RESEARCH DOSSIER
Behavioral Ecology and Fitness Consequences of Cognition
Abstract. The evolutionary significance of cognition depends on its consequences in environments where organisms obtain resources, avoid danger, reproduce, and survive. This monograph examines the behavioral ecology of cognitive variation, asking when information acquisition, learning, memory, inhibitory control, flexibility, and problem solving affect ecological performance and components of Darwinian fitness. It integrates behavioral ecology, life-history theory, experimental evolution, comparative cognition, and longitudinal field studies through a critical narrative synthesis of original experiments and peer-reviewed reviews. Case analyses cover learning and nectar collection in bumblebees, spatial memory and survival in mountain chickadees, reproductive trade-offs in great tits, cognition and social organization in Australian magpies, mate acquisition in bowerbirds, and the costs of improved learning in fruit flies. The review distinguishes a favorable laboratory score from ecologically effective behavior, a fitness proxy from measured survival or reproduction, and an observed association from a demonstrated causal mechanism or historical adaptation. A 2025 systematic synthesis of 45 studies and 211 reported cognition–fitness relationships provides a contemporary measure of the field's mixed evidence. The resulting account identifies the environmental contingencies, trade-offs, measurement designs, and inferential standards necessary to evaluate selection on cognition and specifies research priorities for an integrated evolutionary ecology of information processing.
Keywords: behavioral ecology; cognitive ecology; animal cognition; natural selection; Darwinian fitness; individual variation; spatial memory; associative learning; cognitive flexibility; innovation; reproductive success; survival; ecological validity; behavioral syndromes; phenotypic selection; life-history trade-offs; comparative psychology; experimental evolution; fitness consequences of cognition
Introduction: The Ecological Question of Cognition
Behavioral ecology investigates how the behavior of organisms depends on ecological opportunities, constraints, competition, and natural selection. Within this enterprise, cognitive ecology asks how information is acquired, stored, transformed, and used in ways that matter to organisms encountering particular ecological problems. A feeding bird may remember food locations, a pollinator may learn which flowers are rewarding, and a social animal may alter its behavior after observing others. The evolutionary relevance of each capacity depends on what the resulting behavior accomplishes in the conditions of the organism's life. This inquiry extends the comparative study of cognitive mechanisms toward the demographic and genetic processes through which cognition may evolve. Dukas (2004) and Morand-Ferron et al. (2016) articulate the importance and the practical obstacles of this field-based perspective.
The focal research object is the relationship between measurable differences in cognitive traits and ecologically meaningful outcomes, particularly survival and reproduction, within natural or experimentally manipulated populations. Its evidentiary problem has several levels. First, a cognitive test must operationalize a construct with acceptable reliability and validity. Second, the construct must contribute to an action or information-use process within the animal's ecology. Third, the action must affect access to resources, avoidance of mortality, mating, parental investment, or another component of fitness. Finally, investigators must decide whether selection acts on the cognitive trait itself, on a correlated phenotype, or on a broader developmental or ecological configuration. These distinctions are central to the review by Morand-Ferron et al. (2016).
The empirical record resists the simple claim that superior cognitive scores necessarily enhance fitness. An animal may learn faster yet spend more energy or time collecting information. A problem solver may produce a larger clutch but also abandon nests more frequently. Survival may favor a specialized memory system during a food-scarce winter while reproductive output in another season remains influenced primarily by age, opportunity, and social competition. The relevant causal process is therefore conditional: performance matters in relation to specific environmental distributions of rewards, risks, and developmental histories. Evidence from Cole et al. (2012), Mery et al. (2003), and Sonnenberg et al. (2019) gives this problem experimental and field-based substance.
This manuscript is a critical narrative research synthesis. Its sources comprise original experimental reports, observational field investigations, methodological and theoretical contributions, and a separately identified systematic review. Studies are examined for the unit of measurement, plausible mechanisms, ecological context, endpoint, design, and alternative explanations. There has been no new animal experimentation, registered comprehensive search protocol, original pooled effect-size computation, or independent reanalysis of archived participant-level data. Accordingly, the review uses the 2025 systematic evidence synthesis as an external empirical anchor while retaining its own subject-specific interpretive organization. Madden et al. (2025).
Conceptual Foundations: Cognition, Performance, and Fitness
Cognition refers here to processes by which an organism obtains, retains, processes, and uses information to regulate behavior. This working definition includes discrimination, associative and spatial learning, memory retrieval, information updating, and some forms of inhibitory or flexible action selection. It remains agnostic about whether different species implement equivalent performances through identical neural mechanisms. A task score is a measurement of behavior in an experimental setting, whereas a cognitive phenotype is an inferred characteristic whose definition depends on the task, context, and measurement model. A latent general-ability score assembled from several tasks raises different questions from performance on a specific location-memory task. Morand-Ferron et al. (2016).
Ecological performance means an action's effectiveness in relation to a resource or challenge: nectar collection per unit time, successful recovery of stored food, efficient prey identification, avoidance of a predator, acquisition of a mate, or successful provisioning of offspring. These measures are attractive because they connect cognitive measurements to an organism's ordinary environment. Nevertheless, different ecological outcomes occupy different distances from reproductive success. A temporarily increased food intake may contribute to improved body condition, and improved condition may contribute to successful reproduction, but either connection can be modified by disease, predation, weather, competition, and parental strategy. Raine et al. (2008) provide a particularly instructive example of using natural foraging success as a fitness-related proxy.
Darwinian fitness denotes relative genetic contribution to future generations under a specified population and environmental setting. Researchers often estimate its components through survival, mating success, clutch size, fledging success, recruitment, and lifetime reproductive output. Those components are related but cannot be substituted without a causal argument. A male's number of mating partners measures one aspect of sexual success; the same measure does not directly reveal offspring survival or total genetic contribution. For females, clutch size can be higher while nest abandonment diminishes realized offspring production. The great tit experiment of Cole et al. (2012) demonstrates why multiple life-history outcomes should be assessed simultaneously.
At the population level, a selection gradient is the statistical relation between a phenotype and relative fitness, conditional on other measured phenotypes. The magnitude and sign of this relation may change with ecological conditions, trait distributions, age, sex, and social organization. A positive association between a cognitive score and survival can make selection plausible; establishing evolution requires additional information about heritable variation, the persistence of selection, and changes across generations. Phenotypic selection, heritability, adaptive function, and phylogenetic history are distinct inferential targets. The influential separation of proximate and ultimate questions in Tinbergen (1963) continues to discipline such claims.
