Behavioral Flexibility in Dogs: Adaptation, Learning, and Problem Solving

1. Introduction

1.1 Defining Behavioral Flexibility

Behavioral flexibility refers to the capacity to modify behavior when environmental contingencies change – when a previously successful strategy stops working, when new information becomes available, or when the rules governing an outcome are altered. It is generally understood to encompass several related processes: adapting to new rules, switching between problem-solving strategies, inhibiting behaviors that are no longer effective, incorporating new information into ongoing behavior, and generalizing learned strategies across contexts.

Behavioral flexibility is closely related to, but conceptually distinct from, behavioral persistence – the tendency to continue an established behavior despite changing or ambiguous feedback. Persistence is not inherently maladaptive: maintaining a successful strategy in the face of occasional non-reinforcement is often the correct response, as discussed in the context of reinforcement schedules and extinction resistance (see Reinforcement Schedules in Dogs). Adaptive behavior depends on an appropriate balance between persistence (maintaining effective strategies, tolerating short-term non-reward) and flexibility (abandoning strategies that have genuinely stopped working, in favor of new ones). Behavioral flexibility, in this framing, is not simply "more change is better" but rather the capacity to detect when change is warranted and to implement it efficiently.

1.2 Why Flexibility Matters for Dogs

Two considerations make behavioral flexibility particularly relevant to dogs. First, domestic dogs live in environments substantially structured by humans – households, training routines, work assignments – in which the rules governing reinforcement, social interaction, and acceptable behavior can and do change, sometimes deliberately (as training progresses) and sometimes inadvertently (as household routines or handler behavior shift). A dog's capacity to adapt to these changes efficiently has direct practical consequences for training outcomes and for the quality of the human-dog relationship.

Second, behavioral flexibility – or its absence – has been proposed as a unifying lens for understanding a range of behavior problems. Many presentations seen in behavior therapy, including certain forms of reactivity, anxiety-related behaviors, compulsive-like behaviors, and frustration-related problems, can be described, at least in part, as a difficulty in updating behavior despite changing circumstances – the dog continues to respond to a situation in a fixed way even when that response is no longer adaptive or when the situation itself has changed. This framing does not replace more specific diagnostic or functional accounts of these problems, but it offers a complementary perspective that links them to a broader cognitive construct with its own evidence base.

1.3 Evolutionary and Domestication Context

The evolutionary value of behavioral flexibility is most plausible in environments characterized by variability and unpredictability – where fixed, inflexible response patterns would be poorly suited to changing food availability, social structures, or environmental conditions. For the ancestors of domestic dogs, and for dogs throughout the process of domestication, flexibility in foraging strategy, social behavior, and responsiveness to novel environmental conditions plausibly conferred adaptive advantages.

Domestication itself may have specifically shaped certain forms of flexibility. The capacity to adjust behavior in response to human social signals – a domain extensively documented in the social learning and gesture-comprehension literature (see Social Learning in Dogs and Dogs and Human Gestures: How Dogs Understand Us) – can itself be understood as a domain-specific form of behavioral flexibility: the capacity to update behavior based on a particular class of environmental information (human communicative signals) that would have been less relevant to the dog's wild ancestors. Whether domestication has produced general enhancements to behavioral flexibility, as opposed to domain-specific enhancements concentrated in human-directed contexts, has not been directly tested and remains an open question.

2. Theoretical Framework

2.1 Cognitive Mechanisms: Inhibitory Control, Working Memory, and Attention Shifting

Behavioral flexibility is generally understood not as a single unitary ability but as an emergent property of the interaction between several more basic cognitive processes, often grouped under the umbrella of executive function.

Inhibitory control refers to the capacity to suppress a response that is currently prepotent – the "default" or previously reinforced response – in favor of an alternative. Behavioral flexibility necessarily involves inhibitory control: before a new behavior can be implemented in place of an old one, the old behavior must be suppressed. A dog that has learned to approach a particular location for food, and for whom that location is no longer rewarded, must inhibit the learned approach response before an alternative response can be expressed. Failures of behavioral flexibility are, in many cases, failures of inhibition rather than failures to learn the new contingency itself – the dog may "know," in some sense, that the old location is no longer rewarded, but continues to respond to it because the previously learned response has not been adequately inhibited. This distinction – between failure to learn and failure to inhibit – is central to the interpretation of reversal learning studies discussed in Section 3.

For a detailed discussion of inhibitory control and its neurobiological basis in dogs, see Prefrontal Cortex and Self-Control in Dogs.

Working memory refers to the capacity to hold and manipulate information over short time periods in the service of ongoing behavior. Flexible behavior often requires working memory: tracking which strategies have recently been successful or unsuccessful, holding in mind the current rule governing an outcome, or comparing current information against recently experienced alternatives. The relationship between working memory capacity and behavioral flexibility has been extensively studied in humans, where working memory limitations are a significant predictor of flexibility deficits; the corresponding relationship in dogs has received comparatively little direct study.

Attention shifting refers to the capacity to redirect attention from one source of information to another – for instance, from a previously relevant stimulus dimension (such as the location of an object) to a newly relevant one (such as its color or shape). Attention shifting is closely related to, and in some theoretical frameworks considered a component of, inhibitory control: shifting attention away from a previously relevant stimulus dimension requires inhibiting the attentional bias toward that dimension.

2.2 Prediction Error as a Driver of Flexibility

The prediction error framework – discussed extensively in the context of reinforcement learning (Schultz et al., 1997; see Prediction Error in Dogs: The Core Mechanism of Learning) – provides a mechanistic account of why and when behavioral flexibility is engaged. When an organism's expectations about the relationship between behavior and outcome are violated – when a previously reinforced response is no longer reinforced, or when a previously unreinforced response begins to be reinforced – this generates a prediction error signal. According to this framework, prediction error signals function as a teaching signal that drives the updating of behavior: large or persistent prediction errors signal that the current behavioral strategy is no longer appropriate and that alternative strategies should be considered.

