The transformation of quinpirole-induced behavior produced by a home-cage in the open field was striking. Checking, and an incessant locomotion, disappeared, as the rat entered the home-cage and spent its time there. However, after about 40 minutes, checking resurfaced, as the rat began to exit the home-cage for brief periods of time and to visit the previous spot of concern. A novel cage had no such effects on quinpirole-induced behavior. Below, we consider possible reasons for the effectiveness of the home-cage in arresting checking, discuss whether the obtained findings reveal the phenomenological experience of a compulsion in the rat, highlight the implications of the study for an animal model of OCD checking, and relate controllability of behavior to compulsive checking and stereotyped behavior induced by psychostimulant drugs.
'Safety' Cues, Gradient of Suppression and Reward
The home-cage, but not a novel cage, was effective in suppressing checking. Thus, some psychological attribute of the familiar container accounts for its effectiveness. To the extent that OCD checking is an exaggerated form of normal checking regarding one's well-being and security , and a similar relationship exists between quinpirole-induced and normal checking in the rat, then it is reasonable to suppose that the suspension of quinpirole checking emanates from a sense of 'safety' provided by the familiar contextual cues of the home-cage. However, the effectiveness of such 'safety' cues was time-limited, as quinpirole rats resumed their checking of the open field environment after spending 40 min in the home cage. This resumption could be the outcome of a habituation process to 'safety' cues of the home cage, occurring in the presence of continual drug-activated stimuli for checking. A similar dynamic process of a changing balance between 'safety' and 'checking' cues may account for the variable success that OCD patients have in resisting their compulsions .
While cage familiarity produced the expected suspension of checking, the time at which the cage was introduced into the open field did not show the predicted effect. Based on the notion that quinpirole-induced behavior may be related to the activation of dopamine reward circuits [24–26], we expected that an animal fully engaged in compulsive checking would find it more difficult to suspend this activity than an animal which has not yet started to check. However, regardless of the time at which the home cage was introduced into the open field, quinpirole rats entered it quickly and remained there for 40 min. Thus, both the suspension of checking behavior and the duration of this suspension, was independent of prior ongoing checking. This observation raises to question whether quinpirole-induced checking is indeed related to positive reward stimulation or whether, on the contrary, its repeated performance is propelled by factors with a negative valence. Moreover, it raises the question of whether the controlled variable in the rat's behavior is the duration of staying in the home cage or the length of time that the checking can be suspended. Unfortunately, the design of the present study is inadequate to answer these questions unambiguously. Nevertheless, we suggest that the latter alternatives are the more likely answers because they are consistent with our speculation that quinpirole rats may experience their checking activity as 'compulsive,' as discussed below.
Experience of Compulsion
As noted by Reed , the primary criterial attribute of OCD is the experience of compulsion. Yet, so few authors have tried to "elucidate or analyse compulsion itself. The normal ploy is simply to ascribe the adjective 'compulsive' to such nouns as 'ideas,' 'thoughts,' or 'impulses.' The meaning of 'compulsive' is never examined; it seems to be regarded as so self-evident as to be unworthy of study or exposition... It might well be enquired how it is possible to study a phenomenon without first examining the very factor that defines it" [11, p. 120, italics in original]. According to Reed, to be called 'compulsive' in a clinical sense, the subject must find that the urge to perform the behavior "is intrusive and ego-dystonic, that he feels it is absurd, and that he struggles unsuccessfully to resist it" [11, p. 11]. That is to say, the salient experiential features of 'compulsion', as defined by the first and the last of the above three criteria, are a reluctance and an ultimately doomed resistence to engage in the behavior. From the above definitional criteria, it also follows that compulsions afford the subject with "no gratification or good cheer" as the subject fights "a losing battle with something which is not acceptable to him - a battle, moreover, which seems interminably protracted, exhausting, and thus, distressing" [11, p. 7]. In other words, the compulsive experience is not a pleasant one.
Although they do not prove it, the findings of the present study are consistent with the notion that the quinpirole rat experiences its checking activity as compulsive. The definitional criteria of compulsion predict that because the compulsive experience is distressing, subjects would at first avoid engagement in compulsive rituals if given an opportunity to do so. However, in the face of the compulsion-evoking situation, this avoidance would become unavailing, and, inevitably, the compulsive behavior would emerge. In other words, the expected behavioral pattern associated with phenomenological compulsion is but a temporary suppression of compulsive rituals. This is precisely the pattern found here for quinpirole-induced checking. Specifically, when provided with the opportunity to do something else, quinpirole rats did not engage in checking behavior but instead sought out and remained in their home-cage placed in an unattractive location of the open field. Ultimately, however, they did resume their checking activity. According to this schema, therefore, quinpirole-induced checking does seem compulsive.
