Animal Cognition

, Volume 15, Issue 6, pp 1143–1150

Habituation and dishabituation during object play in kennel-housed dogs


  • Anne J. Pullen
    • Anthrozoology Institute, School of Clinical Veterinary SciencesUniversity of Bristol
  • Ralph J. N. Merrill
    • WALTHAM® Centre for Pet Nutrition
    • Anthrozoology Institute, School of Clinical Veterinary SciencesUniversity of Bristol
Original Paper

DOI: 10.1007/s10071-012-0538-2

Cite this article as:
Pullen, A.J., Merrill, R.J.N. & Bradshaw, J.W.S. Anim Cogn (2012) 15: 1143. doi:10.1007/s10071-012-0538-2


Domestic dogs are reported to show intense but transient neophilia towards novel objects. Here, we examine habituation and dishabituation to manipulable objects by kennel-housed dogs. Labrador retrievers (N = 16) were repeatedly presented with one toy for successive 30-s periods until interaction ceased. At this point (habituation), a different toy was presented that contrasted with the first in both colour and odour (since the dog’s saliva would have accumulated on the first), colour alone, or odour alone. No effect of the type of contrast was detected in the number of presentations to habituation, the difference in duration of interaction between the first presentation of the first toy and the presentation of the second toy (recovery), or the duration of interaction with the second toy (dishabituation). Varying the time interval between successive presentations of the first toy up to habituation between 10 s and 10 min had no effect on the number of presentations to habituation, nor did it alter the extent of dishabituation. Varying the delay from habituation to presentation of the second toy, between 10 s and 15 min, affected neither the recovery nor the dishabituation. Overall, the study indicates that loss of interest in the object during object-orientated play in this species is due to habituation to the overall stimulus properties of the toy rather than to any single sensory modality and is also atypical in its insensitivity to the interval between presentations.


Domestic dogHabituationDishabituationObject play


The domestic dog is renowned for routinely engaging in play even when adult, and object-orientated social play is commonly used by owners to interact with dogs and enrich their environment (Rooney et al. 2000). Provision of manipulable objects can be beneficial for the welfare of individually housed dogs when preferred play objects (henceforth ‘toys’) are offered (Wells 2004b), but the extent to which dogs interact with such toys appears to vary considerably. In laboratory kennels, beagles have been observed to display more long-lasting interest in toys, interacting with them for about a quarter of the observed time (Hubrecht 1993), than did dogs in rehoming kennels, which interacted for less than a tenth of the observed time (Wells 2004a). Dogs display powerful neophilia towards novel toys (Kaulfuß and Mills 2008), and anecdotally, they can also rapidly lose interest in particular toys. The behavioural mechanisms that may lead to the latter, which are likely to include habituation, do not appear to have been investigated systematically.

Habituation, ‘the response decrement as a result of repeated stimulation’ (Harris 1943), is likely to impact on the effectiveness of any inanimate enrichment (Tarou and Bashaw 2007). For example, successive presentations of enrichment objects to chimpanzees rapidly led to habituation towards those objects over a 3-day period (Celli et al. 2003). Habituation has even been observed when manipulable and play-inducing objects were offered, affecting their usefulness as long-term enrichments (Line et al. 1991; Maki and Bloomsmith 1989 cited in Tarou and Bashaw 2007). Dishabituation is defined as any resumption of the behavioural response, which may be less than or equal to that seen prior to habituation (Holmes 1912 cited in Thompson 2009), that occurs when the sensory characteristics of an object presented to the animal contrast sufficiently with those of the object to which habituation has built up. Such dishabituation could potentially be used to restore interest in object play and is therefore one of the key factors that could affect the value of enrichment using objects.

Object-orientated play in domestic cats is motivated as if it were a type of predatory behaviour (Hall and Bradshaw 1998), with habituation rapidly inhibiting exploration of the object unless its sensory characteristics change, as would happen during successful predatory behaviour. In this species, the dishabituation when the sensory characteristics of the toy are changed can exceed that seen with the first presentation of the first toy (Hall et al. 2002), termed post-inhibitory rebound (Kennedy 1985; Roper 1984). Since the domestic dog also evolved from a predatory species (Coppinger and Schneider 1995), it is possible this species has a similar underlying motivation for object-orientated play to that observed in cats (Hall et al. 2002). Social play in dogs is considered to be a means of learning appropriate social interaction (Feddersen-Petersen 2008). However, solitary play is less common, and its construction and motivation are poorly documented (Feddersen-Petersen 2008; Horwitz et al. 2002).

