes in heart rate (Candland & N ~ F , 1968) has already. been mentioned: while Satinder (1968) reported a n q a h v e correlation Kith grooniin~ over four trials.

.A logical extension of intercorrelating open- field variables is the study of their relation- ship to measures obtained on other behavioral tests. Such correlations may afford the ad- vantages previously mentioned of increased sensitivity. reliability, and economy a? well as providing valuable information on the apparatus specificity of aqv apparent rela- tionships. Perhaps the best example o i this specificity is afforded by ambulation. Thus, although some workers report good correla- tions between ambulation scores obtained in different apparatuses (e.g., ambulation in the open field, endosed maze: and an exploratory box [Stretch, 1960, cited by Broadhurst & Eysendt, 1964]), most have not been so en-

OPES-FLELD TEST 497

thusiastic. Particularly common have been studiff of ambulation a s measured by activ- i ty wheel and open field. These studies have yielded consistent reports of significant but not particularly high correlations (Mano- sexitz, 1970; Weasner, Finger, PI. Read, 1960; correbtion coefficient of .41). Con- siderahly lower correlations have been found where motor activity is measured by cage tilting or stabilimetn.. Here i t is obvious that the highly coordinated motor wheel activity may be only slightly similar to the multitude of part and whole body movements regis- tered on the stabilimeter. I t should be noted that even simultaneous measurement of motor activity may yield relatively low correlations. Thus, simultaneous mex?urement of total open-field activity over 30 minutes gave co- efficients of .65, .65, and .49 for cage tilt- ing- beam interruption, cage tilting- flwr contact, and heam interruption- floor contact, respectively (Sparks 8: Lodiard, 1966). Open- field ncliuiry is sometimes used synonynously with ambuhtion, but in \iew of the term's varying usage by different workers and the relatively low correlations above, it is best avoided in this contest.

Ambulation h3s also been found to be cor- related .+th m w u r e s whose ambulatory na- ture is less obvious. \Voods, Ruckelshaus, and Bowling (1960) found significant cor- relations of the Hebb-\Villiams' maze error scores ~ 4 t h ambulation in rats reared under free or restricted eovironments. They there- fore suggested that the often reported finding of lower error scores on maze and discrimina- tion tasks in animals reared under conditions of increased environmental complexity might be due to a reduction in conflicting explora- tory tendencies rather than to an increase in learning ability per se. This suges ts an a yet apparently untried u u g e of the open field, that is, the use of one or more of its parameters as covariate adjusters in cases in which related behaviors are believed to be confounding variables. Thus, in the above study the u s e o f ambulation as a covariate oi error scores might yield a clearer pictuk of the origin of the error score difference.

Satinder (196S), using measures of ambu- lation, defecation, urination! grooming, and

rearing, found little relation to conditioned escape avoidance behavior in the rat ex- cept for positive correlations of avoidances and intertrial crossings m'th ambuktion and a negative relationship between intertrial crolsings and defecation.

L i t and Collola-Hemon- (1969) have de- scribed the relationship of their previously de- scribed open-fieldderived m w u r e s (nonspe- cific excitability level, inhibition, and lability) to both behavioral and physiological measures. Thus, an analysis of individualdifierences in learning a double-T maze revealed h t these three factors accounted for a major portion, of variance on measures of learning, running time between starting box and first runmy, and decision time on choice points. Further- more, nonspecific excitability level was found to van. systematically with indiridual dii- ferences in such diverse physiological m a - -

ures as growth rate, body temperature, caloric intake, qualitative food preference, and endo- crine functions; a a ide range to be sure and an interesting example of what will hope- fully be one of many demonstrations of phys- iological substrates of open-field behavior.

In view o i its multiple intercorrelated rari- a b l s , mosi of &em oi doubtful meaning and their correlations with variables obtained in other tests, the open held would seem to lend itself ideally t o factor-analytic studies. In- deed, such an approach mould seem to aflord an optimal method of interpretation and de-' termination o f the factorial structure of open- field behavior, yet such itudies remain a rarity. Since it furnishes an excellent example of the potential of this method? the study by \\?himbey and Denenberg (1467) is described in some detail.

IVistar rats were assigned to one of 1G treatment combinations that resulted irom completely crossing four independent e.xperi- ential conditions, each a t two levels, yielding a 2 :< 2 X 2 :< 2 iactorid design: The four variables ac re handling oi the subject's mother during the latter's infancy, handling of the subject during infancy, housing o i molher and young betureen birth and weaning, and, finally, housing ior the first 21 days after weaning. Commencing a t 220 days of age, scores were obtained on a numhzr of widely

i - . . . - . . . . . . . - . _-'

. . . . . . .

. . . .. . - . - . . . . . . . . , -. :-.. : : . . . . . . . .