The notion that rats experience compulsion under quinpirole is consistent with another finding. Quinpirole rats checked more, the longer they remained inside the home cage (Figure 4 and Figure 5). Such a result would be expected if staying in the home cage was indeed the rat's attempt of resistance to compulsions because the effectiveness of this strategy should decline as a function of home-cage time, based on an expected process of habituation to home-cage ritual-suppressive ('safety') cues.
While the obtained results are consistent with the hypothesis of a compulsive experience, the data are also open to alternative interpretations. For instance, rather than a presumed avoidance of checking under quinpirole, the observed pattern of results could merely reflect the choice between two relatively pleasant activities, namely, staying in the home cage and checking of the open field. Or, rather than using the home cage as a vehicle to refrain from checking, the quinpirole rat may be merely using it to escape the open space of the open field environment, a suggestion bolstered by the fact that not only quinpirole- but also saline-treated rats preferred the home cage. Clearly, to discount the alternative interpretations, the current preference-like paradigm should be refined to require a certain amount of work to access the home cage, an amount that is beyond the interest of saline rats. Because quinpirole rats would need to exert extra effort for the opportunity to suspend their checking behavior, this would offer a more compelling test of whether checking behavior does, or does not, provide "good cheer."
It should be noted that the present findings do not address the remaining criterion for clinical compulsion, namely, the presence of insight into the "senselessness" of the compulsion. Consequently, for a full test of whether or not quinpirole rats experience their activity as "compulsive," future studies must address this criterion as well.
Animal Model of OCD
We noted previously  that compulsive checking may share mechanisms with drug-induced sensitization, as both phenomena are induced by quinpirole and exhibit similar features [23,27,28]. However, a blanket inference that all drugs that induce sensitization produce compulsive behavior seems unwarranted without a series of validation studies as being done for quinpirole [16; present study]. In fact, one must consider that the relevant factor for the genesis of compulsive checking may not be the mere induction of sensitization to quinpirole. Instead, the relevant factor may be repeated exposure to quinpirole in a specific environment, namely, an environment (such as a large open field) that evokes checking behavior readily. The latter possibility has merit because the psychological and physical characteristics of an environment affect not only the nature of the acute response to psychostimulant drugs [29–36] but also the amount and the type of behavior that is sensitized with chronic drug treatment [23,27,28][37–43].
Our focus on quinpirole-induced checking as a possible animal model of OCD, stemmed from serendipitous observations of an apparent surface similarity of the drug-induced behavior to that of the motor compulsions of patients with OCD [44–46]. While there exist different types of validity by which to evaluate animal models of psychiatric disorders [47–50], our process of validation of the quinpirole preparation does not fit neatly into any one of the described types. Specifically, our validation strategy involves the identification of the essential behavioral, psychological, and neurobiological properties of the human disorder and testing whether the same properties are present in the quinpirole preparation. In this context, we asked previously what characteristics define the spatiotemporal structure of OCD checking and examined whether the same features are found in the behavior of quinpirole-treated rats . Moreover, we examined whether a pharmacological agent used in the treatment of OCD produces an amelioration of checking in the quinpirole preparation . Similarly, in the present study, we investigated whether the expression of checking in the quinpirole preparation is subject to external inhibitory control as it is the case for compulsive checking in OCD. Other ongoing studies examine the presence of additional attributes of the human disorder.
Although we would view our strategy as striving for face validity, this term is used in a more restricted sense by Geyer and Markou . These authors consider that "face validity refers to the superficial similarity in symptomatology between the model and the disorder," and that this type of validity is of little scientific use, being difficult to defend rigorously because of, invariably, "subjective arbitrary arguments." Clearly, such a description does not apply to the pursuit of validating the quinpirole preparation, as there is nothing arbitrary or subjective in testing for critical properties shared by OCD compulsions and quinpirole-induced checking, especially when those are defined and measured in a strictly objective manner. Of course, the task to identify which properties of OCD are the crucial ones, and how to define them operationally for measurement in the animal, is not a trivial one. However, its difficulties do not imply that the approach lacks rigor. One should note also that our strategy does not seek "superficial similarity" but, on the contrary, asks whether the disease-defining attributes of the human disorder are present in the animal preparation. Thus, our evaluation of the quinpirole preparation as to its face validity, extends well beyond the scope of such a validation as described by Geyer and Markou .