Historical Development of Cognitive Ecology
Charles Darwin made variation and natural selection central to interpreting behavioral differences, but the integrated modern study of cognition and fitness emerged through several scientific traditions. Natural historians described behaviors in conditions close to an organism's ecology; ethologists developed systematic comparative observation and questions concerning mechanism, development, function, and history; experimental psychologists investigated learning under controlled contingencies; and behavioral ecologists constructed formal models of adaptive decisions. The resulting field inherited both powerful methods and unresolved tensions about whether laboratory tests preserve the biological significance of behavior. Tinbergen (1963) remains a historical landmark because it organizes complementary explanatory questions.
Behavioral ecology's emphasis on decision outcomes brought economics and formal optimization into the study of feeding, competition, and reproduction. The animal's behavior became interpretable as a response to changing payoffs, risk, and information. Learning complicates static optimality because organisms may initially possess uncertain estimates of reward; experience can improve decisions, but collecting information consumes time and may expose individuals to danger. Stephens (1991) analyzed how the rate and regularity of environmental change influence the value of learning, providing a theoretical bridge between ecological payoffs and cognitive plasticity.
Late twentieth- and early twenty-first-century comparative studies connected brain measures, innovation, behavioral flexibility, and establishment success across species. In birds introduced into novel environments, Sol et al. (2005) examined the relationship among brain size, innovation, and success in unfamiliar settings. Such comparisons suggest important macroevolutionary patterns but cannot reveal an individual learning mechanism merely from species-level correlations. A further transition therefore placed individual animals in longitudinal or experimentally informed designs, allowing investigators to relate task performance to subsequent survival and breeding.
The modern field also developed a countertradition emphasizing cost, ecological specialization, and imperfect measurement. Mery et al. (2003) experimentally demonstrated that selection for improved learning ability could entail reduced larval competitive capacity. Morand-Ferron et al. (2016) integrated methodological challenges in measuring cognition in the wild. Madden et al. (2025) then systematically mapped published links among cognition, behavior, and fitness and showed how uncertain the general relation remains. The historical direction is toward stronger alignment of cognitive measurement, ecological mechanism, and demographic consequence.
Selection Theory and the Behavioral Consequence Chain
The causal structure of interest can be represented as cognitive trait → information-dependent behavior → ecological outcome → fitness component. Every arrow corresponds to an empirical proposition. A chickadee with more accurate spatial memory may retrieve cached seeds more successfully; those seeds may improve winter energy balance; improved balance may increase survival. Each stage can be investigated through distinct measurements, and an unmeasured stage should remain an explicit hypothesis. In the mountain chickadee work, the decisive measured comparison involved cognitive task performance and later survival, with food caching as the species' ecological justification. Sonnenberg et al. (2019).
A second route may run through social and reproductive behavior. Information about companions or social conditions can alter access to collective resources. An individual's problem-solving or discrimination performance might then covary with group size, dominance, partner choice, or offspring provisioning. The relation between task scores and reproduction may be mediated by social interactions as much as by individual foraging skill. Ashton et al. (2018) offer evidence of cognitive differences in a cooperatively breeding bird and illustrate how social ecology enters any account of fitness.
Confounding creates a separate route: early-life nutrition, physiological condition, developmental stress, age, motivational state, exploratory tendency, and habitat quality may influence both an animal's test score and its subsequent survival. Unless such pathways are measured or manipulated, the observed association supports prediction more directly than causal mediation. Even statistical covariate adjustment may be incomplete because unrecorded common causes remain possible. A rigorous analysis asks which variables are plausible common causes, which are genuine mediators, and which are consequences of the focal trait. Conditioning on a mediator or a collider can itself introduce bias.
Selection may also favor combinations of traits. Better memory is potentially valuable only when an animal explores sufficiently to acquire useful locations and survives the exploration process. A lower threshold for sampling unfamiliar food can yield more opportunities and greater exposure to toxins. Cognitive performance cannot be assigned a universal adaptive value independently of its relation to morphology, energy budget, social partners, and ecological structure. The conceptual framework developed by Sih et al. (2012) connects cognition to behavioral syndromes and provides a mechanism for studying such trait combinations.
Optimality, Learning, and Environmental Predictability
Learning produces value by allowing behavior to track informative regularities. If resource qualities or environmental signals change predictably, an organism can improve its expected return by updating behavior from experience. Where conditions remain constant throughout life, a reliably inherited preference may achieve similar performance at a lower developmental or computational cost. Where change is extremely rapid or cues are weakly correlated with future outcomes, the organism may learn patterns that have already expired. Theoretical treatments therefore identify environmental autocorrelation, cue reliability, life span, and sampling expense as determinants of the selective value of learning. Stephens (1991).
A concrete learning problem requires a payoff structure. Suppose a pollinator encounters flowers whose visual appearance is correlated with nectar reward. Learning improves choice when the reward association persists across enough visits to repay initial errors. The value of updating depends on flowers' turnover, the distribution of competitors, and the cost of investigating an unrewarding alternative. A larger learning rate can be beneficial in a changing field yet harmful when stochastic reward fluctuations lead the individual to abandon a productive option prematurely. Cognitive optimality therefore concerns an organism–environment relationship, not maximal scores on an abstract test.
The distinction between a prepared preference and an acquired response is particularly important for evolutionary experiments. Dunlap et al. (2009) studied components of environmental change in relation to the evolution of learning and unlearned preference. The central insight is that an ecological regime can favor different mixes of innate response and plastic updating. It is insufficient to say that changing environments select for more learning; investigators must specify the type, predictability, and scale of change and how acquired information can improve reproductive payoffs.
Models that maximize expected foraging return also require attention to variation and risk. Some animals may benefit from strategies that decrease the likelihood of catastrophic shortfall rather than maximize average intake. A cognitive rule that appears suboptimal under a laboratory's constant payoffs may prove adaptive when field resources are depleted, competitors vary, or errors carry asymmetric costs. The appropriate analysis compares candidate decision rules using actual ecological distributions while recognizing that organisms face constraints on perception, time, and energy. This principle motivates the scientific link to ecological rationality without collapsing the dossier into a general theory of heuristics.
Cognitive Specialization and Ecological Niches
Ecology supplies patterned problems that different cognitive systems solve. Food-caching birds must retain locations of stored resources, migratory species must organize spatial and temporal information across large scales, and nectar feeders benefit from learning reward-associated sensory signals. The presence of a specialization can be hypothesized through task relevance, field performance, developmental trajectories, and phylogenetic comparison. Its evolutionary origin must be evaluated separately through historical and population-level evidence. A species' exceptional performance on a specialized task offers evidence of capacity, while its selective explanation requires more. Dukas (2004).