From an animal welfare perspective, establishing the motivation behind habituation to objects, and the nature of any subsequent recovery in response that restores object-orientated play, should help to predict the effectiveness of inanimate enrichments, the time interval needed between successive presentations of any enrichment for effective or prolonged enrichment, and which cues lead to both habituation and dishabituation.

The study described here aimed to determine the following: the time taken by dogs to habituate to a manipulable, interactive, ‘play’ object over repeated presentations (experiments 1 and 2); the extent of dishabituation when visual and/or olfactory characteristics of the play object were changed (experiment 1); and the effect on dishabituation of changing the interval between habituation and the presentation of the contrasting play object (experiment 3).

General methodology


The study was carried out at the WALTHAM® Centre for Pet Nutrition, Leicestershire. The dogs were housed in pairs, in pens arranged around an octagonal central court (Loveridge 1998). The dogs had constant access to an indoor pen (3.05 m2) with a raised, bed area (facing a central area and other dog pens) and an outdoor covered area (4.15 m2); additional access to an outdoor paddock was available during the day with access to large dog toys in the paddocks such as Aussie hanging balls (Aussie dog products, AU), as well as platforms and staging to climb on. The indoor pen area could be sectioned off using a lockable dog flap. Overnight, pens were provided with fleece bedding and nylon chews, considered safe for unsupervised interaction. In addition to exercise in paddocks with other dogs throughout the day, each dog received at least one half hour of structured interaction with a person per day: this could consist of either a walk with off lead interaction in an enclosed field, interactive play in paddocks with kennel staff, or obedience training. The kennel staff also spent time interacting individually with dogs in their pens or paddocks on an ad hoc basis. Destructible toys were only made available when the dog was interacting with kennel staff due to the risk of destruction and ingestion during unsupervised interaction. The dogs could maintain visual contact with kennel staff throughout the day. All the dogs were clicker trained for basic obedience commands such as ‘sit’ and ‘wait’ on a daily basis. This continued throughout the study period.

Study area

The experiments were undertaken in a room in one of the dog housing areas (approximately 4.2 m × 4.0 m). The dogs were acclimatised to the room prior to the studies as part of their daily walks.

Study subjects

Sixteen adult (1–8 years) Labrador retrievers (LR) (9 neutered males, 4 neutered females, 3 intact females) from 9 litters were randomly chosen from the dogs at the centre. The dogs were normally used for non-invasive feeding trials. All dogs were used for experiment 1. Seven of these dogs, all of which had interacted with and subsequently habituated to the toy, were studied in experiment 2. Fifteen of the dogs were used in experiment 3: one dog that had not shown any interest in the toys in experiment 1 was not tested.

Experiment 1: The effect of visual and olfactory cues on habituation


Experiments were carried out on days 1, 3, and 5 of a five-day period, with a distraction day on days 2 and 4 to reduce the expectation of subsequent experimental days and to minimise any carryover of habituation due to previous treatments. All dogs were presented with experiment 1a on day 1, half the dogs were given experiment 1b on day 3 and then 1c on day 5, and the other half were given 1c on day 3 and then 1b on day 5.

Experiment 1a: Visual and olfactory cues

Two manipulable objects were used for the experiment on day one: identical soft zoomorphic toys (20 cm × 15 cm × 8 cm; Chubleez ‘Plush Squeaky Sniffer Dog Toy’, UK) with an internal squeaker that sounded when mouthed by the dog, varying only in colour: one was blue and one brown/yellow. The dogs were taken to the test room individually and given 2 min to settle. The experimenter, an unfamiliar female, 26 years, then placed the first toy on the floor, 0.5 m from the door (Presentation 1.1, hereafter abbreviated P1.1) and left the room. After 30 s, she removed the toy from the test room, ending P1.1. Ten seconds later, the same toy was presented again in exactly the same way as before (P1.2). P1.2 was then repeated until the dog no longer interacted with the toy for more than the first 10 s of the presentation: this was taken to indicate the point at which the dog had habituated to the first toy and was designated the final presentation of toy 1, P1F. This toy was then removed; after a 10-s delay, the second toy was placed on the floor (P2). After 30 s, this toy was removed and the experiment ended. If any dog showed no interest in toy 1 after the first three presentations, or the dog continued to interact with toy 1 after 10 presentations and therefore had failed to habituate, the experiment was terminated at that point.