The usefulness of an animal model is particularly striking when findings from the model reveal an attribute of the human disorder hitherto unappreciated. While it is still premature for such studies using the quinpirole preparation, nevertheless an incidental observation from the model may prove revealing. Specifically, because quinpirole is a dopamine agonist, and to the extent that the drug does indeed induce compulsive checking behavior, then the model predicts an involvement of dopamine systems in OCD compulsive checking. Indeed, at a point in time when serotonin was thought to be the primary neurotransmitter system in OCD, our early observations with quinpirole were one of the few experimental findings that the authors employed to derive their then novel hypothesis that dopamine may play a role in OCD , a notion favored now for a particular subtype of OCD [52,53]. It follows, therefore, that the quinpirole preparation may turn out to be particularly useful for elucidating the role of dopamine circuits in OCD.
Stereotypy, Compulsions and Voluntary Control
The checking behavior of rats under quinpirole had been described as "flexible, yet recurrent" . That is, even though the moment-to-moment flow of checking behavior under quinpirole is unpredictable, checking activity repeats itself on a larger time scale and hence the overall spatiotemporal structure of checking under quinpirole is highly predictable. The presence of recurrent behavior under quinpirole may give a reader the impression that the rat is engaged in a motor automatism uninfluenced by external stimuli or internal cues, akin to "stereotyped behavior" induced by psychostimulant drugs [36,54,55]. Such is not the case, however. Not only does the actual behavior appear as relatively spontaneous to an observer [16,22], but also it is closely coupled to environmental stimuli , and is even subject to interruption, as shown here. Nevertheless, we suggest below that the difference in the nature of quinpirole-induced compulsive checking behavior and psychostimulant-induced "stereotyped behavior" is one of degree rather than kind.
Stereotyped behavior induced by psychostimulant drugs is often conceptualized as consisting of movements that are repetitive, aimless and involuntary [54,55]. The view that the behavior is "involuntary" may imply that there are no controls over it, as is apparent from the early labels of the drug-induced behavior as "compulsion" , "compulsory"  or "forced" . In fact, one method of scoring "stereotypy" evaluated whether prodding the animal would disrupt the drug-induced behavior . However, it is now abundantly clear that psychostimulant-induced behavior is subject to modulation by environmental, psychological and experiential factors [36,38]. Importantly, recent studies showed that rats treated chronically with amphetamine can leam to suppress even sensitized stereotyped movements to obtain milk reward [55,59,60]. Thus, as pointed out by Wolgin , stereotyped behavior is not irrepressible, being subject to control by the organism as well as external stimuli. Therefore, the notion that stereotyped behavior is an uncontrollable motor automatism is not justified.
Because the stereotyped behavior induced by psychostimulant drugs is subject to control by external and internal stimuli, therefore, it is not a qualitatively different phenomenon than the quinpirole-induced compulsive checking. However, the two may differ in the degree of controllability, with quinpirole-induced checking being more open to control than the stereotyped behavior induced by other psychostimulant drugs. This suggestion stems from the observation that the variability in spontaneous behavior is greater under quinpirole than amphetamine [61,62], indicating a higher potential for flexibility under quinpirole than amphetamine (and related psychostimulants). Consequently, a wider range and intensity of stimuli may be effective in influencing behavior under quinpirole than amphetamine. Alternatively, the domain of responses induced by quinpirole versus amphetamine-like drugs may be more open to modulation. Such difference in degree of controllability may be related to the specific mode of action of the various compounds and/or to the dosage of the drugs used.
Using the notion of controllability, stereotypy and compulsions can be viewed as sequential points on a continuum along a dimension of spontaneity [30,63]. Normal behavior is at one end of spontaneity as it represents behavior that is free to vary with changes in external and internal stimuli and is readily open to voluntary suspension. The other end, a loss in spontaneity, is represented by stereotyped behavior in that it reflects behavior with a narrow range of possible responses, few effective stimuli to modify it, and a limited capacity to suspend ongoing activity. Accordingly, compulsions fall to the right of stereotyped behavior in that there are relatively many stimuli that can modify compulsive behavior and the behavior can be suspended relatively more easily than stereotyped behavior. Thus, stereotypy and compulsive behavior may be differentiated by the degree of loss in spontaneity and in particular the extent to which the behavior can be suspended by the organism.