This framework integrates naturally with the broader account of behavioral flexibility. A change in environmental contingencies – the classic trigger for flexible behavioral adaptation – is, from the prediction error perspective, simply a situation in which previously accurate predictions become inaccurate, generating the prediction errors that drive behavioral updating. The rate and magnitude of behavioral change following a contingency shift can therefore be understood, in part, as a function of how prediction error signals are generated, weighted, and used to update behavior – processes that are themselves subject to individual variation, developmental change, and modulation by stress and arousal, each discussed in subsequent sections.

3. Evidence in Dogs

3.1 Reversal Learning: The Primary Experimental Paradigm

The dominant experimental paradigm for studying behavioral flexibility in dogs, as in many other species, is reversal learning. In a typical reversal learning task, a dog first learns a discrimination – for instance, that one of two locations, objects, or stimuli (A) is associated with a food reward while the other (B) is not. Once this discrimination is reliably learned, the contingencies are reversed without warning: the previously unrewarded option (B) now produces the reward, while the previously rewarded option (A) no longer does. The dog's task is to detect this change and shift its responding accordingly.

Reversal learning performance is typically quantified by the number of trials required to reach a new learning criterion following the reversal, and by the pattern of errors made during this adjustment period – particularly perseverative errors, in which the dog continues to respond to the previously rewarded option despite its no longer being reinforced. Reversal learning is widely considered to index a combination of inhibitory control (suppressing the previously learned response) and new learning (acquiring the new contingency), and the relative contribution of each is a recurring theme in the interpretation of reversal learning data.

3.2 Findings on Age and Cognitive Decline

The most substantial body of dog-specific reversal learning research concerns age-related change. Piotti et al. (2018) tested 107 pet dogs, divided into younger (2.5–6.5 years) and older (8–14.5 years) groups, on discrimination and reversal learning tasks using two spatial paradigms – one based on the location of stimuli and one based on their visual characteristics (e.g., color, shape). Younger dogs learned both the initial discrimination and the subsequent reversal significantly faster than older dogs. A notable methodological finding was that the location-based version of the task was more effective for studying reversal learning in older dogs, since most older dogs failed to reach the learning criterion within the trial cutoff on the characteristic-based version – an important consideration for the design of future studies and for clinical assessment tools.

Wallis et al. (2016), studying a cohort of 95 pet Border Collies ranging from 5 months to 13 years of age using touchscreen-based discrimination, reversal learning, and reasoning-by-exclusion tasks, similarly found that cognitive flexibility – as indexed by reversal learning performance – declined with age, consistent with findings in other domains of canine cognitive aging (see Cognitive Dysfunction Syndrome in Dogs). As a single-breed study, the generalizability of these specific findings to other breeds has not been established, though the broad pattern of age-related decline is consistent with the more breed-diverse sample examined by Piotti et al. (2018).

A further refinement comes from a serial reversal learning study (cited in the broader aging literature) that found age did not straightforwardly predict overall learning rate but that both older and younger dogs showed longer streaks of perseverative errors at certain points during testing, suggesting that age-related cognitive decline in dogs may be better characterized by intermittent episodes of impaired performance ("bouts" of cognitive dysfunction) rather than a uniform, continuously declining trajectory. This finding has practical implications for the assessment of age-related cognitive decline: a single testing session may not adequately capture an individual dog's typical level of functioning, and repeated or longitudinal assessment may provide a more accurate picture.

3.3 Findings on Stress, Arousal, and Cognitive Bias

A related body of evidence comes from the cognitive bias (judgment bias) literature, which, while not using reversal learning paradigms directly, examines a related construct: how an individual's affective state influences its interpretation of and response to ambiguous information – itself arguably a form of behavioral and cognitive flexibility under uncertainty. Mendl et al. (2010) found that dogs exhibiting separation-related behavior showed a more "pessimistic" cognitive bias – responding to ambiguous cues as though they predicted the less favorable outcome – compared to dogs without such behavior, an early and influential demonstration that affective state and cognitive interpretation are linked in dogs in ways that parallel findings in other species.

Subsequent work has extended this line of research, including studies examining the relationship between training methods and cognitive bias (with at least one study finding that dogs trained using a higher number of aversive methods showed more pessimistic biases) and methodological work examining the reliability and replicability of cognitive bias measures in dogs, which has highlighted substantial individual variability and measurement challenges (discussed further in Section 3.4).

While cognitive bias and reversal learning are not identical constructs, both are sensitive to affective state and stress, and both have been proposed as indicators of the broader relationship between emotional state and flexible cognition discussed in Section 6.

3.4 Methodological Considerations and Limitations of the Current Evidence

Several methodological issues affect the interpretation of the dog-specific behavioral flexibility literature and should be made explicit.

Reliance on a narrow set of paradigms. The large majority of direct dog-specific evidence on behavioral flexibility comes from reversal learning tasks, frequently in spatial or simple visual discrimination formats. While reversal learning is a well-validated paradigm with a long history in comparative cognition, it represents one operationalization of behavioral flexibility among several possible ones, and the degree to which reversal learning performance predicts flexibility in other domains – social behavior, problem solving with novel objects, adaptation to changed household routines – has not been directly tested in dogs.

Confounding of inhibition and learning. As noted in Section 2.1, reversal learning performance reflects a combination of inhibitory control (suppressing the previously rewarded response) and new learning (acquiring the new contingency). Most dog studies do not fully disentangle these components, meaning that a dog's relatively poor reversal learning performance could in principle reflect a specific deficit in inhibition, a specific deficit in new learning, or both, with different implications for each.