Specialization can generate uneven performance profiles. A highly accurate spatial memory system might deliver little advantage on reversal learning, inhibitory control, or an unfamiliar manipulatory task. Such heterogeneity matters because batteries that average across unrelated tasks may obscure the trait under ecological selection. Conversely, a latent factor that predicts performance across several contexts can be useful if measurement invariance and common-cause assumptions are supported. The relevant analytical question is whether cognition has a domain-specific, domain-general, or multidimensional structure in a particular species and ecological setting.
Environmentally demanding habitats may create stronger benefits of an ecological specialization. High-elevation mountain chickadees face winter conditions that make stored food especially important, motivating predictions about spatial learning, memory, and survival. Observed altitude differences are informative because they align a hypothesized selection pressure with a focal phenotype. Nonetheless, habitat comparisons are vulnerable to correlated environmental differences, demographic sorting, and developmental plasticity. Longitudinal within-population survival analyses provide a more direct test of whether individual differences in spatial performance track viability under those conditions. Sonnenberg et al. (2019).
A comprehensive theory also allows multiple solutions to the same niche challenge. Animals may improve food availability by remembering individual cache sites, making more stores, searching preferentially in familiar areas, following others, or altering social tolerance. Behavioral equivalence at the level of resource acquisition does not establish cognitive equivalence at the level of mechanism. Experimental perturbation and repeated observation can distinguish these strategies, and their costs may determine which route is favored under a particular ecology.
Foraging Decisions and Information Acquisition
Foraging combines perception, memory, evaluation, search, and action under constraints of time and energy. Individuals must assess cues to food quality, learn which locations were rewarding, adjust to depletion, and choose whether further sampling will repay its cost. These activities make foraging a natural testing ground for ecological cognition because field outcomes can often be quantified through intake rates, visit choices, patch residence, and travel behavior. Yet each measure answers a limited question. Net energy gain is one ecological outcome; survival or recruitment is an additional demographic outcome that may depend on other factors.
The logic of information acquisition predicts selective attention to cues that have sufficient reliability and consequence. If two flower colors are associated with different nectar quantities, a bee may improve returns by assigning value to those colors and updating choice after errors. If floral rewards change, the ability to reverse a preference may become more important than initial learning speed. These components are measurable, but the experimenter must specify whether the task captures the distribution and timing of rewards in the wild. Raine et al. (2008) were able to connect a standardized associative-learning task to colony-level natural nectar collection.
Foraging also creates opportunities for phenotypic confounding. Animals willing to approach unfamiliar objects may complete learning tasks faster because they sample more frequently, while food-motivated animals may participate more readily and harvest more in nature. Body size and locomotor speed can independently influence collection rate. An adequate design therefore separates errors per opportunity, number of opportunities, persistence, motivation, and movement capacity. The original bumblebee investigators tested several such alternatives, an important improvement over merely correlating an aggregate laboratory score with field output.
Ecological outcomes become most interpretable when measured over comparable intervals. Daily nectar collection, seasonal food storage, and lifetime reproductive success integrate over different timescales. A temporary advantage can dissipate as competitors adjust or as organisms compensate for shortfalls. Multiple field seasons and repeated cognitive assessments can establish whether trait effects are robust across weather, resource abundance, and age. This longitudinal approach links foraging behavior to life-history consequences more directly than isolated laboratory trials.
Bumblebee Learning and Natural Nectar Collection
The study by Raine et al. (2008) is a landmark field-linked investigation because it paired experimental learning measurements in bumblebees with natural foraging by the same colonies. Workers learned an association between floral color and reward under controlled conditions, and investigators obtained colony-level rates of nectar collection in the field. This design provided an explicit bridge from a laboratory learning phenotype to an ecologically meaningful outcome while retaining the limits of observational inference about the final relationship.
Across 12 colonies, learning rates differed and were negatively associated with a parameter indexing learning time, while nectar collection was higher in faster-learning colonies. The original paper reported a colony-level Pearson correlation of approximately −0.59 between the learning-rate parameter and nectar collection, with the negative sign arising because a smaller parameter denoted faster learning. The field consequence was substantial: the slowest-learning colonies collected about 40 percent less nectar than the fastest-learning colonies. The published analysis addressed possible effects of worker size and color preference. These exact sample and effect descriptions derive from the original research report. Raine et al. (2008).
The case establishes the ecological predictive relevance of learning but has boundaries. Colonies, rather than individual bees, were the units for the decisive comparison; colony environment and unmeasured physiological differences could have influenced both learning and collection. Nectar intake is an important fitness-related resource measure, but the study did not demonstrate subsequent colony reproductive output or the lifetime genetic contribution of workers. The authors themselves presented a correlational approach as a major advance while acknowledging the need for causal demonstration. Its value rests precisely on the clear, measurable relation between a cognitive task and field resource acquisition.
The study also illustrates why selection cannot be deduced from an effect size alone. For learning rate to evolve in response to the observed relationship, there must be an appropriate source of transmissible variation and a corresponding effect on reproductive contribution. Under another ecology, floral reward regularities may differ and faster updating may offer less benefit. The bumblebee result should therefore be understood as evidence that learning variation can carry ecological value in one natural foraging system, together with a methodological model for measuring the intervening behavior.
Food Caching, Spatial Memory, and Winter Survival
The mountain chickadee studies address a crucial demand in evolutionary cognition: linking a cognitive phenotype to a direct component of fitness. Food-caching birds store supplies that may become important during resource shortages, so remembering cache locations has an intelligible ecological function. This creates a well-specified behavioral hypothesis: individuals that perform better on suitable spatial learning tasks may be more likely to recover resources and survive challenging seasons. The advantage of this system is the opportunity to follow identifiable wild birds from one winter to another. Sonnenberg et al. (2019).
Sonnenberg et al. (2019) tested spatial learning and memory and a reversal task in wild mountain chickadees living at high elevation. Their comparisons separated adults and first-year juveniles and used follow-up observations of birds surviving into their second winter. Adults performed better than juveniles on the cognitive tasks. Critically, juveniles that survived into the next winter had shown better spatial learning performance than juveniles that did not survive. Among individuals retested across winters, spatial and reversal performance did not show significant year-to-year differences, helping address whether observed differences simply reflected improvement with age.
