Meijer & Robbers recently reported wheel-running (WR) in wild mice . Arguing that because WR occurs in non-captive animals, with bout lengths resembling those from a laboratory study, this ‘falsifies one criterion for stereotypic behaviour’ [1, p. 1]. They suggest that this should reassure researchers who use WR to investigate exercise but are also concerned that it ‘merely signifies neurosis or stereotypy’ [1, p. 1]. However, concluding that this should ‘help alleviate the main concern regarding the use of running wheels in research on exercise’ [1, p. 1] is premature. Here, we propose better ways to assess whether WR is pathological, making three main points: observing WR in wild animals does not demonstrate that laboratory animals' WR is normal, because abnormal behaviours often develop from normal ones (see text below and electronic supplementary material); how to distinguish stereotypic from non-stereotypic behaviour depends on the definitions used, but has never been based on bout length alone (see electronic supplementary material); and clear, useful objective methods exist for identifying behaviours that reflect underlying pathology (see text below).
Stereotypic behaviours (SBs), such as repetitive pacing and object-biting, are common in captivity . Taxon-typical, they often develop from normal intention, redirected or displacement activities performed when highly motivated behaviour is thwarted. Repetition may be further sustained by changes in neural circuitry (e.g. in the basal ganglia) that mediate behavioural inhibition and cause perseveration (functionless repetition) [2–4]. As in humans (see below), SBs can be malfunctional: for example, stereotypic horses and mink are poor at winning food in tasks requiring flexibility [4,5], and stereotypic mink gain fewer copulations in mate choice tasks . Many have asked whether WR is stereotypic (e.g. [1,7–9]), not least because it can consume many hours a day and persist for weeks (see electronic supplementary material). Using the standard, descriptive definition of SB (repetitive, unvarying and apparently goalless ; see electronic supplementary material), WR clearly is [7,8], but this leaves unanswered the key question: can WR be pathological, too [1,8,9]?
Current data suggest that WR is heterogeneous (e.g. [1,7,9]), occurring, respectively, in adaptive, maladaptive and malfunctional forms. Thus, as well as in nature, WR occurs in large, semi-natural enclosures . It is reinforcing, and may be stress-reducing [8,10]. Providing wheels can boost litter size  and reduce bar-mouthing [9,11], an SB shown by caged rodents highly motivated to escape . Together, this suggests that WR, even in laboratory environments, can represent normal, adaptive, intrinsically rewarding running.
However, WR varies with circumstance, increasing as current environmental quality declines. It is thus higher in barren cages than enriched ones [12–16]; and in animals exposed to noise, rough handling, intra-specific aggression or predator cues [12,14,17–20], or deprived of food [8,21], mates  or alcohol (if addicted ). When elicited by sub-optimal circumstances like these, WR may represent thwarted, unsuccessful, and thence maladaptive attempts to relocate . Consistent with this, WR is also highest in environments eliciting the most bar-mouthing in hamsters , and can be reduced with anxiolytics .
Could WR also reflect pathology? In humans, abnormal repetitive behaviours typify autism, and impulsive, obsessive-compulsive and stereotyped movement disorders [2–4,23,24]. For WR to indicate malfunction (e.g. [1,9]), it should therefore demonstrably reflect alterations in brain function that are disabling (e.g. that interfere with social, occupational or other important areas of functioning ). High WR rodents do show altered basal ganglia physiology (e.g. [7,25]). Genotypes prone to addiction (e.g. ) and neurological plaques  also perform more WR. As for impairment, WR may cause paw injuries, tail malformations or even lethal weight loss (e.g. [9,21,27]) and promote infanticide . Missing, however, is research systematically comparing how WR differs between normal subjects and those affected by brain malfunction.
We therefore suggest raising genetically similar, individually identified rodents (see the electronic supplementary material) to adulthood in environments varying in quality (from large, semi-naturalistic and low-stress to small, barren and chronically stressful), to induce differential neurological development [2,23]. We predict that impoverished rearing will elevate WR, regardless of the test environment (WR thus not merely reflecting dispersal). This should covary with perseverative errors (e.g. during extinction/reversal learning) and basal ganglial changes (e.g. regional cytochrome oxidase activity ). Furthermore, the most affected subjects should be most impaired (e.g. poorest at maternal care or attracting mates, and/or in tasks requiring flexibility). Were these predictions confirmed, then WR's fine details should be analysed (bout lengths; 24 h time budgets; inter-bout and day-to-day variation; fidelity of running direction). This would reveal how WR changes as malfunction increases, so yielding useful diagnostics for researchers keen to avoid forms caused by underlying pathology.
Overall, WR in wild animals is fascinating. However, it should not be used to ‘explain away’ or normalize WR in laboratory rodents, because pathological activities often develop from normal ones. To identify whether WR reflects pathology, we advise using the objective methods that exist for identifying when repetitive behaviour patterns reflect underlying malfunction and for elucidating underlying mechanisms. These methods come from clinical practice and neuroscience, and have recently revolutionized the understanding of captive animals' SBs [2–4]. They could do likewise for WR. Meanwhile, for researchers concerned that WR may not model normal exercise, we suggest either recognizing that findings may relate more to ‘exercise addicts’ than normal exercising humans, or, instead, using enriched, low-stress housing to minimize risks of frustration or brain malfunction.
G.M. and H.W. co-wrote the original submission. G.M. led revisions, with H.W. editing/commenting to improve the MS. Both approved the final submission.
We declare we have no competing interests.
G.M. thanks NSERC for funding.
Thanks to Maria Diez-Leon, Mike Walker, Innes Cuthill and referees for helpful comments.
Comment to: Meijer JH, Robbers Y. 2014 Wheel running in the wild. Proc. R. Soc. B 281, http://dx.doi.org/10.1098/rspb.2014.0210
- Received April 10, 2015.
- Accepted July 10, 2015.
- © 2016 The Author(s)
Published by the Royal Society. All rights reserved.