The experiment was performed on all 16 dogs on day 1, using fresh toys for each dog. Eight of the dogs were presented with a blue toy first and eight dogs with a brown/yellow toy to balance any effect of preference for toy colour. The toys were washed prior to the start of the experiment (and between all subsequent test days) using the washing powder used to wash the dogs’ bedding. Any handling of the toys was carried out using rubber gloves to minimise any preference due to deposited human scent.

The toys and test room were cleaned between every dog studied to avoid any effects of scent from the preceding dogs.

The dogs were then randomly allocated to one of two groups, so that half received experiment 1b on day 3 and experiment 1c on day 5 while the other dogs received experiment 1c on day 3 and experiment 1b on day 5.

Experiment 1b: Olfactory cues

Experiment 1a was repeated, except that both toy 1 and toy 2 were the same colour. This restricted the contrast, between the last presentation of toy 1 (P1F) and the only presentation of toy 2 (P2), to the absence of the dogs’ own olfactory cues that were assumed to have accumulated on the toy through saliva, etc., during the repeated presentations of toy 1.

Experiment 1c: Visual cues

Experiment 1a was repeated, but the dogs were presented with a different toy of the same colour every time toy 1 was presented, followed by a clean toy of the opposite colour once habituation had been reached (P2).

Distraction days

The dog was taken to the test room and given 2 min to settle. The experimenter entered the test room, but no toy was given. After 30 s, the experimenter entered the room in the same way as she had done when removing the toy and then exited the room. This was repeated a further 10 times to reduce the anticipation of a toy being presented every time the experimenter entered the room.

Data recording

Interactions with the toys were recorded using remote video cameras mounted above the test room. An interaction was defined as anything other than sniffing or accidental contact, so included contact with the mouth or paw, such as mouthing, chewing, and pawing at the toy. Intra-(AJP) and inter-observer reliability (AJP, JWSB) had been established for this measure in the previous studies using dogs at the same establishment (Pullen 2011; Pullen et al. 2010, 2012).

The video recordings were analysed using Observer 5.0 (Noldus, Wageningen) to determine total duration of interaction with the toy during each presentation (seconds) and the number of presentations to habituation.

Statistical analysis

Data were analysed using SPSS 14.0 (SPSS Inc, Chicago). None of the data were normally distributed, so hypotheses were examined using nonparametric tests. The dog that failed to habituate was excluded from further analysis.

Proportions of dogs reaching criterion for habituation were compared between experiments by a Cochran Q-test. A Friedman test was applied to compare the effect of treatments 1a, 1b, and 1c on the number of presentations of toy 1 to habituation. Wilcoxon tests were then used to look at pairwise comparisons between the three treatments. Recovery (defined as the difference in duration of interaction between P1.1 and P2) and dishabituation (difference in duration of interaction between P1F and P2) were compared between the three treatment groups (1a, 1b, and 1c) using Kruskal–Wallis tests. The magnitude of the recovery was tested against a null hypothesis of zero with Wilcoxon tests.


The number of dogs habituating within 10 presentations was similar for all three treatments (69, 63, and 50 %: Cochran Q = 0.5, P > 0.50). One dog showed no interest in the toys (Table 1).
Table 1

Habituation during the three trials in experiment 1 for all 16 dogs










No habituation




No interest




‘Habituated’ dogs stopped interacting with the toy within the first 10 s of a presentation between the second and the tenth presentation of toy 1. ‘No habituation’ dogs did not stop interacting within the first 10 s of any of the 10 presentations. ‘No interest’ dogs did not interact with the toy for more than 10 s during any of the first three presentations

For the dogs that habituated in one or more treatments (N = 11), the median number of presentations to habitation was 5.0 when colour and olfactory cues were altered together (1a), and 4.0 when they were altered independently (1b and 1c). No overall difference was apparent between treatments (Friedman chi-squared = 2.31, P < 0.32). All pairs of treatments were then compared to assess whether some were more similar than others (noting that treatments 1a and 1b were identical up to the point of habituation), but the distribution of P values appeared to be random (experiment 1a vs. 1b, Wilcoxon T = 43, P < 0.21; experiment 1a vs. 1c, T = 36, P < 0.42; experiment 1b vs. 1c, T = 48, P < 0.10).