The evidence is substantially closer to selection than a task-only species comparison because survival was measured in the same ecological population. It nevertheless remains a study of phenotypic selection associations rather than a direct experiment that selectively changes an animal's memory and measures its descendants. Differences in developmental history, condition, territory quality, or other correlated characteristics remain plausible components of the causal process. Direct measures of cache retrieval, energetic status, and survival across multiple environmental regimes would further clarify mediation.
The ecological interpretation is powerful because the predicted consequence derives from a known functional challenge. A spatial memory score cannot be treated as a universal fitness predictor, but in a food-caching bird during a resource-limited winter it has a clear behavioral rationale. The study establishes that survival-related cognitive variation can be observed in wild populations and places viable selection on spatial cognition within reach of longitudinal field methodology.
Spatial Cognition and Reproductive Investment
Survival is one fitness component; reproduction introduces additional pathways by which cognitive variation might matter. Spatially proficient birds may enter the breeding season in better condition after recovering stored food more effectively. Mates may adjust reproductive investment in relation to a partner's condition or provisioning capacity. An observed positive association between a male cognitive score and offspring production, however, could reflect female plastic investment, a shared environment, or other correlated traits rather than the inheritance of an adaptive cognitive allele. Each of these mechanisms generates different predictions. Branch et al. (2019).
In wild mountain chickadees, Branch et al. (2019) examined female reproductive investment in relation to the spatial cognitive performance of their male partners. Under favorable breeding conditions, females paired with males that performed better on spatial cognition tasks laid larger clutches and fledged larger broods than females paired with lower-scoring males. These are field reproductive endpoints, making the study more informative than a mating preference assay alone. The relationship was environmentally contingent, which is central to interpreting the selective value of the focal trait.
The potential mechanisms include direct material benefits through partner provisioning, female assessment of mate quality, and indirect benefits from descendants inheriting advantageous traits. The existence of a fitness association does not uniquely identify which mechanism operates. A decisive test of genetic benefits would require information on parentage, heritability, offspring performance, and environmental mediation. Likewise, claims that females consciously recognize spatial memory as such require independent behavioral evidence. The research is most securely read as a link between a partner-associated cognitive phenotype and realized reproductive investment.
Combined with survival data from the same biological system, the study offers an emerging picture of cognition embedded in different stages of the annual cycle. Winter selection can favor resource-relevant memory; breeding outcomes can depend on mate associations and ecological conditions. A complete fitness account would integrate the survival and reproduction of individuals and descendants over multiple years rather than transferring the sign of one association across all components.
Problem Solving and Fitness Trade-offs in Great Tits
Experimental assessment of novel problem solving has been used to investigate whether differences in behavioral flexibility influence reproduction. The study by Cole et al. (2012) temporarily brought wild great tits into captivity for a problem-solving task and then examined subsequent breeding performance in the wild. This approach paired an individualized cognitive phenotype with field life-history outcomes while limiting the period of experimental confinement.
The researchers studied 468 great tits and reported that females classified as problem solvers produced larger clutches than nonsolvers. They examined alternatives related to breeding timing, habitat, offspring quality, and adult survival, and interpreted aspects of foraging behavior as consistent with more efficient exploitation of the environment. Yet the beneficial association with clutch size had a striking counterpart: problem-solving birds were more likely to desert nests. The joint pattern produced little or no net directional selection on the measured problem-solving trait within the observed reproductive outcomes. Cole et al. (2012).
This case shows why selection should be estimated on integrated outcomes rather than on a favorable component selected after the fact. If a cognitive phenotype is associated with larger initial reproductive investment and higher probability of losing that investment, total reproductive success may remain unchanged. It also illustrates a problem of task interpretation: solving a novel apparatus could depend on exploration and persistence as well as inference or learning. These traits may themselves contribute to nest abandonment or responsiveness to disturbances.
The scientific result is a concrete life-history trade-off rather than a simple positive or negative association of intelligence with fitness. It suggests that the evolutionary persistence of variation in cognitive performance may reflect opposing selection pressures distributed across behavioral contexts. Longitudinal studies measuring individuals repeatedly across breeding seasons could determine whether these pressures change with experience and ecological conditions.
Sexual Selection and Bowerbird Problem Solving
Cognition can have reproductive consequences when individuals use learned or flexible behavior during courtship, territory management, or construction of sexual displays. Satin bowerbirds offer a model because males build and maintain elaborate structures used during courtship. Display-related behavior includes responses to changing objects and interference, creating opportunities to study problem solving in a reproductively meaningful context. Keagy et al. (2009).
Keagy et al. (2009) reported that measures of male problem-solving ability predicted mating success. The association is consistent with cognition influencing a male's capacity to maintain an attractive bower or respond to ecological and social challenges. However, the task-to-mating pathway requires further behavioral identification. Persistent or exploratory males may perform well in experimental challenges and also devote more effort to displays, yielding a correlation without a uniquely cognitive mechanism. Mating success additionally captures only a portion of lifetime reproductive success.
Sexual selection may favor cognitive traits indirectly through the observable products of skill, such as display construction, rather than through direct assessment of internal cognitive capacity by potential mates. A female may respond to reliably maintained display properties whose production happens to depend on memory, motor coordination, attention, and learned routines. This interpretation connects cognitive performance with the ecological production of sexual signals while avoiding assumptions about what individuals perceive in others' minds. The literature review by Boogert et al. (2011) describes broader questions in mate choice for cognitive traits.
The bowerbird case also invites cross-seasonal measurement. Mating success may vary strongly with ecological conditions and display opportunities, and the successful male's contribution to descendants depends on additional factors. Assessing both sexual and viability selection can determine whether courtship-related cognitive advantages are accompanied by resource or survival costs. This is a general principle: every claimed advantage must be placed within a lifetime balance of selective consequences.
Constitutive and Operating Costs of Learning
A comprehensive evolutionary theory requires an account of the costs of acquiring and maintaining cognitive capacity. A constitutive cost is incurred through the presence or maintenance of machinery that makes a cognitive trait possible, independently of whether the trait is used on a particular occasion. An operating cost arises when learning or information processing is deployed and may include energy, time, errors, exposure, or interference with other activities. Distinguishing these categories matters because the same learning capacity can have different fitness effects across environments with differing informational demands. Mery et al. (2003).
The experimental evolution study by Mery et al. (2003) selected replicated populations of fruit flies for improved learning. The evolved improvement was associated with reduced larval competitive ability compared with control populations. The investigators tested inbreeding as an alternative explanation through crossing procedures. The result provides evidence of a constitutive trade-off: improved learning capacity carried a fitness-related disadvantage even when the capacity itself was not being expressed in the competitive assay. This is an unusually strong example of causal evolutionary manipulation because lines underwent experimentally imposed selection.