Individual variability and measurement reliability. The cognitive bias literature, in particular, has highlighted substantial individual variability in dog performance and raised questions about the test-retest reliability of some measures – an issue directly relevant to behavioral flexibility research more broadly, since one-time testing (the norm in most dog cognition studies) may not adequately characterize an individual's typical performance, particularly given the evidence (Section 3.2) that performance may vary across testing occasions even within the same individual.

Heavy reliance on a small number of research groups and paradigms. As with several other areas of canine cognition research discussed elsewhere on this site, the dog-specific reversal learning literature is concentrated among a relatively small number of research groups, predominantly in Europe, using broadly similar touchscreen or apparatus-based paradigms. This concentration does not invalidate the findings but means that the generalizability of specific quantitative findings (e.g., precise age-related decline rates) across different populations, breeds, and testing contexts has not been extensively cross-validated.

Translation from laboratory paradigms to everyday flexibility. Perhaps most importantly, the relationship between performance on structured reversal learning tasks – which involve a clearly defined, single contingency reversal in a controlled testing environment – and the kind of behavioral flexibility relevant to everyday life, training, and behavior therapy (adapting to multiple, often poorly-defined and gradually shifting contingencies across complex social and physical environments) has not been directly established. The applied relevance of laboratory reversal learning findings to training and behavior therapy contexts (Sections 9 and 10) should therefore be understood as a reasonable extrapolation rather than a directly demonstrated link.

4. Behavioral Flexibility and Problem Solving

While reversal learning has dominated the dog-specific literature on behavioral flexibility (Sections 3.1–3.4), a related body of research has examined how dogs approach novel physical problems – tasks requiring the discovery, rather than the updating, of a behavioral solution. This research provides a complementary perspective: where reversal learning asks how readily a dog abandons a previously correct response, problem-solving research asks how a dog explores, persists, and innovates when no previously correct response exists at all.

4.1 Means-End Tasks and Physical Cognition

A long-running line of research has examined whether dogs understand the physical, causal relationships between objects – for instance, whether pulling a string will bring an attached reward within reach. Early studies using string-pulling tasks found that dogs performed poorly, appearing to rely on a simple proximity rule (approaching whichever string end was closest to the reward) rather than on an understanding of the connection between string and reward. Range, Hentrup, and Virányi (2011), however, found that dogs could solve a related "support problem" – choosing between a board with a reward placed on top of it versus a board with a reward placed beside it – even when the proximity rule would have led them astray, suggesting that dogs can attend to means-end relationships under at least some task conditions. Subsequent work (Riemer et al., 2013; Müller et al., 2014) found that dogs' performance on these tasks is sensitive to specific perceptual and task features, with dogs sometimes succeeding through attention to connectivity and sometimes falling back on proximity-based strategies depending on how the task is structured.

For behavioral flexibility, the relevance of this literature lies less in the specific question of means-end understanding and more in what it reveals about strategy use under uncertainty: dogs appear to have access to multiple possible strategies (proximity-based, connectivity-based) for approaching a novel physical problem, and which strategy is deployed depends on the specific features of the task and, plausibly, on the individual dog's prior experience. This is consistent with the broader framing of behavioral flexibility as involving the availability and appropriate selection of alternative strategies, rather than the presence or absence of a single "correct" response.

4.2 The Unsolvable Task and the Exploration-Exploitation Boundary

A second relevant paradigm is the "unsolvable task," in which dogs first learn to solve a physical problem (e.g., a container that can be opened to access food) and are then presented with a version of the same apparatus that has been rendered impossible to solve. The behavioral question is how dogs respond to this abrupt removal of a previously available solution: how long they continue attempting the original strategy (a measure of persistence, conceptually related to the perseverative responding discussed in the reversal learning literature, Section 3.1), and whether and when they shift to alternative behaviors – including, distinctively in dogs, gazing toward a nearby human.

This paradigm sits at the boundary between exploration and exploitation: continuing to exploit a strategy that has worked before (here, the previously successful manipulation) versus exploring alternatives (manipulating the apparatus differently, or seeking information/assistance from a human partner) when the original strategy fails. The finding that dogs' tendency to shift toward human-directed gazing in this paradigm is influenced by factors including age and experience, living situation, and prior success on the task (Marshall-Pescini et al., and the broader literature reviewed in Mendes et al., 2021) indicates that this exploration-exploitation shift is itself subject to individual and experiential variation – directly paralleling the individual difference factors discussed for reversal learning in Section 8.

A point of caution regarding interpretation is warranted here, and connects directly to the discussion of behavioral suppression elsewhere in this collection (see Learned Helplessness in Dogs): a dog that quickly stops attempting to solve an unsolvable task could, in principle, be displaying either an efficient, flexible shift toward an alternative strategy (seeking human assistance) or a more generalized reduction in persistence that is harder to distinguish from giving up. The unsolvable task literature has generally interpreted human-directed gazing as the former – an active, communicative strategy shift – but the broader point, that reduced behavioral output is ambiguous between adaptive flexibility and maladaptive shutdown, applies here as much as it does in the contexts discussed in the learned helplessness literature.

4.3 Functional Fixedness

A concept with a long history in human cognitive psychology, functional fixedness refers to a cognitive bias toward continuing to perceive or use an object or strategy in terms of its previously established function or use, even when this is no longer effective and an alternative use would solve the current problem. A dog that has consistently obtained food from a particular container, location, or manipulation, and that continues to direct effort toward that container, location, or manipulation even after it has stopped producing food – persisting with option "A" specifically, rather than exploring alternatives more generally – would be displaying a pattern consistent with functional fixedness.