The degree of dishabituation (difference in duration of interaction between P1F and P2) was similar across the three treatment groups (Kruskal–Wallis = 0.81, P < 0.67) (Fig. 1). Recoveries, the differences between durations of interaction during the first presentations of the two toys (P1.1 and P2), were also similar for all three treatments (Kruskal–Wallis = 0.67, P < 0.72). Treatments could therefore be combined by averaging within dogs. The average dishabituation was highly significant (Wilcoxon T = 65, P < 0.001). The median of the average recovery was 4.0 s, which is not significantly different from zero (Wilcoxon T = 40, P < 0.29), indicating that play returned to approximately its original intensity after the toy was changed.
Fig. 1

Boxplots of duration of interaction (s) with toys at presentation 1.1 (first presentation of toy 1), presentation 1F (final presentation of toy 1 at the point of habituation), and presentation 2 (presentation of toy 2, following habituation to toy 1), by the dogs that habituated within 10 presentations (N = 11, 10, and 8 for experiments 1a, 1b, and 1c, respectively) for visual and olfactory contrast between 1F and 2 (a), visual contrast only (b), and olfactory contrast only (c). Heavy horizontal lines indicate medians, boxes indicate 25th and 75th percentiles, and error bars indicate minima and maxima


All three parts of this experiment confirm that dogs habituate rapidly to manipulable objects and that changes in the sensory characteristics of the objects lead to immediate resumption of play. The extent of the dishabituation shows that the dogs’ motivation for object play had not diminished to any great extent over any of the treatments. Changes in colour and (presumed) odour cues appeared to be equally effective in causing dishabituation; this suggests that the alteration in any cue, whether visual or olfactory, may be sufficient to induce dishabituation. Habituation, therefore, appears to occur in parallel to each of the stimulus properties of the toy, rather than any particular sensory modality taking precedence. A remote possibility that has to be considered is that habituation was primarily driven by cues other than colour or odour, although it is unclear what these could be, and this would not explain why dishabituation is almost complete when only colour and/or odour is changed.

Thompson (2009) suggests that whenever regular presentation of a stimulus occurs, habituation will become evident. All mammals able to exhibit habituation to a stimulus appear capable of showing dishabituation (Harris 1943 cited Thompson 2009). By viewing dishabituation as the ‘neutralisation’ of the habituation, a distinction can then be made between that and ‘fatigue’ or loss of motivation (Humphrey 1933 cited in Thompson 2009). In rats, habituation was found to occur when presented with either auditory or tactile cues. However, generalised habituation was also apparent but not clearly distinguishable between increased habituation and decreased sensitisation to the stimulus (Vogel and Wagner 2005).

Over time, habituation can become more rapid through training and recovery (Thompson 2009) or simply through repeated presentation of unchanging objects (Tarou and Bashaw 2007). This was not evident in the experiment reported here, possibly because each dog was only subjected to 3 experimental periods in experiment one, and a combination of distraction days, delays between experiments (of a number of weeks), and randomisation of trial orders within experiments were all employed to negate these potential effects.

The almost complete dishabituation that occurred when the toy presented was visually identical to the toy to which the dog had become habituated was unexpected. It is highly unlikely that any small differences in visual appearance between the two toys provided the salient contrast, since short-range vision is poorer in dogs than it is in humans (Miller and Murphy 1995). While minor textural differences cannot be ruled out, the most likely explanation for the dishabituation is the contrast in odour that arose from a clean toy being substituted for a toy that had been repeatedly chewed by the dog and therefore would have smelled of its own saliva. This highlights both the dog’s highly sensitive olfaction (Walker et al. 2006) and the salience which dogs attach to what might seem to primates such as humans to be trivial differences in odour.