The authors subsequently investigated costs arising from the use of learning in Mery et al. (2004). Such work shows why an adaptive calculation cannot rely only on reward improvement after learning. If performing an informational task reduces immediate survival or reproduction, the net benefit depends on whether future gains compensate for that loss. The energetic and physiological mechanisms underlying a measured cost may be difficult to isolate, particularly when selection changes multiple correlated traits across generations.
Trade-offs can also maintain variation in a population. Individuals differ in ecological exposure, developmental condition, and strategy; a capacity favored under unpredictable rewards may be costly under stable conditions. Yet identifying a laboratory trade-off is distinct from proving that it maintains polymorphism in a wild population. That stronger claim requires the distribution of environmental states, genetic and developmental transmission, fitness functions for alternative phenotypes, and sufficient temporal observation. The fruit fly studies supply an experimental foundation for such models, not a universal magnitude of cognitive cost.
Flexibility, Innovation, and Ecological Change
Novel environments create opportunities and hazards for cognitive flexibility. Animals encountering unfamiliar foods, predators, competitors, or human-modified habitats may benefit from exploration, innovation, and learning. Behavioral innovation can increase access to resources, but indiscriminate novelty seeking can also produce exposure to toxins or predation. The adaptive value of innovation therefore depends on how novelty is encountered, how errors are corrected, and whether innovations persist long enough to alter demographic outcomes. Tebbich et al. (2016) examine the relation between mechanisms and functions of animal innovation.
Across bird species, Sol et al. (2005) investigated whether relative brain size, innovation, and the ability to become established in novel environments covaried. Such comparative findings made innovation a plausible component of ecological resilience. Their limitations are characteristic of interspecific inference: brain size correlates with numerous ecological and life-history traits, and establishment success is jointly shaped by introduction history, propagule pressure, habitat suitability, and opportunity. An observed species-level pattern therefore motivates mechanistic tests rather than settling them.
Innovation should be measured as a behavioral occurrence, whereas cognitive flexibility refers to the processes that allow modification or transfer of behavior. A novel solution can result from trial-and-error exploration, social observation, changed motivation, or the removal of familiar alternatives. Repeated innovation may also reflect repeated exposure to unsolved problems. Researchers need standardized opportunity measures and task conditions appropriate to each species before comparing rates across populations. The integrated framework of Tebbich et al. (2016) is valuable because it embeds innovation in both its causal mechanisms and its ecological consequences.
For conservation, the distinction has practical importance. Behavioral flexibility might help some species respond to urbanization or climate-driven shifts, while specialized behavior remains effective under familiar stable conditions. A conservation inference must therefore trace a measured cognitive or behavioral trait to actual persistence or reproductive output in the affected population. Apparent success in one kind of anthropogenic novelty need not generalize to all forms of environmental change.
Personality, Exploration, and Behavioral Syndromes
Animal personality research documents repeatable individual differences in traits such as exploration, boldness, activity, and responsiveness to social situations. These characteristics can interact with cognitive testing because a shy individual may withhold responses despite competent discrimination, while a bold individual may sample more stimuli and accumulate greater opportunity for learning. An apparent variation in cognitive ability may partly reflect variation in participation, attention allocation, or response threshold. Sih et al. (2012) present a behavioral ecology framework for linking personality and cognition.
These interactions are scientifically meaningful rather than merely statistical nuisances. A stable exploratory strategy may improve opportunities to collect information but raise predation risk. An individual that updates beliefs rapidly may become flexible in an uncertain world while exhibiting less behavioral consistency. The fitness consequences of a cognitive trait may therefore depend on a correlated behavioral syndrome, with selection acting on multivariate phenotypes. Models measuring only one task score can misattribute selection that actually concerns a trait combination.
Experiments can separate exploration from information processing by measuring approach latency, number of exposures, within-exposure accuracy, and learning curves. Repeated trials enable estimation of individual-specific acquisition rather than simply declaring an individual a solver after a single success. Counterbalanced stimuli and independent measures of motor ability, hunger, or stress help establish which differences reflect cognitive processes. Because such controls can themselves change the animal's motivational ecology, field validation remains essential.
When multiple traits jointly predict survival, ecological interpretation should assess their covariance and effect modification. A fast learner may show a reproductive advantage only among individuals that explore frequently enough to encounter useful novel cues. A cautious, accurate individual may survive better under high predation while achieving lower intake. Such interactions support a richer explanation of maintenance of variation than an assertion that selection always maximizes a single cognitive dimension.
Social Information and Collective Ecological Outcomes
Many organisms obtain information socially. Observing conspecific foraging choices, alarm responses, or breeding activity may reduce the cost of independent exploration. Individuals can also benefit from shared vigilance, cooperative resource access, or repeated exchange. Social information therefore complicates a purely individual account of cognition: the payoff associated with one animal's ability may depend on the capacities and behaviors of others. The social environment is an ecological source of information and also an arena of competition.
Group-level processes create different units of analysis. A well-informed individual may guide companions toward food, and companions may benefit without possessing the same ability. Conversely, group competition can diminish the advantage of knowledge when others follow and deplete discovered resources. These dynamics can cause the direct fitness effect of cognition to differ from its consequences for social partners. The within-species magpie study by Ashton et al. (2018) connects social grouping with cognition and reproduction, although isolating the specific informational mechanism requires further experimentation.
Fitness effects are also frequency dependent. A rare innovator can gain temporarily by discovering a resource that competitors have not exploited; once many animals acquire the same information, the advantage may diminish. Social learning can spread a useful behavior while reducing the incentive to acquire information independently. An evolutionary interpretation therefore needs to consider the distribution of alternative strategies and the temporal dynamics of information transmission within populations.
Comparing individuals of different social status requires appropriate adjustment for age, opportunity, kinship, and habitat. Mixed models may represent group identity and relatedness, while experiments altering access to information can directly test social transmission. The strongest accounts link observed information flow to a specific action and a subsequently measured fitness component, rather than inferring adaptive social intelligence from group size alone.
Development, Experience, and Life-History Timing
Cognitive phenotypes develop through interactions among inherited architecture, age, nutrition, learning history, and ecological opportunity. An animal with more extensive prior experience in a task domain may score higher without possessing a greater underlying learning rate or general capacity. Age can influence sensory ability, motivation, and physiological performance as well as memory. Comparing individuals at a single time point therefore conflates developmental state and stable between-individual variation unless the study includes relevant controls or repeated measurements.