. . . .- . - 4 9 s . . . . . . ROGER S. \V.U,SH ASD:ROBERT ,A. CL;M>iISS

. . ., used :measures of exploration, emotionality,

' . avoidance learning, consummatory behavior, and- the. Like, yielding a tntal of 3.7 scores in- cluding open-field ambulation and defecation

. on Dags I-4-and a f i e i an intervening period - . . of testing on several other apparatuses, on

Day 14. Outstanding feature of theanalysis were the relatively lox correlations of the first day's ambulation score with those o i

.other da-s and the positive coirdation di Day I ambulation wi thal l defecation scores -in-cnmpsrison to the negative currelations of ambulation on other days. The two orthogo- nal factorsacco'unting for most' of . tbe vari- ance of Lhese variables were interpreted as cmotional reaclivily and cxpforation. Defe- cation loaded only on emoliond reacliaily. whereas ambulation proved factorially com- plex, :loading positively on . czplorution but changing i t s loading on emotional reactkit? from.positive on Day 1 to negatirre for other days. In addition, defecation was found Lo load on a Lhird factor wGch was named con- summation+limination (Denenberg, 1969b). T h e factorial complexity-of the ambulation score Frovidcs, for the firs: time, an explana. tion f o r w m e of the lar'gedumber of contra- d i c t o ~ f i n d i n g s and interpretations of .this

-parameter. Ambulation: has for . many y a r s been the traditional memr-e of rat explora- tion, while others h u e used il (sometimes in conjunction with defecationbut sometimes not) as an index of emotionality, fear, or escape behavior. It can. now be seen that it may-well he indicative of esploration as well as .either high or low emotionality but that intekpretation i s next to impossible where only one trial is e'ven and the information from deiecation scores is not taken into ac- count. In view of the similarity of the within- to the between-trizls decline in zmhula t i~n in the rat, i t i s interesting,to speculate as to whether there might be a similar reversal of loading on emotionality within trials, espe- cially for Trial' 1 . The advantage conferred by the finding.that the two major parameters of the open field' load well o n these two important dimensions of animal behavior is immense, but i t -must be noted that the gen-

and for different .test conditions remains un- proved. One has only to remember the in- c r k n g defecation across. trials iound in the mouse, as compared w i t h t h e decrease of . the 'rat, to realize the relevance of these. qualifications.

\\%at then are the iactors determining t h e qualitative and quantitative nature of be: llavior erjnced by a subject in the "pen field? Basically, & is true: of any behavioral test , this behavior represents the interaction of the suhject with the experimental situation. This concept of operationalism, that is, the interaction between that which is measuring and that which is being measured &d the inseparability of the properlies and efiectj of one irom the other, might well be extended from physics to the behavioral science. Fur- thermore, each o i these factors influencing be- bavior may be divided into interacting sub- components which of course may he iurther subdivided ad infiniturn. Division will mainly be carried only to the first stage in this study: since it is a t this level o i behavioral de:ermi- nation which most open-field studies are con- cerned with and because of the incresing complexity a n d diminishing returm trhich iurther subdi\ision brings.

Sutject,variables are perhaps best clavified under genetic, developmental, and experien- .tial headings. Genetic c o m p n e n b of species, 'strain, and sex are most often discussed. Developmental features usually refer t o the physiological maturation procwes, but o n e might subsume biological rhythms under this beading. .E.xperiential variables have been c l s i f i ed by a variety o i criteria,. most com- monly by the nature of the stimulus and the development phase a t xrhich it was given. For the purpose of the present discussion, the most uxiul classification is into experience p i n e d prior to testing, manipulation of the subject involved in bringing it to the expe~i- mental situation, and e.rperience of the ex- perimental situation up to the instant a t which hehavior is beine measured. -.

. ' erality of this finding for other species and Some idea o i the interactional tornplexity strains, for different experiential backgrounds, of behavioral determination may be gained

:*.L ;., .

OPEN-FIELD TEST 499

from looking at studies which have simul- taneously examined the effects of several vari- ables from just one of the above divisions. Denenberg and his co-workers (Denenberg, Karas, Rosenberg, 8i Schell, 196s; Denenberg Sr Rosenberg, 1963; Denenberg.& Whimbey, 1969) manipulated the handling of the-sub- jects, their pretieaning.and postweanins en- vironments, using both male a n d female Wistar rats and found. a significant four-way interaction among all variables o n open-field ambulation. . '

Coming now to the test situation side of the equation, Lhis may be divided in to the test environment and measurement procedure. The test environment. has previously bcen considered in some detail under the headings Techniques and Apparatus Characteristics, The Testing Environment, and Procedural Details. The nonequality of results,obtained by difierent methods of measurement of sup- posedly the same behavior has a h previously been mentioned? and interactions between treatmqnt and measurement techniques are well known in the literature. Such an inter- action was remrtrd bv IVeasner et al. (19601.

mental situation, stimulation afforded by the experimental apparatus, stimulation afforded by the test environment, e . ~ r i e n c e of the test situation up to the inrtant of measure- ment, and method o f measurement. The im- plications of this. for open-field studies are discussed below but may also be generalized to other areas of behavioral research.