Furthermore, it emphasises that consideration should be given to interpretations based upon contrasts in odour in experiments that involve objects that could have been contaminated by odours that are detectable by dogs but not by humans. Such changes in odour could result from contact with either dogs or humans. For example, in the movie clip presented by Kaminski et al. (2004), the subject dog repeatedly sniffs and mouths the novel object in the set before being given the novel cue word to retrieve it. He also sniffs the familiar objects in the set; presumably, these emit odours derived from himself and his owner. His behaviour suggests that the unfamiliar odour of the new object makes it particularly salient and therefore likely to be retrieved spontaneously, unless overridden by a command previously associated with a specific object.

Experiment 2: The effect of time interval between presentations on habituation


The same two types of toy used in experiment 1 were used for experiment 2. The method of presentation used in experiment 1a (the same toy for Presentation 1 followed by the other colour toy for P2) was used since it had caused the greatest number of dogs to habituate in experiment 1 and was therefore considered the most effective combination to lead to habituation in future experiments.

The dog was taken to the test room and given 2 min to settle. It was then presented with the first toy (P1.1). After 30 s, the toy was removed from the room. After a fixed time interval that varied between treatments, the same toy was presented again (P1.2). This continued until the dog no longer interacted with the toy for more than the first 10 s of the presentation (P1F) (up to 10 presentations). At this point, the toy was removed. Following a 10-s interval, the second toy (P2) was presented. After 30 s, the toy was removed, signalling the end of the experiment.

Each dog was studied on odd-numbered days over a 9-day period. The treatments consisted of five different delay intervals (either 10, 30 s, 1, 3, or 10 min) arranged in a balanced order for each dog using a Latin square design. On each day, the same delay was inserted between all presentations, until habituation occurred. Distraction, as for experiment 1, was performed on the days between each study day.

Comparisons between the number of presentations to habituation were made between the five treatments (time intervals) using a Friedman test. Recovery and dishabituation were compared between the five time intervals using Kruskal–Wallis tests.


The median number of presentations of toy 1 to habituation was 4.0 for the 10, 30 s, 3, and 10 min between-presentation intervals, and 3.0 for the 1-min interval: overall, no effect of interval was apparent (Friedman chi-squared = 2.48, P < 0.65). The average durations of interaction for habituation and dishabituation (Fig. 2) were very similar to those in experiments 1a–c (Fig. 1).
Fig. 2

Boxplot of duration of interaction (s) with toys at presentation 1.1 (first presentation of toy 1), presentation 1F (final presentation of toy 1 at the point of habituation), and presentation 2 (presentation of toy 2, following habituation to toy 1) (means for all 6 time intervals combined) by the seven dogs that habituated within 10 presentations in experiment 2


Habituation over short time intervals appeared to be insensitive to the time between presentations. From the consistent median of four presentations to habituation for all time intervals of 10 min or less, dogs seem to have a pattern of habituation that is insensitive to the temporary absence of the stimuli that are becoming habituated, indicating that they remember those precise characteristics for at least 10 min. Thompson (2009) suggests, from a review of earlier studies by Harris (1943), Glanzer (1953), and Welker (1961), that ‘the more rapid frequency of stimulation, the more rapid and/or more pronounced the habituation’. In this study, even when the inter-stimulus intervals were brief, neither was found to be the case. This lack of sensitivity to interruption is markedly different to that observed in ‘classic’ habituation experiments. It is unlikely that ‘ceiling’ effects played a part, since few dogs played with the object for the full 30 s even on the first presentation. Thus, it is possible that object play behaviour in dogs is terminated by mechanisms that are different from, even if superficially similar to, habituation: the same may apply to object play in another member of the Carnivora, the domestic cat Felis silvestris catus (Hall et al. 2002). Extending the time interval between presentations to 15 min or longer would presumably eventually increase the number of exposures required to reach the habituation criterion. However, if the time period between each presentation was very long, the dogs might not associate the presentations together as belonging to the same play session, therefore potentially introducing a different motivational basis for any change in response compared to the shorter time periods. This might be possible to investigate further using distractions (e.g. feeding or social play) during the intervals between presentations of the toys.

Experiment 3: The effect of delay on dishabituation


The procedure was identical to experiment 2, except that the interval between each presentation of toy 1 (P1) was fixed at 10 s, and the treatments consisted of varying interval between the final presentation of toy 1 (P1F) and the presentation of toy 2 (P2) (10, 30 s, 1, 5, 10, or 15 min).