Life-history timing changes the value of cognition. Juveniles may benefit from rapid learning when establishing foraging routines but face elevated mortality while sampling. Adults may possess refined ecological knowledge while remaining constrained by parental duties or territorial commitments. Cognitive improvements can matter most at particular transitions: dispersal, the onset of a harsh season, first breeding, or entry into a new social group. Field designs should therefore align measurement schedules with these biologically important periods.
The mountain chickadee evidence is valuable because it jointly considered age-class differences and longitudinal change. Sonnenberg et al. (2019) observed better average task performance in adults yet found no significant change for the retested survivors across winters. The combination of age structure and repeated measures constrained a simple explanation that all survival-linked cognitive differences arise from within-individual age-related improvement. It does not settle every developmental mechanism, but it shows why survival selection can be investigated more convincingly with repeated measurements than with a single cross-sectional comparison.
Developmental environments can produce apparent selection if early advantages influence both cognition and survival. Nest quality or juvenile condition, for example, might generate better task performance and enhanced survival through partially independent pathways. Long-term pedigrees, cross-fostering, and experimentally controlled learning opportunities can help distinguish inherited variation from environmental covariance. These methods are most informative when repeated across ecologically different years.
Measuring Cognitive Phenotypes in Natural Populations
A cognitive measurement gains interpretive force through construct validity: the test must assess the proposed information-processing property rather than an uncontrolled collection of motivation, activity, sensory ability, and task familiarity. Ecological validity adds a separate question about whether the property is relevant to the natural setting. A task can be highly repeatable yet weakly related to an organism's actual ecological problems. Conversely, an ecologically realistic task can have poor reliability when the investigator cannot standardize exposure or observation. Morand-Ferron et al. (2016).
Repeatability estimates the consistency of between-individual differences under repeated measurement and supplies an upper constraint on the reproducibility of an observed phenotype. Repeated sessions can separate initial exploration, rate of acquisition, asymptotic accuracy, retention, and reversal learning. A single binary success score loses information and can confound speed with willingness to persist. Longitudinal survival analyses are especially sensitive to measurement error: noisier cognitive scores attenuate associations and may obscure selection.
Test design should preserve species-appropriate sensory and motor opportunities. An apparatus requiring manipulation may systematically disadvantage individuals whose natural cognitive specialization concerns acoustic, spatial, or visual navigation. Different species may reach identical correct answers through different perceptual rules, so task equivalence does not necessarily imply mechanism equivalence. Within-species ecological studies reduce some cross-species comparability problems but still need controls for age, sex, energetic state, and social position.
Field technologies can improve ecological correspondence by recording individually identified visits, choices, locations, and survival across seasons. Automated feeders, radio-frequency identification, remote tracking, and repeated standardized trials expand the range of observable behaviors without requiring continuous capture. Nevertheless, machine-recorded performance can change when animals differ in willingness to approach devices, and detection probability is rarely identical across all individuals. Transparent observation models and robust sensitivity analyses are necessary to separate cognition from the research infrastructure itself.
From Correlation to Causal Explanation
Correlations between cognition and ecological outcomes can arise through a direct effect, a mediated behavioral pathway, a shared cause, reverse causality, or selection into the observed sample. Consider a field study in which high-scoring animals survive more frequently. Better memory may improve resource recovery, or healthier animals may both perform better and survive longer. A third pathway can occur when only individuals surviving to testing are eligible for measurement, leaving an already selected sample. The causal diagram should therefore precede statistical adjustment.
Controlled manipulation can strengthen causal inference if it changes a cognitive opportunity without simultaneously altering multiple ecological conditions. Randomized access to information, changed reward contingencies, or experimentally modified spatial demands may isolate a portion of the pathway. In some designs the intervention changes learning experience, not inherited capacity, and thus supports an effect of acquired information rather than genetic selection on cognition. Experimental evolution, as used by Mery et al. (2003), is informative about the response to selection, but its laboratory environment restricts transferability to wild ecologies.
Observational studies can improve identification through repeated individual measures, demographic follow-up, measurements of plausible confounders, and preregistered hypotheses. Multiyear designs allow investigators to ask whether cognitive performance measured before an ecological event predicts a subsequent outcome. Such temporal ordering strengthens prediction but does not eliminate all common-cause explanations. Where an investigator controls for intermediate behavior, the estimated direct effect of cognition may differ from its total mediated effect; the estimand must be specified.
Causal evidence is often assembled across studies rather than delivered by a single test. A laboratory experiment may establish that an animal can learn relevant cues; field observations may show the corresponding behavior; a longitudinal study may associate the cognitive phenotype with survival; and a breeding or selection experiment may assess inherited variation. Together they provide converging support while retaining uncertainty about any unmeasured pathway. The requirement is coherence among the levels of explanation, not an expectation that every design performs all functions.
Quantifying Selection on Cognitive Traits
Phenotypic selection is commonly examined by relating relative fitness to standardized trait values. A directional selection coefficient describes how fitness changes with the trait after accounting for relevant covariates; nonlinear terms can reveal stabilizing or disruptive patterns; interactions can test whether effects depend on sex, age, or ecological regime. When several cognitive and behavioral traits are correlated, multiple regression or multivariate selection models can help distinguish direct associations from selection on a correlated phenotype. Their interpretation remains conditional on measurement quality and the causal structure of included covariates.
The response variable matters fundamentally. Survival to the next breeding season can be modeled through event-history or multistate methods when observation and detection probabilities vary. Reproductive outcomes may require count or zero-inflated models when many individuals fail to reproduce at all. Lifetime reproductive success demands sufficient longitudinal follow-up. Effect sizes from different endpoints cannot be pooled into a single conceptual category of fitness without attention to their scale and biological meaning.
Selection gradients can vary among environments, creating temporal or spatial fluctuations in the value of a cognitive trait. A bird's memory may be vital during a cold winter but contribute little under abundant food. A problem-solving phenotype may increase clutch size only to face a compensating cost in nest retention. Evaluating the fitness effect therefore requires estimates across ecological states, not only a point estimate averaged over them. The studies of Cole et al. (2012) and Sonnenberg et al. (2019) exemplify different endpoint choices.
A claim that selection will produce evolutionary change further requires heritable variation and consideration of genetic correlations, developmental plasticity, and population dynamics. A survival advantage associated with a phenotype does not establish narrow-sense heritability. Pedigrees, genomic relatedness, or controlled breeding designs can estimate transmissible variance, while repeated ecological data can characterize the selective environment. Only their integration supports a prediction about evolutionary response.