This interactional n a t u r e o i the behavior places severe limitations on the extent to which generalizations czin be made from the results of any individual esperiment. I t also increases the potential importance of changes in any one factor 01 the interaction equa- tion, so that apparently Lribial alterations of any one stage of the chain reaction may re- sult in totally unexprcled changes in hebav- ior. The importance of this for coniparisons between treatment groups rests on the differ- ential efiects of any 'changes. hleier (1963): in discussing the conditions which modify behavioral development, has stated:

Thc rondilions that have been drrcribed (c.8.. handling, Litler sire, malcroal w a i n and prepartal cxperirncc, cage sire and complexity. lo name a f rw] . wcrr ~rev ious ly releqatrd to the groccles

demonstrated that' photoelectrically reg. limbo a: m i n t e n a n t e routines .which s e r e inire-

istered home-cage ambulation and revolving- quentty recorded 2nd codihed, and even then ior only limited or local conrumplion. Now we rcaliur

wheel counts yielded different results for that variablrs are in their poteoo. equivalent treatments. In comparing such t o cffrrt behavior change that is drmonlrable for measures of behaxior obtained by-difierent prrio& often a t e n d i n g far into adulthood. \Ye tffhniques, there are several factors to be canno: help hut wonder how many of the now-

c k i c controrersie in the behavior sciences could considered. First, there is the question of .,,be rewired were detiled information avail- tibether the behaviors being measured by able on the maintenance routines employed by L ~ P

each method a r e in fact thesame. Presum- protaaonisls in thore controversies. (P. :7) ably they can never be identical; the study of S p a r k and Lockard (1966) using cage tilt, beam interruption, and floor contact m w u r e s of activity afiords a gross e m p l r of this. Second, there is the queslion of dif- ferential sensitivities of techniques,. as regards bolh ~hreshold and sensitivity to increments of behavior. And third, turning full circle: '

there is the matter of interaction b e t ~ e e n the subject and the measurement apparatus. The concept of operationalism fits well here.

In summary then, any behavior represents the determinant o i the interaction of genetic badground, maturation, biological rhythms, experience prior to testing, stimulation in- volved in bringing the subject into the experi-

4 note of despair. has been c a t by Henderson (1970), who stated "for the time being investigators must be aware of the pos- sibilities that early environmental interactions nith cenotype may limit the validity of their Findings to their OUT unique laboratory situa- t i o d ' (p. 509). This statement stems from the author's work on effects of genetic and early environmental interactions on a number of subsequent behaviors (Henderson, 1963, 1969); including open-field defecation, in which it was obvious that there tias no single relationship between prior treatment and subsequent defecation across difierent mouse strains. >foreover, this is only one of the posible 72 one-way, let alone multiaay,.inter-

actions derivable from the nine major inde- pendent variables listed earlier! As early as 1958, King listed variables likely to be of importance in determining subsequent behav- ior and pointed to the lack of concern by investigators for control oi variables other than theone in which they themselves were interested. Today, little has changed except for the enormous r o w t h of the list .of vari- ables known to be important. I t is small wonder then that contradictions and failures of replication abound in open-field and much other behavioral research.

-What then cyl be done short of abandon- ing a nomothetic for an ideographic approach. First, experimenters must bccome a m r e of the nonsdditive nature of behabioral determi- nation. With this in mind: one can see that studies which examine only one factor in imlation are of considerably lesj value than those which permit consideration of a vari- able in light of the effects oi others. To this end Henderwn (1969) recommended the use of considerably more elaborate multifactorisl multivariate desi.ps than have previously been customary and felt that drawbacks o i size and multiway interactions are a neces- sary price to be paid if this area is to advance beyond its present limitations. .\ further need . . which will enlarge such designs even more i s the .need for large numbers of factorial levels in view of the nonlinearity of response to su many variables. Furthengore, those rele- vant variables not included in the design must be ~ rupu lous ly controlled, and' the leveln a t wbich t h e were controlled must be reported.

INTERPRETATION OF OPEN-FIELD MEASZTRES

Ihat then does t h e open field measure? Interpretation may be in terms of under- lying constructs or on the basis of the behav: ior's presumed purposi\-e or adaptive nature. Underlying constructs may be suggested by the face validity or anthropomorphic interpreta- tion of a particular behavior, by resemblance to a natural behavior pattern or to other constructs, or by factor analysis. I n t ~ i t i \ ~ e or face validity interpretations have been the 'norm until verv recentlv: the diificulties and

widespread ignorance of the ethologid signifi- cance of much behavior studied in laboratory situations (Henderson, 1968). .