Functional fixedness, framed this way, is closely related to but not identical with the perseverative errors documented in reversal learning (Section 3.1): both describe continued responding to a previously effective option after it has stopped being effective, but functional fixedness as a construct emphasizes the cognitive "lock-in" to a specific object or strategy's established function, which may be a useful framing for understanding certain real-world behavior patterns – for instance, a dog that continues to attempt to access a particular resource through a particular previously-successful route or method, even when the route has been blocked and alternative routes are available and have not themselves been tried. Direct, dedicated studies of functional fixedness as such in dogs are, to this author's knowledge, not well developed; the construct is introduced here primarily as a conceptually useful framing that connects the reversal learning literature (Section 3.1) to applied problem-solving contexts, rather than as an area with its own substantial dog-specific evidence base.

4.4 Innovation and Individual Problem-Solving Styles

Across the problem-solving literature more broadly (including substantial work in birds and primates, with more limited direct work in dogs), individuals within a population frequently show consistent differences in problem-solving style – some individuals explore a wide range of strategies relatively quickly ("fast explorers"), while others persist longer with a smaller number of strategies before switching ("slow explorers" or more conservative problem solvers). Whether such stable individual styles exist in dogs, how they relate to the individual difference factors discussed in Section 8 (genetics, early experience, training history, emotional state), and whether they predict performance on reversal learning or other flexibility measures, represents a promising but largely unexplored area for dog-specific research.

5. Research Gaps and Future Directions

Building on the limitations identified in Section 3.4, several specific research gaps deserve explicit attention as priorities for future work.

Direct studies of stress effects on flexibility in dogs. While the relationship between stress and reduced behavioral flexibility is well established in rodent and human research (Section 6), and while the cognitive bias literature in dogs (Section 3.3) is consistent with a relationship between affective state and cognitive processing, direct experimental studies manipulating stress or arousal and measuring subsequent reversal learning or other flexibility paradigms in dogs are, to this author's knowledge, limited. Given the practical importance of this relationship for training and behavior therapy (Section 10), this represents one of the more consequential gaps in the current evidence base.

Ecologically valid flexibility paradigms. The development of paradigms that capture behavioral flexibility in contexts more similar to everyday training and household environments – multi-step problem-solving tasks, tasks involving social information, tasks with gradually rather than abruptly shifting contingencies – would help address the translation gap identified in Section 3.4.

Longitudinal and repeated-measures designs. Given the evidence that flexibility-related performance may vary across testing occasions even within individuals (Section 3.2), longitudinal designs that track individual dogs over time, both within and across developmental stages, would provide a more accurate picture of both individual differences and developmental trajectories than the cross-sectional designs that currently dominate the literature.

Breed and individual difference research with adequate power. The sources of individual variation in behavioral flexibility (Section 8) – genetics, early experience, training history, temperament – have been proposed based on theoretical considerations and on findings from adjacent literatures, but direct, adequately powered studies examining these factors specifically in relation to behavioral flexibility in dogs are sparse.

Integration of neurobiological and behavioral measures. As discussed in Section 6, the neurobiological account of stress effects on flexibility (shifting control from prefrontal to more automatized systems) has substantial support in rodent and human research but has not been directly tested in dogs using combined physiological and behavioral measures. Studies combining cortisol or heart rate variability measures with reversal learning or other flexibility tasks in dogs would help establish whether the rodent/human model applies directly.

6. Underlying Mechanisms and the Effect of Stress

6.1 The Prefrontal Cortex

The prefrontal cortex (PFC) occupies a central role in virtually all contemporary accounts of behavioral flexibility, consistent with its broader role in executive function, decision-making, planning, and behavioral inhibition. In the context of reversal learning specifically, regions of the PFC – particularly orbitofrontal and medial prefrontal subregions in the rodent and primate literature – are implicated in representing the current value of different response options and in updating these representations when contingencies change. Damage to or disruption of these regions in rodent and primate studies reliably produces perseverative deficits: continued responding to a previously rewarded option despite its no longer being reinforced, the same error pattern that defines poor reversal learning performance.

For dogs, the prefrontal cortex has been discussed in detail in the context of self-control and impulse regulation (see Prefrontal Cortex and Self-Control in Dogs). The application of the specific orbitofrontal/medial prefrontal distinctions found in the rodent and primate reversal learning literature to dogs has not been directly investigated, and the functional organization of the canine prefrontal cortex with respect to flexibility specifically remains to be characterized.

6.2 Dopamine, Exploration, and Behavioral Updating

Dopaminergic signaling, discussed extensively in relation to reinforcement learning and prediction error (see Dopamine and Learning in Canine Neurochemistry and Prediction Error in Dogs), is also implicated in behavioral flexibility through at least two related routes. First, as discussed in Section 2.2, dopaminergic prediction error signals are proposed to function as a teaching signal that drives behavioral updating when contingencies change – a more responsive or sensitive prediction error system would, on this account, be expected to support faster behavioral adjustment following a contingency shift. Second, dopaminergic signaling has been linked to exploratory behavior more generally – the propensity to sample novel or currently non-optimal options, which is itself relevant to flexibility insofar as exploration of alternatives is a prerequisite for discovering that contingencies have changed.

The relationship between these two roles – dopamine as a driver of behavioral updating in response to prediction error, and dopamine as a driver of exploration independent of any specific prediction error – is an area of active discussion in the broader neuroscience literature and has not been disentangled in dogs specifically.

6.3 Acute Stress: A Shift from Flexible to Automatized Responding

One of the most consistent findings across the stress and cognition literature, established primarily in rodent and human research, is that acute stress produces a shift in the balance of control over behavior – away from systems associated with flexible, goal-directed behavior (broadly associated with prefrontal cortex function) and toward systems associated with more automatized, habitual responding (broadly associated with striatal circuits). Under this account, an organism under acute stress becomes more likely to fall back on previously established, well-rehearsed response patterns and less likely to flexibly update behavior in response to new information – even when such updating would be adaptive.