Comparative and Phylogenetic Inference
Comparative evidence can reveal broad ecological correlations among species with different cognitive architectures and life histories. Relative brain size, observed innovation, social structure, and colonization success have been used to investigate associations between cognition and ecological flexibility. Such patterns raise hypotheses about selection pressures operating over evolutionary time. Yet shared ancestry makes species observations statistically nonindependent, while ecological covariates can explain part of an apparent cognitive association. Sol et al. (2005) provide a historically important comparative example.
Phylogenetic control is therefore necessary but insufficient. Accounting for a common ancestor can correct part of the dependence among species without making brain volume a validated measure of a specific cognitive process or establishing causality. Two species may display similar outcomes by distinct neural and behavioral routes, and two related species may diverge ecologically while retaining homologous mechanisms. Investigators must specify whether the question concerns historical origin, selection on current variation, or functional equivalence.
Within-species studies provide a complementary unit of analysis because they can relate multiple individuals' measured phenotypes to the same local demographic processes. Such studies improve directness but may have limited external validity when conducted during one season, in one habitat, or with a small number of social groups. The strongest research program combines species-appropriate experiments, population-level field measurements, and phylogenetically informed comparisons. Evidence at one level is most useful when it generates testable predictions for the others.
The distinction also limits easy narratives about brains getting bigger because intelligence is better. Brain size can covary with development, body size, sensory demands, longevity, and energetic budgets. The consequences of a neural investment depend on what information it processes and how that information alters decisions under local ecological constraints. Ecological cognition requires behavioral mediation and measurable fitness outcomes before assigning an adaptive function to a particular anatomical difference.
The Evidence Landscape in 2025
The systematic review by Madden et al. (2025) supplies an unusually useful quantitative overview of research linking individual cognitive performance to fitness in nonhuman animals. It assembled 45 studies involving 26 species and described 211 cognition–fitness relationships. More than 70 percent of the raw reported relationships were statistically nonsignificant, and statistically significant links became less likely after accounting for covariates. The authors judged the available evidence incomplete and inconclusive as a foundation for broad generalizations about cognition and fitness.
The review also found informative heterogeneity. Significant associations tended to be positive, although not universally so; relationships appeared more likely when researchers considered general rather than narrowly specific cognitive entities and when the fitness endpoint concerned survival rather than reproductive output. These patterns identify promising hypotheses, but the underlying studies varied in tasks, taxa, and endpoint definitions. The higher frequency of significant survival associations does not itself prove that survival selection is stronger in all species, because study designs and measurement error also differ by outcome. Madden et al. (2025).
The systematic review provides an empirical corrective to a publication-driven emphasis on conspicuous positive case studies. An association that fails to reach conventional significance is not proof of zero effect; it may arise from low power, inaccurate measurement, unfavorable ecological timing, or real heterogeneity. Conversely, a statistically significant correlation within one population does not imply a stable evolutionary advantage across environments. The distribution of results encourages a field-wide focus on estimated effect sizes, uncertainty intervals, preregistration, longitudinal replication, and better causal models.
Other contemporary contributions frame these problems through the diversity of cognitive capacities and the need for individual-based evolutionary analysis. Sol et al. (2025) treat the evolution of cognition as a puzzle requiring stronger ecological linkage. Sheehan et al. (2025) emphasize connections between individual fitness and cognitive evolution. Leadbeater et al. (2025) situate the evidence in a broader inquiry into how selection shapes diverse animal minds. These contributions support a research direction rather than a new single-factor consensus.
Critical Scientific Discussion
The strongest evidence for an evolutionary ecology of cognition comes from convergence among plausible ecological function, ecologically appropriate measurement, temporally ordered fitness outcomes, and manipulations where feasible. The bumblebee study links learning speed to natural resource collection. The mountain chickadee study links spatial learning to subsequent winter survival. The great tit study reveals compensating reproductive effects. The magpie and bowerbird studies connect cognitive performance to social or mating outcomes. These results are complementary because they demonstrate different segments of the mechanism-to-fitness chain; none supplies a universal selection coefficient for cognition as a whole.
The concept of fitness has frequently been stretched across resource proxies, clutch counts, mating outcomes, and survival probabilities. Clear reporting should identify the measured endpoint in every claim and reserve descriptions of direct fitness for demographic or genetic contribution where these are actually measured. A correlate of reproduction can be informative without representing lifetime reproduction, and a resource-gain measure can be evolutionarily relevant without directly measuring selection. The original authors' own formulations should be respected while their results are located on an explicit evidentiary continuum. Raine et al. (2008); Cole et al. (2012).
Measurement non-equivalence presents a second challenge. Comparing fast versus slow learners presupposes that the task separates information acquisition from performance speed, sensory thresholds, motivation, exploration, and prior experience. General cognitive factors can summarize correlations among tasks, but their biological meaning depends on the construction and replicability of the battery. Studies should report task reliability, exposure histories, participation rates, missing observations, and model diagnostics. The field's conceptual review by Morand-Ferron et al. (2016) remains relevant because technical improvements do not replace careful construct definition.
Environmental contingency is a third challenge that also offers the deepest theoretical insight. A trait can be beneficial under predictable resource signals and costly when sampling opportunities are dangerous. The same phenotype can improve one fitness component and diminish another. Stable individual differences may be maintained through fluctuating selection, frequency dependence, or trade-offs, but demonstrating any such mechanism requires evidence for its actual operation. Stephens (1991) and Mery et al. (2003) provide complementary theoretical and experimental foundations.
The broadest contemporary limitation is statistical and inferential generalization. The Madden et al. (2025) reveals that many observed associations are uncertain and that the evidentiary landscape is uneven across species and endpoints. Publication bias may favor striking effects, and multiple correlated cognitive tests create risks of selective reporting. Small samples can yield unstable estimated gradients, while uncorrected group or population dependence can overstate precision. A mature synthesis must therefore present its strongest field results alongside the prevalence of null or ambiguous findings and describe what additional research would change a conclusion.
Research Priorities and Cross-System Perspectives
A priority for field science is the prospective study of cognitive phenotypes measured before major environmental challenges. Longitudinal identification of juveniles, repeated cognitive testing, direct tracking of food acquisition, and survival and reproduction measurements can identify the relevant ecological intervals. Data should span variable years to estimate whether cognitive performance remains predictive when the resource regime changes. Whenever possible, investigators should document both the hypothesized mediator and its fitness consequence rather than skipping from task score to demographic outcome.