The construct of emotionality bas been the major one for open-field work. Although rarely defined except in circular operational t e r n (Denenberg, 1969b), i t can be thought of as an entity underlying the nonspecific affective components of behavior. We say nonspecific components because. an. adequate stimulus and appropriate goal-oriented behav- ior cannot be identified, although anthmpo- morphic. analysisor experimental manipula- tion suggests the presence of an aiiective state (Ader. 1969). ii'hile initially b a e d on an- thropomorphic interpretation, the construct has been factoridly validated (Zi'himbey 8: Denenberg. 1467). .%lLhough Ridely ajsumed to underlie a large number of behaviors, fac- torial loadings have as yet been demonstrated only for defecation and.ambulation. Ilinskis (1970) attempted to test the validity of deie- cation, an~bulation, latency, rearing, aashing, inner drcle activity, and urination as indexes ol emotionality in rats..His criteria of valid- ity were, first, intertrial decrease of scores acrosr, 4 days; second, efiects 01. prior open- field testing to reduce scores; and third, ei- fects of alterations in anibient auditory and tisual stimulation. The adequan. of these criteria would seem to be debatable, and in iact Ivinskis concluded that ol the three, only stimulus variation was satisfactory. Using this criterion, of the seven parameters tested, only t w d e i e c a t i o n .and latency-ould be considered valid indexes of emotionality. The necesit).. for interpretive caution when using the o~en-field test is even more aDDarent in . . light of this study, especially when i t is re- membered that of these two m w r e s only defecation has been found to be .acceptably reliable (Ivin~lris, 1968).

Another widely referred to aflective con- struct, which would seem to be ven. d o s e t o emotionality, is that of fear. It does not seem certain to u h a t exent the behavioral andphysiulogical components of fear can be difierentiated from those of emotionality, and the two t e r n often seem to be used synony- mously. Since measures of emotionality have , ,

dangers of which are compounded by the most often been obtained in situations wbicb

CLD TEST '501

anthropomorphically would seem likely to in- duce fear, one wonders if there is an advan- tage to be gained a t this stage by trying to difierentiate them except denotatively in terms of the specific experimental stimulus situation.

Specific. affective behaviors and constructs have naturally been rare in view of the ab- sence of adequate stimuli in most open-field testing. However, with two subjects in the field 'simultaneously; the construct of gregari- ousness has been used to explain differences in average distance between and time spent in contact by the subjects (Latan6 et a l . , 1970).

Exploration has been a much used con- struct whose prime measure has been ambu- lation. While this assumption has been vali- dated by factor analysis, its limitations have been pointed out by hIcCall e t al. '(1969). Once again, many other parameters are thought to be indicative of exploration (sniff- ing, rearing, etc.), but factor-analytic valida- tion is still awaited. At the psychophysio- logical level, arousal-habituation has proved extremely useful, especially in the work of U t and Gollova-Hemon with their derived constructs o f nonspwihc excitability level, inhibition, and lability.

One of the ' most pressing needs in the formulation of open-field constructs is an increased reporting of validity studies.

The erlablishiog of validity u always tenuous, for in the long rua validit!: b determined by whether the %ajority oi inrrsti~alars, or at least t h c most outspoken, believes the external criterion to hare been adequak. (Candland k Nagy. 1969, p. 841)

Such has largely been tlie case in the open field, but this consensual validity must now rest on the bg i s of further criterion and construct validity testing, for which, in addi- tion to the correlational studies which form the basis of most convergent validity testing (h'unnally, 1967), factor analysis would seem particularly suited.

I t is obvious in reviewing this area that many of the all-too-numerous discrepancies, failures of replications, and contradictions stem from preventable methodological and experimental design causes. Many o i these

have been enumerated already. but their inclusion in a concluding section seems more than warranted in v iew of the danger of "the growing body of literature in this area suffocating under its own weight" and in the hope of helping to "convert the current paper explosion in this area into a knowledge explosion" (Hendewn, 1969, p. 867).

First, with regard to procedural matters, there is a need for a more precise specifics- tion.of subjects including their genetic back- ground, prior treatment, apparatus, testing conditions, measurement techniques, and tem- poral factors. Where replication or compari-' son is the aim, all possible factors other than the one under manipulation must be suupulously controlled. ' h regards experimental design, tbere is a

need for the recognition of the essentially nonadditive nature of behabioral determina- tion. \Vith this in mind experiments which manipulate only one independent variable would be better replacd by multiway iac- torial designs, which because of their econ- omy andnonassumption of additivity yield vastly more information and whose limita- tions regarding generality of conclusions can be a r s e w d in light of interaction terms. Similarly, there is a need for increased num- bers of levels i n factors under investigation. Testing subjects under. light and.-dark tells nothing about the precise relationship be- tween behavior and level of illumination unless demonstrably untenable mumptions of linearity are made. Similarly, comparing two species or strains tells nothing of the genetic basis of behavioral differences: whereas an approach such as the diallel cross, which af- fords a large number of precisely quantifiable genetic levels, allows very detailed psycho- genetic analysis. Failure to perform temporal analyses of behavior discards large amounts of information and may result in diminished sensitivity.

For dependent parameters, there is a great need for reliability and validity testing for a wide range of conditions and subjects. >.Ian? papers base far-reachinr assumptions and conclusions on parameters whose reliability and validity remain unproven. The wider use of physiological variables, particularly bio-

so? ROGER X. WALSH k\KJ ROBERT A. C U J 1 M ~ ' S

chemical and electropbysiologicaI, .and the . . .-

determination of their relationship to behav- ioral would seem to hold consider; able prom&.