For dogs, direct experimental demonstrations of this specific shift using reversal learning or comparable flexibility paradigms under manipulated acute stress conditions are, to this author's knowledge, limited. However, the broader pattern – that dogs under higher arousal or stress show behavior that is less responsive to changing contingencies and more characterized by fixed, often previously-learned response patterns – is consistent with extensive applied observation and with the cognitive bias findings discussed in Section 3.3. The mechanistic account from the rodent and human literature provides a plausible explanation for this pattern in dogs, but should be understood as an extrapolation pending direct dog-specific testing.

6.4 Chronic Stress: Rigidity, Reduced Exploration, and Worse Problem Solving

Chronic stress is associated, across species, with a more persistent shift toward behavioral rigidity: reduced exploration of novel options, a narrower behavioral repertoire, and poorer performance on tasks requiring the updating of established response patterns. This pattern overlaps substantially with findings discussed in the context of chronic stress and learned helplessness (see The Neurobiology of Chronic Stress in Dogs and Learned Helplessness in Dogs). Indeed, the reduced behavioral variability and reduced exploration documented in chronically stressed dogs in the chronic stress literature can be understood, in part, as a manifestation of reduced behavioral flexibility: a narrowing of the behavioral repertoire and a reduced capacity or willingness to sample alternative responses.

This convergence is theoretically important. It suggests that behavioral flexibility is not an isolated cognitive variable but is functionally embedded within the broader stress-response system: chronic activation of stress physiology (Section 5.3 of the chronic stress article; the HPA axis, sustained cortisol elevation) appears to produce, as one of its behavioral consequences, a reduction in exactly the kind of flexible, exploratory, contingency-updating behavior discussed throughout this article. For dogs in behavior therapy contexts with histories of chronic stress, this suggests that interventions aimed directly at building behavioral flexibility (Section 9) may be most effective when implemented alongside, rather than instead of, interventions aimed at reducing the underlying chronic stress burden.

6.5 Anxiety and Reactivity

The relationship between anxiety, reactivity, and behavioral flexibility deserves specific mention given its clinical relevance. Anxious or reactive responses are, almost by definition, characterized by a degree of behavioral fixedness: the dog responds to a trigger in a consistent, often escalating way, with limited apparent sensitivity to information that might, in principle, support an alternative response (e.g., information indicating that the trigger is not, on this occasion, dangerous). Whether this fixedness reflects a general reduction in behavioral flexibility, a domain-specific failure of flexibility concentrated on threat-related stimuli, or simply the operation of a well-established and highly reinforced (through negative reinforcement, in the case of avoidance responses) behavioral pattern that has not been directly challenged, is not fully resolved. The general stress-and-flexibility framework discussed in this section (Sections 6.3–6.4) provides one plausible account – heightened arousal in the presence of the trigger shifts control toward automatized responding – but a complete account would need to integrate this with the specific learning history (often involving substantial negative reinforcement) that typically underlies reactive behavior patterns. For further discussion, see Reactivity in Dogs: A Neurological Perspective and Anxiety in Dogs: Neurobiology.

6.6 Perceived Control as a Determinant of Flexibility

A theme that runs through Sections 6.3–6.5, but that warrants explicit treatment in its own right, is the role of perceived control – an organism's experience of, or expectation regarding, the degree to which its behavior influences environmental outcomes. This theme connects most directly to the discussion of agency in Learned Helplessness in Dogs, where the 2016 reformulation of learned helplessness theory (Maier & Seligman, 2016) proposes that the detection of control – mediated by the medial prefrontal cortex – is the actively learned component that determines whether an organism responds to challenging circumstances with active, flexible engagement or with default passivity.

The connection to behavioral flexibility as discussed in this article is direct, though it has not been empirically tested as such. If an organism's history has established an expectation that its behavior generally produces effects – that exploring alternatives, testing new strategies, and responding to changed contingencies tends to be "worth it" because behavior matters – this expectation may itself support the willingness to explore and update behavior that defines flexibility. Conversely, an organism with a history of experiences in which behavior and outcome were decoupled – the uncontrollable, unpredictable conditions discussed in the learned helplessness literature – may show reduced exploration and reduced behavioral updating not because of an acute stress response in the moment (Section 6.3) or a chronic elevation of stress physiology (Section 6.4) specifically, but because the basic expectation that behavioral variation is informative or productive has been weakened.

This framing suggests that perceived control may operate as a kind of background condition for behavioral flexibility – not a moment-to-moment modulator in the way that acute arousal is, but a more stable, history-dependent factor that determines the baseline propensity to explore and update behavior at all. If this is correct, then experiences that build perceived control – the choice- and agency-focused interventions discussed in the context of learned helplessness – would be expected to support behavioral flexibility not only by reducing stress (Section 6.4) but through this more direct route: by maintaining or restoring the expectation that exploring alternatives and adjusting behavior is a productive use of behavioral effort. This represents a theoretically motivated but currently untested hypothesis, and an integration of the agency-focused and flexibility-focused literatures represents a promising direction for future work.

7. Development Across the Lifespan

7.1 Puppies: High Plasticity, High Exploration

The developmental period of puppyhood is characterized by high neural plasticity and a strong propensity toward exploratory behavior – both broadly consistent with, and plausibly contributing to, high behavioral flexibility during this period. From the perspective of behavioral flexibility specifically, the puppy period can be understood as a developmental window in which the behavioral repertoire is being actively constructed and in which the cost of exploring alternative responses (in terms of foregone reward from a known, reliable response) is relatively low, since few responses have yet become highly established.