Each dog was studied on odd-numbered days over an 11-day period, with treatments arranged in a balanced order for each dog using a Latin square design. Each study day was followed by a distraction day, as per experiment 1.

Recovery and dishabituation were also compared between the six time intervals of the experiment 3 data using Kruskal–Wallis tests.


The median number of presentations of toy 1 to habituation was 4.5. Averaging across all intervals, the durations of play for the first presentations of the two toys (P1.1 and P2) were similar to each other and to those in experiment 1 (Fig. 3). Varying the time interval (10 s up to 15 min) between the final presentation of toy 1, that is, once habituation had been reached, and the presentation of P2 did not affect either the recovery (Kruskal–Wallis = 1.56, P < 0.91) or the dishabituation (Kruskal–Wallis = 6.05, P < 0.30).
Fig. 3

Boxplot of duration of interaction (s) with toys at presentation 1.1 (first presentation of toy 1), presentation 1F (final presentation of toy 1 at the point of habituation), and presentation 2 (presentation of toy 2, following habituation to toy 1) (means for all 5 time intervals combined) by the 12 dogs that habituated within 10 presentations in experiment 3


Dishabituation appeared to be insensitive to the duration of the delay since habituation. However, the relatively short maximum time interval used (15 min) may not have been sufficient to induce an observable effect. Dishabituation in cats did not diminish until the delay reached 25 min (Hall et al. 2002), so the same response might occur in dogs if the time interval was extended beyond 15 min. The large post-inhibitory rebound occurring in cats when the time interval was reduced to 5 min, such that play became more intense than at the first presentation, was not observed in the dogs at this or any shorter time interval, suggesting that both habituation and dishabituation during object play may be qualitatively different in dogs as compared to cats (Hall et al. 2002). However, several dogs played for almost the whole of the 30 s available in P1.1 (see Fig. 3), and therefore, any increase in play when P2 was presented would have only been detectable for the dogs that did not play throughout P1.1.

The motivation behind play is likely to be a contributing factor in the extent of dishabituation. In cats, object play appears to be a form of redirected hunting behaviour, which can lead to dishabituation that reaches a greater level than the initial play interaction (Hall et al. 2002). In contrast, object play in dogs may be an extension of juvenile play, resulting from the neotenisation of dogs through the domestication process (Bradshaw and Brown 1990; Frank and Frank 1982). This difference between the domestic cat and the domestic dog in their putative motivations for play may explain the differences observed between the species in their responses to dishabituation.


The almost complete dishabituation observed in both these experiments when superficially trivial aspects of the toys were changed is consistent with the powerful neophilia for toys demonstrated by Kaulfuß and Mills (2008). However, the unexpected lack of sensitivity to interruption of both habituation and dishabituation suggests that more complex inhibitory mechanisms may be involved in this species, apparently reinforcing the effects of ‘classical’ habituation.

Since the results obtained here are broadly similar to those found for domestic cats, it is clear that such habituation and dishabituation are common phenomena in the object play of adult domesticated carnivores. Whether this also applies to juveniles, which engage in more play than adults do, remains to be tested.

More generally, our findings may be usefully applied to environmental enrichment for kennelled dogs. It may be possible to rekindle interest in objects provided to stimulate play by altering any cue from the toy that is perceptible by the dog, so long as the toy remains a desired play object and motivation to play has not diminished (Celli et al. 2003; Tarou and Bashaw 2007). If the latter is the case, play is unlikely to restart, at least within 30 min or so, regardless of the changes made.


The authors would like to thank the BBSRC and WALTHAM® Centre for Pet Nutrition for funding the project (studentship to AJP). We are also grateful to WCPN for providing the facilities and dogs to enable the research to be undertaken.

Conflict of interest

Ralph Merrill is an employee of the WALTHAM® Centre for Pet Nutrition. Anne Pullen was in receipt of a studentship partially funded by the WALTHAM® Centre for Pet Nutrition. John Bradshaw has no financial relationship with either of the bodies that sponsored the research.

Ethical standards

Approval was obtained from the University of Bristol Ethics Committee for all the experiments described in this paper. As part of this process, confirmation was obtained that none of the procedures used required a licence from the UK Home Office.

Copyright information

© Springer-Verlag 2012