,As regards interpretation of behaviors, there is a need for mnstruct validation' for which factor analpis and ethologiul data would seem to be particularly relevant. The former tendency to regard variable and construct as almost equivalent and constant across sub- jkts and treatments must be abandoned ex- cept where construct validation indicates that it is permissible. Individual dependent param- eters shouldnot be considered in isolation but in light of changes in other parameters. This has been demonstrated particularly ior arnbu- lation. Finally, the dangers' oi unqualified generalization acmss genetic, experimental, and testing backgrounds must be appreciated.

Broadhunt, P. L. Determinants of emotionality in the rat: I. Situational factors. Bn'tirk Journal of Pryckology. 1957, 4S. 1-12,

Bmadhurst, P. L. Detcrminanls of emationalitg in the rat: 11. Antecrdent iactors. British l o u m l of Psychology, 1958, 49, 12-20, (a)

Broadburst, P. L. DctcrminanU of emotidnatity in the rat: 111. Strain diflcrcnres. Journol of C m - parafiuc snd Ph?siological Prycholo~.~, 1958, 51, 55-59. !b)

Broadhunt. P. L. Psychogenetits of emolionalily in the rat. .4nnnlr 01 Ihc Scw York Acodcmy o] Sdmcc, 1969, 159, 806824.

Broadhunt, P. L., k Eysenck, H. J. lnterprctationr of erploralory behavior in the rat. I n F I . J. Eyrenck (Ed.), Ezpcn'mcntr inmotivation. Elms- ford. X.Y . : Pergamoa Prru, 1964.

Bronstein. P. hf. Open fic!d behavior 01 the rat as a iunctian of ape: Cross sectional and l on~ tud ina l invertigaliozs. J o m l 01 Comparative ond Phyr- iologicnl Psychology. 19i2, 80. 335-341.

Bruell. J . H. Brhasioral hetcrasb. In J . Hirwh (Ed. ) , Brhozior-Gcr.t!ic rlnal?sis. S e w York: n1cCra~-HiJI . 1967.

Candland. D. K.. '9: S ~ F , Z. h1. Thr Open-Field: REFEKESCE S O T E Some comparative data. Annolr 01 ihc rVew Yolk

rlrodnny o! Science, 1469, 159. 831-Sjt. 1. Cumminr. R. A,, Vialsh, R. N.. & B u d u - O k n , Clark, C.. Gorman, D., & Vrrnadlkir, A. EBects o! 0. E. EnvironmenuUy induced cbango in opcn- p r e ~ a t a l administration of prychorropic drugs on field beha\,ior. Paw< prc-tcd a t Lhe 44th annual behavior of d.,,e!oping rats. D~~~~~~~~~~~~ meeting oi Lhc Australian and KenZealand ASO- chobioiogy iO, ,, 2:5-2j5, ciation ior the .Aduancement of Science, Sydncy, August 19i2. Collins. R . L. What clw does thc defecation score

measure? Proccrdingr of the Aznval Conoentioc of the Am~n'can P~ycholoj-icol .4srodotion, 1966.

REFERESCES

AbeI. E. L. Habituation as a factor in carly handling. Jou rnd of Comporoli;~ and Phyrio- logical Prycholoqy, l F i l , 74, ?IF-222.

Ader, R. Adrenocortical function and the measure- - men1 of emotiooality. .4snolr of !he n'cw Ymk Acodrm? 01 S d m c t , 1969. 159. 791-S95.

Archer, J . Tests for emotionalit? in rots azd mice: A revira. Animal Bthocior, lFi3, 2 1 , 20'135.

Barrett, R. J., 8 Ray, 0. S. Behavior in Lhe open- field. Lashley I l l m ; r e , l u t t l e box, and Sidman avoidance as a function of ma in , sex and age. Dcorlopmcnfal Psychology, 19i0, 3, i3-i i .

Battig. K. Drug r6erts on caploration of a combincd maze and open-&Id system by rats. Aznalr o! the Xtw l'ork .4rodtmy o! Sdcnce, 1969, 159, 892-S97.

Bindra, D., & Spinner, N. Raponse to dinerent degrcer of novelty: T h t incidence o l various activities. Jourw! of thc Erpcrimcnfol Analyk ol Behatior; lqj8, 1. 341-353.

Blirard. D. A. Situational determinants of open-field lehavio; in m u murcullis. Bn'fish Jrmrnol of Pry- cho!ogy. 1971. 62. 243-252.

B&k, I. B.. & Errman, \V. B. C:ou-fh hormone 'administration during prrpnanc?.: A behrviornl diflerencc in oflrprinq rats: Solure, 196j, 205. 11361137.

I , 147-1;s. Dcfries. J . C. H e w n , J. @., Ross, D., b- Howard

h1. A gumtirative genetic analysis of change in o ~ n field behavior of mice. Ps>,chonom'c Science, 1969, 17. 152-154.