This has direct relevance to the sensitive period literature in dogs (see Sensitive Period in Puppies: Brain Development and Behavior): the high flexibility characteristic of this period may represent both an opportunity (experiences during this window may have disproportionate influence on the subsequent behavioral repertoire) and, from the perspective of this article, a developmental baseline against which later changes in flexibility can be understood.

7.2 Adult Dogs: Balancing Experience and Flexibility

Adulthood is generally characterized, across the lifespan literatures of multiple species, as a period of relative balance between the accumulated benefits of experience (well-established, often highly efficient response patterns for familiar situations) and continued capacity for flexible adjustment when circumstances genuinely change. This is consistent with the general finding (Section 3.2) that younger adult dogs outperform older dogs on reversal learning tasks, while themselves likely outperforming very young puppies on tasks requiring the application of established knowledge (though direct puppy-vs-adult reversal learning comparisons using comparable paradigms are limited).

The practical implication is that adult dogs are likely to represent something of an optimal point for many forms of training that require both the application of established skills and adaptation to new requirements – though this should not be taken to imply that puppies or senior dogs cannot learn or adapt, only that the relative balance of these capacities may differ across the lifespan.

7.3 Senior Dogs: Declining Flexibility and Cognitive Aging

The most robust dog-specific evidence on lifespan changes in behavioral flexibility concerns decline in older dogs, discussed in detail in Section 3.2. Piotti et al. (2018) and Wallis et al. (2016) both found reduced reversal learning performance in older compared to younger dogs, consistent with the broader literature on canine cognitive aging.

This decline in behavioral flexibility is closely connected to the broader phenomenon of cognitive dysfunction syndrome (CDS) in aging dogs, in which behavioral flexibility – along with memory, spatial awareness, and social responsiveness – is among the cognitive domains affected. The finding (Section 3.2) that age-related decline in flexibility may manifest as intermittent "bouts" of impaired performance rather than a uniform decline is potentially relevant to the early detection of CDS, since it suggests that owners and clinicians should attend to variability and inconsistency in a dog's responses to familiar versus changed situations, not only to a general overall decline. For a detailed discussion of cognitive aging in dogs, see Cognitive Dysfunction Syndrome in Dogs.

8. Individual Differences

Why do some dogs adapt readily to new rules, environments, or problem-solving demands, while others show persistent difficulty? Several sources of individual variation have been proposed, though – consistent with the methodological discussion in Section 3.4 – the degree to which each has been directly demonstrated to affect behavioral flexibility specifically in dogs (as opposed to being a plausible inference from related findings) varies considerably.

Genetics. Heritable variation in temperament traits associated with novelty-seeking, fearfulness, and general behavioral style plausibly contributes to individual variation in behavioral flexibility, given that these traits are known to vary heritably in dogs and are conceptually related to the exploration and stress-reactivity factors discussed in Sections 6.3–6.4. Direct genetic association studies specifically targeting behavioral flexibility measures in dogs are, to this author's knowledge, not well developed.

Early experiences. Given the sensitive period literature discussed in Section 7.1, early experiences – particularly the degree of environmental variability, novelty exposure, and opportunities for exploration and problem-solving during the socialization period – represent a plausible mechanism by which individual differences in adult behavioral flexibility could arise. This is consistent with the broader principle, well established in other domains of canine development, that early experience shapes later behavioral tendencies, though the specific application to behavioral flexibility as operationalized in this article has not been directly tested.

Environmental complexity. Dogs living in environments that present varied, changing demands – multiple routines, varied social contacts, diverse physical environments – may have more frequent opportunities to practice and therefore develop behavioral flexibility, compared to dogs in highly stable, unchanging environments. This parallels findings from the environmental enrichment literature in other species, where complex and changing environments are associated with improved performance on flexibility-related cognitive tasks.

Training history. The relationship between training history and behavioral flexibility is plausibly bidirectional and is one of the more practically significant individual difference factors discussed in this article. Training approaches that emphasize a single, fixed solution to a problem, with limited variation in cues, contexts, or acceptable response variants, may provide fewer opportunities for the development of flexibility-relevant skills than approaches that incorporate variation and problem-solving (discussed further in Section 9). Conversely, dogs with a training history that has emphasized flexibility-relevant skills might be expected to show better performance on subsequent flexibility tasks – though, as with the other factors in this section, this represents a plausible hypothesis rather than a directly demonstrated finding in the dog-specific literature.

Emotional state. As discussed in Sections 3.3 and 6, an individual dog's current and characteristic emotional state – its baseline level of arousal, anxiety, and stress – is among the factors most directly and consistently linked to flexibility-related performance, both through the acute stress mechanisms discussed in Section 6.3 and through the chronic effects discussed in Section 6.4. Of the individual difference factors discussed in this section, the relationship between emotional state and flexibility is arguably the best supported by both dog-specific evidence (the cognitive bias literature) and the broader cross-species literature on stress and cognition.

9. Implications for Dog Training

9.1 Encourage Exploration and Problem-Solving

Training approaches that incorporate opportunities for the dog to generate, test, and adjust its own behavioral strategies – as opposed to approaches that specify a single correct response and provide no scope for variation – may, based on the theoretical framework developed in this article, support the development and maintenance of behavioral flexibility. Shaping-based training, in which a behavior is built up through the reinforcement of successive approximations and in which the dog's own behavioral variability is a resource rather than an obstacle, is a practical example of an approach that inherently incorporates exploration.

This recommendation should be understood as following from the general theoretical framework (flexibility benefits from opportunities to detect and respond to changing contingencies, which requires behavioral variability to sample) rather than from direct dog-specific evidence that such training produces measurably improved performance on flexibility tasks such as reversal learning – a link that, per Section 3.4, has not been directly established.