Delb=rrc. B.. Dumas, G., 8. Guionniere. 1 . .an autosa ted open-fieid method. P~ychophormaco- login. 1970, 18, 227-230.

DcnenScrg. V. H. Opcn-field behavior in the :a!: What docs it mean? Annals. a! the iYcut Yorh .4codrmy of Scirnrc, 1969, 159, SS2-859. ( b )

Denncnberg. Y. H.. liaras, C. G., Xmenberg, I;., 6. Scbell, 5. Programming liie histories: An crpcri- menlal d e s i q and initial results. Dcudopmrn!al Pryckobiilogy, 1963, 1 , 3-9.

Denenberg. \'. H., & Rorenberg. I;. M. Proqram- minq liic historits: EKrcts of maternal and enri- ronmcntal variables upon open-field behavior. Dcvelopmeald P rydob io lo~y , 1965, 1, 93-96.

Denenberg, . H.. \Vrhmer, F., Wcrhofl, J.. B Za:ron,, hl. S. Effects of portweaning enrichment and isolation upon emotionality and brain weight in the mouse. Physiology and Bchmior, 1969. 4. 403406.

Dcnezberg, \.'. H.. k Whimbey. A . E. Experimental programming oi life bbtories: Towards a n rxperi- mental idmm of individual diffcrcnrrs. Deorlop- m m t d Prychobiol~gy, 1963, I, 55-59.

CLD TEST

&on, L. K., & Deiries, J. C. Development of o p n - field behavior in mice: Effects of age and e r p r i - en-. Dmelopmmtd Ps?chobiology, 1965, 1. 100- 107. (a) .

D h n , L. K., & Defrier, 1. C. Effects of illumina- tion on o p n held h h a r i o r in mice. Journal of ComparDlive rmd Phjriological Psychology, 1968, 66, 605-805. (h)

Ehrlich. A,, & R u m , S. ErploratoryhebaCior of - the black fooled ferret. Canodion Journal of PI?-

chology, 1958, 12, 235-241.- Eyxnck, H. J., & Broadhurst, P. L. Introduction.

In H. J. Eysenck (Ed.), E r p e n m n l r in molioo- lion.~Elmslord, N.Y.: .Pergamon Press. 1964.

Fox, hf. W., & Spencer; J. W. Exploratory behav- ior in the dog: Erperimcntal or age depndcnt? Dcvrlopmcnlal Psychobiology. 1969, 2, 68-74.

Fuller, J. L., Chamben, R . .M., & Fuller. R. P. EBects of cortisonr and of adreoalcctamy on aclivlty and emotional behavior of mire. P s ~ c h o - ~omol ic Ucdicine, 1956, 18, 234-242.

Furcbtgott, E.. Wechkin, S.'. & Dces. J, \\'. Opeo- held exploration as a iunction of a g . Journal of Comparo!ive.and Phpsiologicol Psychology, 1961, 54, 386-3-355.

Hall, C. S. Emotional behavior in the rat: Dcfeca- Lion and urination k measurcs of individual dii- ferences in emotionality. Joumal o! Comparath:c P s ~ c h ~ l o g y , 1934, 18, 385403.

Hall, C. S. Emotional behavior in the rat: n l . The relationshio between emotiomlitv and ambu- lator). aclivity. l o u d of Comparntiue Plychol- ogy, 1936, 22. 345-352.

Hall, C. S., B'Rdlechey? E. L. A s!udy of the rat's behavior in a 6e!d: A contribu:ion to method in comparative psychology. Uniumrity o,! California Publications i n P v c h o l o ~ ~ . 1932. 6. 1-12. .. .

Henderson. S. D. The ranlaundinz rficcu of mnetic variables in early experience research: Can we ignore them? Dcseloptentol P~ychobiology. 1965, 1. 14+152.

Henderson, N. D. Prior treatment effects on open- - field emotion?lity: The "red for representative

design. Annals of the Ncm Yorb Acodemy of Sdcnce, 1969. 159, 8E-868.

Henderson, X. D. Genetic influences oo the behavior of mice can he obscured by laboratov rearing. Journal o! Cornparalive ond Ph.vriologicd Prp- chology. 1970. 72, 505-511.

Hoicr, M. A . Cardiac and respiratory function during sudden prolonged immohililv in wild rodenla. P~ychoromeric h f ed id~e , 1970, 32, 633- 647.

Hughes, R. N. Social facilitation of locomotion aod erploratio? in rats. BritLh J o u t d of P5pchofog.v. 1969, bO? 365-358.

Ivinskir, .4. The reliabilily of behavioral measurer oblsined in t he open-field. Aur!mlim Journal of

Ivinskis, A. A study of validity of opca-held mea- sures. AuslroL'on J w m d of Psychology, 1970, 22, 175-15.3. - - ~-..

Jerrard, L. E., 8; Brunnel? 8. X. Open-6cld bchav- ior of hipporampal icu'onrd rats and hamsten.

Journal of Cornparalivc and Phydological Psy- chology, 1965, 66, 50W502.