9.2 Avoid Over-Reliance on Rigid Routines

Highly rigid training and management routines – in which cues, contexts, and reinforcement contingencies are held maximally constant – may produce highly reliable performance within that specific, unchanging context but may not support the dog's capacity to adapt when context does change (a new location, a new handler, a change in the cue itself). This is related to, but distinct from, the well-established principle of generalization training (Section 9.3): rigid routines may not only fail to promote generalization of a specific behavior but may more broadly fail to provide the dog with experience in adapting to change as a general skill.

The practical balance here mirrors the persistence/flexibility balance discussed in Section 1.1: some degree of routine and consistency is valuable and indeed necessary for efficient learning (as discussed extensively in the context of reinforcement schedules), but routines that are so rigid as to eliminate all variability may come at a cost to the dog's broader adaptive capacity.

9.3 Use Variable Contexts to Support Generalization

Training a behavior across a range of contexts – different locations, different handler positions, different levels of distraction, minor variations in the cue itself – is a well-established principle for ensuring that a trained behavior generalizes beyond the specific context in which it was initially trained. From the perspective of this article, context variation during training can also be understood as providing the dog with practice in a core component of behavioral flexibility: recognizing that a given response remains appropriate despite variation in surrounding conditions, and, where it does not, adjusting accordingly.

9.4 Manage Arousal

Given the relationship between acute stress/arousal and reduced behavioral flexibility discussed in Section 6.3, maintaining a dog's arousal within an appropriate range during training – avoiding both insufficient engagement and excessive arousal – is directly relevant to the dog's capacity to flexibly acquire and adjust behavior during training sessions. A dog in a state of high arousal may show reduced sensitivity to the specific contingencies being trained, falling back instead on highly established response patterns (which may or may not be the response the trainer is attempting to establish). This provides an additional, flexibility-focused rationale for the arousal management principles discussed in Arousal Regulation in Dogs: Neurophysiology and Training.

9.5 Play as a Driver of Behavioral Flexibility

Play behavior occupies a distinctive position with respect to behavioral flexibility, and its inclusion here connects this article to the broader discussion of play development in dogs (see Neurobiology of Play Development in Dogs). Several features of play make it plausibly relevant as a context in which behavioral flexibility is exercised and potentially developed.

First, play is characterized by high behavioral variability: the same play sequence rarely unfolds identically twice, and play partners (whether conspecifics or humans) continuously introduce variation that must be responded to. This provides frequent, naturally-occurring opportunities for the kind of attention shifting and response updating discussed in Section 2.1, in a context where the consequences of a "wrong" response are minimal.

Second, and relatedly, play is characterized by reduced stakes. Unlike many problem-solving or training contexts, the cost of trying an ineffective or unexpected behavior during play is typically very low – there is no lost reward, no negative consequence, simply a shift in the play interaction. This low-cost environment may be particularly conducive to exploration of behavioral alternatives, in a way that more consequential contexts (where errors carry a cost, whether the loss of a reward or an aversive outcome) are not. From the perspective of the exploration-exploitation framing introduced in Section 4, play may function as a context in which exploration is particularly cheap, and therefore particularly likely to occur.

Third, play frequently involves role-switching and rapid behavioral transitions – chasing and being chased, offering and withholding a toy, escalating and de-escalating arousal – each of which requires the kind of inhibition of an ongoing response and shift to an alternative response that is central to the definition of behavioral flexibility (Section 2.1).

Taken together, these features support the hypothesis that play functions, among its other roles, as a context that exercises and potentially supports the development of behavioral flexibility – through repeated, low-stakes practice in exploration, response switching, and adaptation to a continuously changing partner-driven environment. This hypothesis is consistent with the broader ethological and developmental literature on play across species, where play is widely understood to serve a practice function for behaviors and cognitive capacities relevant to non-play contexts. Direct, dog-specific evidence linking play behavior (e.g., frequency, quality, or diversity of play) to performance on behavioral flexibility measures such as reversal learning has not, to this author's knowledge, been established, and this represents a further specific opportunity for future research that would connect two currently separate areas of this collection.

10. Behavioral Flexibility and Behavior Therapy

The framework developed in this article offers a complementary lens for understanding a range of presentations commonly encountered in behavior therapy. Reactivity, certain anxiety-related presentations, compulsive-like behaviors, and frustration-related problems can each, in different ways, be described as involving a degree of behavioral inflexibility: a fixed response pattern that persists despite changing circumstances, despite feedback that the response is not producing the desired outcome, or despite the availability of alternative responses that would be more adaptive.

Framed this way, a component of behavior modification for these presentations can be understood as building behavioral flexibility specifically: providing the dog with experiences in which alternative responses are available, reinforced, and successful, thereby creating the conditions (per the prediction error framework, Section 2.2) under which the previously fixed response pattern can be updated.

This framing should be treated as a complementary perspective rather than a replacement for established functional and diagnostic approaches to these presentations. It does, however, suggest a specific practical emphasis: behavior modification programs that focus exclusively on suppressing an undesired response (without providing and reinforcing alternative responses) may address the surface behavior without addressing what, from this perspective, may be a more fundamental issue – the dog's repertoire of available, accessible alternative responses in the relevant context. Building new behavioral options, rather than only reducing old ones, is consistent with the broader differential reinforcement approaches discussed in the context of reinforcement schedules (see Reinforcement Schedules in Dogs) and is directly relevant to the agency-focused approaches discussed in the context of learned helplessness (see Learned Helplessness in Dogs).

11. Conclusion

Behavioral flexibility – the capacity to adapt behavior in response to changing environmental contingencies – is a foundational construct that connects several of the themes addressed elsewhere in this collection: prediction error and reinforcement learning, the effects of acute and chronic stress, cognitive aging, and the agency-focused approaches to behavior modification discussed in relation to learned helplessness. Rather than being a narrow, specialized ability relevant only to laboratory cognition tasks, behavioral flexibility appears to be functionally embedded within the broader systems governing learning, stress response, and motivation.