King, D. L. Comparisoo bctwceo two metbods of demonstrating whtcdners of emotionality rariahlu in me. J o d 01 Comporarivc and Phyriologkal Psychology. 1970, 72, 238-243.

King, 1. .+. Parameters relevant to d c t e r m i k n ~ the effech of carly experienv upon the adult behav- ior of animals. Psycholofical BJlcl in. 1958, 53. 46-55.

Xrsiak, ?.I., & Janku. I. hlruurement of pharmaco- logical dcp-on of erploratorv activity in mice:

A contribution la the problem of time economy and wnsitirity. Psychophormocologia, 1971, 21, 118-130.

Labarba,R. C., & White, J . L. Litter sizevarialiom . and cmationalreacti\ity in BALBJc mice. J o m a l 01 Compomlicr and Ph?riulogicol Prychology, 1971, 75, 254-257.

L i t , J., & GoUo\=-Remon, E. Permanent efircts of nutritional 2nd cndocrinolopical iotrrvention in early ontagcny on the level of nan-rw&c uc i t - ahililp and on lability (cmatioaslity). Ann& of Ihc 'YLW York Acadcmy of Science, 1969,129, 710- . . . 720.

Latan&, B., Cappll . H., & Joy, 1'. Social depriva- tion, housing density. and gregariousncrs in rats. Journal o! Comparative and Phyriologicol Pry- chology, 1970. 70. 221-227.

Le\ine, S., Haltmeyer. G., Karas, G., & Denenberg, V. Physiological and behavioral eRecU of infantile stimulation. Physiology and.Bchocim, 1967, 2, 5s- 59.

Livewy, P. J., & Egger, G. J. A;e as a factor in opn-field rpspohiivmtrs in the white rat. Jou+nd of Comgnroliur and Phyriologicnl Psychology, 19iO. 73, 93-99.

Lynn. R. ' Arlnlion, orourd, onr! the on'enlation rcaclion. Elmford , N.P.: Pergamon Prerr, 1966.

3fanoresitr. 31. Early enriranrnenml enrichment and mouse behavior. Journnl. of Comparoliae ond Phyriologicol Psychology, 1970. 71, 459466.

Martinek, Z., & Li t , J . Long term stability o: indi- vidual differences io exploratory behavior and rate of habituation in dogs. Phyriologio Bohemoslo- unico, 1969, 18. 217-225.

3IcCall. R . Caretaker rffert in rats. Dcr!dnprrm!d P$?cholog?. 1969. I. 771.

3lcCall. R.. Lester, >I., & Dolan. C. Differential rearing and the crploration stimuli in the open- field. Uct~clopmcnlal Pryrholog?. 1969, I. ;jO-:t?.

>llrClearn. G. E. Strain diffcrenvs in adirit) ' in mice: ln6ucnre of i!lumination. J o m a l of Com- porntioa ond Pkpriological Psycholog?, 1960, 53, 142-14:.

hleier, G. W. The half to:d tale and th: half replica- tion. Dsr~r1oprnm:al Psychubioiog~, 1968, 1, 77- '

i8. hlontgomery, K. C. The relation between crplora-

tory behavior and spontaneous alternation in the white rat. J ou rnd 01 Comporoliva end Phjsio- logical Pspchology, 1'351, 44, 581-559.

ROGER S. WALSH. ASD ROBERT A,' CbTfhlTSS

Morrkon. B. J., .& Thatcher. K. Overpopulation cEecu on social reduction.of emotionality in the albino rat. Jmmd 01 Comporolivc and Phyrio- logical Prvchology, 1969. 69. 658-662. - .

Moyer, K. E . EEect of adrenalcctomy on emolional elimination. Journal of Gmctic Psycholog?. 1958, 92, 11-16.

Nag?.. M. Z. , B Forest. E. 1. Open-field behavior o i C3H mim. EEect of size and illumination of field. Pr).chonomic Srjcnrc, 1970, 20, I Q-21.

Nag).. Z., & Glaser, D. N. Open-field behavior of CSiBLj6.J mice: EEect oi illumination, age, and number o i test day. Prvchonomic Science, 1970, 19, 143-145.

N a u , A!. Z., B Holm, AI. Open-field beharior of C3H mice: EErct of earlv handlinx. field illumina- -. tion, and q e a t testing. Ps.vchzomic Scicncc, 1970, 19, 275-275.

Xielson. T . C. Early experience and explorative bc- havior in the white rat. Scondinooian Journal of Psychobgy, 1970. 11. 1-6.

Xielson, IT. C. YSual and tsctual explorstion in rats reared in dificrcnt dcgrers of visual and t x tua l rtimulatioa. Scondinnuion Journal of Pry- chology, 1971, I : , 53-m.

NunnaUy, J . Psychometric fhcor?. Sew TorL: hirGraw-HiU, 1967.

Oldham, J.. & Morlock, H. The eKecu oi open- field size on activlty in the mongolian gerbil. Psychonomic Sdmce. 1970, 20, 2M.

Pare, \{I. P.. & Cullen, J. \<I. Emotional behavior and adrend iunction in the rs:. P ~ ~ r ) c k o l o ~ l c d Reports, 1F65, 16, 283-236.

Paul, C., &.Harlena, J. Mar: learning and open- field. behavior .of sdrenalectomized rats. Prycho- logical Re~or t r . 1962, 15, 263-186.

Poley, N., 8 Royce, J . R. Genotype, maternd stimulation, and faclon o l mouse emotionalit)-.

. J o d o! Comporolir,e a r d Pby~iologicol PI?- tholog?. 1970, 71. 2462.50.

Porter, R. H.. & \\'ehmer. F. hlaternal and inhntil? influences upon erp!oratory brhavior and ema- tional reactivity in the albino rat. D~vrlopmcnlol Psychobioligy, 1969. 2, 1%2j.

Prescott, R. Some behavioral -effects of variables which influence general level oi acti\ity of r a s . dnimal Behavior. 1970, 18, 791-796..

Ray. O., B Hockhauser. S. Grow:ti honnonc and environmental complexity eiiens on beharior in the rat. Dturlopmcnld Psychology, 1569, 1 , 311- 317.

Rheingaid, H. L. The etfect oi a s:rsage environment on the behavior of infants. In B. M. F o s (Ed.), Delnminonts o! infont behavior IV. London: Methuen. 1969.

Sachet:. G. P. R e r p o n ~ to slimulur novelty and complcxily as a iunction of rats early rearing experiences. Journal o! Cnmporativr and P h ~ s i o - logical Psychology, 1967. 63, 569-375.

Satinder, I;. P. A note on the correla:ion bela,ccn open field and escape avoidanct behavior in the rat. Journal 01 Psychology, 1968, 69. 36.

Satinder, I;. P. Effects of dour trails oa open-field behavior in rats. Prr.chological Rrporlr, 1969, 25. 11>116.

S p a r 4 L. 31., 6; LocLald. R. B. Relatioashipr among three meuurrs of home cave activity. Ps~chological Report,, 1966. 19, 7i0.

Tighe. T. G. A handiiog device lor small animals. Journal o! the Ezpcrimental Analyrir 01 Bchno- ior, 1965, 8, 261162.

Tobach, E. bfinipuhtion eflccts, open-held rxperi- cnre, and digestive tramit time in Wktar male and

. female rats. Psychological Rcporfs, 1966,.19, 375- 3;s.

Valle, F. P. EKects of strain, EX, and illumination on open-field behavior of rats. dmericon Journal of. Pficholog?, 1970, 83. 103-111.

Valle, F. P. Rats' performance on repeated ttsls in the open field a r a funclion of age. Psycho- nomir. Sdcncr, 1971, 2.7, 33.L335.

\Valsb, R. S., b Cumminr, R. A. AIechmisms mcdi- sting the prcduction of environmentally induced hrain changes. Psychological. Bu lk fh , 1975, 82, 986-1[)00.

\\blsh, R. S., B Cumminr, R . A. Neural rrspo- to therapeutic environments. In R. 9. Walsh & W. T . Grcenough (Eds.), Ensironmcnlr nr l h n - npy !or brain disorders. Xerr York: Plcaum Press, in prm.

\Vez-e:, \\:. .H. , Finger, F. \Ir.. B Rcad. L. S. Activity changes under food depri\.ation as a iunr- tion of recording device, Journal of Cmpam!:ve

a n d Physiological Psychology, 1960, 33, 410474. \Velis. P. h.. Lawe, G., Shcldoo, 11. H., b Williams,

D. I. E i i ~ t s of iniantil: sfimulatiun and en\.iron- mental iamiliarily on crp!oratary behavia: in tbe rat. Brilirh Journdl gl Pspchology. 1969, 60, 339- 19.1. . .. ~

\Vhimbey. A. E., 6; Denenbrrg, 1'. R. Two indexn- drnt behavioral dimensions in opcn-6cfd perform- ance. J0;ur.d 01 Compnroliwe and Ph~sioiogical Pr?chology, 1967. 63, 5W50.1.

Whittier, J . L., & hlcReynoldr, P. Pcrsis~ing-odourr as a biasing fnc:or in open-field research with mim. Canadian J o m a l of, Prychology, 1965. 1!J, 224-230.

LVim~r, R., & Stems, H. Controllrd visual input and exploratory activily in C57BLj6.J n i ce . Pcrccptval ond J1o:or Skills, 1564, 18. 29S307.

Woods, P. J., Ruckelrhsus, S. I . ! Q bowl in^. D. M. Some eKerts oi "iree" and "rcs;ricud" cnviron-

' mcntal rearing conditions upon adult behavior in the rat. Psyrhological Reports, 1960, 6, 191-2W.

Zimhardo, P. F.. Q Montgomcr).. K. C. EKects of free environment rearing u ~ o n erplora:ar). be- havior. Prychological Repoils, 1957, 3. SSe59:.

(Rcccivrd Janunr). 28. 1575)