The dog-specific evidence base, while informative, remains considerably narrower than the theoretical framework would ideally support. Reversal learning – the dominant paradigm – has provided robust findings on age-related decline (Piotti et al., 2018; Wallis et al., 2016) and has connections to the cognitive bias literature on affective state and judgment (Mendl et al., 2010), but direct dog-specific evidence on the relationship between stress and flexibility, on the sources of individual variation, and on the translation from laboratory reversal learning to the kinds of flexibility relevant to everyday training and behavior therapy, remains limited.

For practitioners, the framework developed in this article supports a coherent set of practical orientations – encouraging exploration, avoiding excessive rigidity, training across varied contexts, managing arousal, and building new behavioral options rather than only suppressing old ones – each grounded in the general theoretical account even where direct dog-specific experimental confirmation is not yet available. As with the other topics in this collection, the responsible application of this framework involves holding both the theoretical coherence of the account and the current limits of the dog-specific evidence in view simultaneously.

Key Insights (Takeaways)

1. Behavioral flexibility – the capacity to adapt behavior to changing contingencies – is best understood not as an isolated cognitive skill but as an emergent property of the interaction between inhibitory control, working memory, attention shifting, and prediction error-driven updating. It is the functional counterpart to behavioral persistence, and adaptive behavior depends on an appropriate balance between the two.

2. The dominant dog-specific evidence comes from reversal learning studies, which have established robust age-related decline (Piotti et al., 2018; Wallis et al., 2016) but which conflate inhibitory and learning components and have not been directly linked to flexibility in everyday training or behavior-therapy contexts. The translation from laboratory reversal learning to applied flexibility should be understood as a reasonable extrapolation rather than a directly demonstrated finding.

3. Stress and arousal are among the best-supported influences on behavioral flexibility, both through acute mechanisms (a shift from flexible, prefrontally-mediated responding to automatized responding under acute stress) and chronic mechanisms (reduced exploration and behavioral rigidity associated with sustained HPA axis activation). This connects behavioral flexibility directly to the chronic stress and learned helplessness literatures discussed elsewhere in this collection.

4. Behavioral flexibility declines with age in dogs, consistent with broader patterns of cognitive aging, though the decline may manifest as intermittent episodes of impaired performance rather than a uniform trajectory – a finding with potential relevance to the early detection of cognitive dysfunction syndrome.

5. Many presentations encountered in behavior therapy – reactivity, certain anxiety-related behaviors, compulsive-like behaviors, frustration problems – can be productively understood, in part, as involving reduced behavioral flexibility: a fixed response pattern that persists despite changing circumstances or available alternatives. This framing supports an emphasis on building and reinforcing new behavioral options, not only suppressing existing ones, though it should complement rather than replace established functional and diagnostic approaches.

6. Physical problem-solving paradigms (means-end tasks, the unsolvable task) complement reversal learning by revealing how dogs balance exploration and persistence when no previously correct solution exists. These paradigms highlight the exploration-exploitation boundary and the concept of functional fixedness – a "lock-in" to a previously successful strategy or object even when it has stopped working.

7. Perceived control and agency, developed in detail in the learned helplessness literature, may function as background conditions that support exploration and behavioral updating more broadly: a stable, history-dependent sense that behavior generally produces effects may be a prerequisite for the willingness to explore alternatives at all. This integration remains largely theoretical in dogs and represents a promising direction for future research connecting these two areas.

References

Maier, S. F., & Seligman, M. E. P. (2016). Learned helplessness at fifty: Insights from neuroscience. Psychological Review, 123(4), 349–367. https://doi.org/10.1037/rev0000033

Mendes, J. W. W., Resende, B., & Savalli, C. (2021). A review of the unsolvable task in dog communication and cognition: Comparing different methodologies. Animal Cognition, 24(5), 907–922. https://doi.org/10.1007/s10071-021-01501-8

Mendl, M., Brooks, J., Basse, C., Burman, O., Paul, E., Blackwell, E., & Casey, R. (2010). Dogs showing separation-related behaviour exhibit a "pessimistic" cognitive bias. Current Biology, 20(19), R839–R840. https://doi.org/10.1016/j.cub.2010.08.030

Müller, C. A., Riemer, S., Virányi, Z., Huber, L., & Range, F. (2014). Dogs learn to solve the support problem based on perceptual cues. Animal Cognition, 17(5), 1071–1080. https://doi.org/10.1007/s10071-014-0739-y

Piotti, P., Szabó, D., Bognár, Z., Egerer, A., Hulsbosch, P., Carson, R. S., & Kubinyi, E. (2018). Effect of age on discrimination learning, reversal learning, and cognitive bias in family dogs. Learning & Behavior, 46(4), 537–553. https://doi.org/10.3758/s13420-018-0357-7

Range, F., Hentrup, M., & Virányi, Z. (2011). Dogs are able to solve a means-end task. Animal Cognition, 14(4), 575–583. https://doi.org/10.1007/s10071-011-0394-5

Riemer, S., Müller, C., Range, F., & Huber, L. (2013). Dogs (Canis familiaris) can learn to attend to connectivity in string pulling tasks. Journal of Comparative Psychology, 127(1), 31–39. https://doi.org/10.1037/a0033202

Schultz, W., Dayan, P., & Montague, P. R. (1997). A neural substrate of prediction and reward. Science, 275(5306), 1593–1599. https://doi.org/10.1126/science.275.5306.1593

Wallis, L. J., Virányi, Z., Müller, C. A., Serisier, S., Huber, L., & Range, F. (2016). Aging effects on discrimination learning, logical reasoning and memory in pet dogs. Age, 38(1), Article 6. https://doi.org/10.1007/s11357-015-9866-x