A second priority is controlled manipulation along the causal chain. Experiments that modify available information, reward structure, or learning opportunity can test whether performance changes lead to ecological gains. Such manipulation should be distinguished from experiments selecting heritable learning traits, because the intervention changes a different causal component. Replicated experimental evolution, cross-fostering, and pedigrees can connect plastic behavior, transmissible variation, and selection response. Pre-registered selection models and openly available task protocols would make cross-population synthesis more credible.
A third priority is measurement pluralism with conceptual precision. Standardized tasks improve within-study comparisons, while species-appropriate tasks preserve ecological meaning. Researchers should report separate measures of acquisition, accuracy, retention, reversal, participation, and motivation before compressing them into an aggregate factor. Formal latent-variable and measurement-error models can determine whether a general performance construct is supported. Selection analyses should specify endpoints and treat survival, reproductive success, and ecological resource yields as distinguishable consequences.
The Artificial Era broadens the comparative vocabulary around information processing, but an evolutionary fitness claim remains anchored in a population undergoing reproduction, survival differences, and transmission. Artificial cognitive systems may be compared with animals in terms of learning, uncertainty management, memory, or resource-limited decision behavior. Their optimization objectives, selection environments, persistence mechanisms, and criteria for success require independent definitions. Analogies between performance and adaptation can guide modeling; they do not transfer biological reproductive fitness unchanged to artificial systems. The comparative opportunity is strongest when each system's informational task and evaluative endpoint are operationally specified.
A final research direction is the coordination of conservation ecology, behavioral science, and demographic modeling. If a species depends on particular cognitive competencies to track rapidly shifting resources, changes in habitat structure can alter the value of those competencies within a few generations. Predictive models will need measured ecological payoffs and trait variation rather than an assumption that more flexible animals always prevail. That agenda connects cognitive ecology to practical questions of population persistence while preserving the distinction between a plausible adaptation narrative and demonstrated selection.
Conclusion
The fitness consequences of cognition arise through specific ecological pathways in which information-dependent behavior changes resource acquisition, survival, mating, or reproduction. Cognition contributes to natural selection when measurable phenotypic differences reliably affect demographic success in a relevant environment and when those differences have a basis on which evolutionary processes can operate. The value of learning, memory, innovation, and flexibility depends on cue reliability, ecological change, biological cost, life-history stage, social conditions, and the alternatives available to the organism. These dependencies define cognition as an ecological and evolutionary research object.
Original investigations have established consequential links at several levels. Bumblebee learning predicts natural nectar collection; food-caching chickadee spatial performance predicts overwinter survival; cognitive differences in wild birds are associated with reproductive outcomes and social ecology; experimental evolution shows that improved learning can entail compensating costs. The great tit finding, in which higher clutch size was offset by nest desertion, exemplifies how opposing life-history effects can weaken overall selection. The evidence supports particular mechanisms in particular systems and encourages direct studies of the transitions between cognitive phenotype, behavior, and demographic consequence.
The field-wide assessment in Madden et al. (2025) places a decisive constraint on general claims: most published cognition–fitness relationships in its reviewed corpus were statistically nonsignificant. That finding defines the contemporary scientific frontier. Future progress depends on validated phenotypes, causal identification, longitudinal fitness measurement, heritability and selection analyses, environmental replication, and explicit accounting for contradictory outcomes. Behavioral ecology thereby supplies cognition with a rigorous functional question: under what conditions, through which actions, and with what net consequences does information processing alter the evolutionary future of organisms?
References
- Ashton, B. J., Ridley, A. R., Edwards, E. K., & Thornton, A. (2018). Cognitive performance is linked to group size and affects fitness in Australian magpies. Nature, 554(7692), 364–367. https://doi.org/10.1038/nature25503
- Boogert, N. J., Fawcett, T. W., & Lefebvre, L. (2011). Mate choice for cognitive traits: A review of the evidence in nonhuman vertebrates. Behavioral Ecology, 22(3), 447–459. https://doi.org/10.1093/beheco/arq173
- Branch, C. L., Pitera, A. M., Kozlovsky, D. Y., Bridge, E. S., & Pravosudov, V. V. (2019). Smart is the new sexy: Female mountain chickadees increase reproductive investment when mated to males with better spatial cognition. Ecology Letters, 22(6), 897–903. https://doi.org/10.1111/ele.13249
- Cole, E. F., Morand-Ferron, J., Hinks, A. E., & Quinn, J. L. (2012). Cognitive ability influences reproductive life history variation in the wild. Current Biology, 22(19), 1808–1812. https://doi.org/10.1016/j.cub.2012.07.051
- Dukas, R. (2004). Evolutionary biology of animal cognition. Annual Review of Ecology, Evolution, and Systematics, 35, 347–374. https://doi.org/10.1146/annurev.ecolsys.35.112202.130152
- Dunlap, A. S., & Stephens, D. W. (2009). Components of change in the evolution of learning and unlearned preference. Proceedings of the Royal Society B: Biological Sciences, 276(1670), 3201–3208. https://doi.org/10.1098/rspb.2009.0602
- Keagy, J., Savard, J.-F., & Borgia, G. (2009). Male satin bowerbird problem-solving ability predicts mating success. Animal Behaviour, 78(4), 809–817. https://doi.org/10.1016/j.anbehav.2009.07.011
- Leadbeater, E., & Thornton, A. (2025). Selection shapes diverse animal minds. Philosophical Transactions of the Royal Society B: Biological Sciences, 380(1929), 20240108. https://doi.org/10.1098/rstb.2024.0108
- Madden, J. R., Ellis, S., Lens, L., & Verbruggen, F. (2025). How does cognition determine an individual's fitness? A systematic review of the links between cognition, behaviour and fitness in non-human animals. Philosophical Transactions of the Royal Society B: Biological Sciences, 380(1929), 20240118. https://doi.org/10.1098/rstb.2024.0118
- Mery, F., & Kawecki, T. J. (2003). A fitness cost of learning ability in Drosophila melanogaster. Proceedings of the Royal Society B: Biological Sciences, 270(1532), 2465–2469. https://doi.org/10.1098/rspb.2003.2548
- Mery, F., & Kawecki, T. J. (2004). An operating cost of learning in Drosophila melanogaster. Animal Behaviour, 68(3), 589–598. https://doi.org/10.1016/j.anbehav.2003.12.005
- Morand-Ferron, J., Cole, E. F., & Quinn, J. L. (2016). Studying the evolutionary ecology of cognition in the wild: A review of practical and conceptual challenges. Biological Reviews, 91(2), 367–389. https://doi.org/10.1111/brv.12174
- Raine, N. E., & Chittka, L. (2008). The correlation of learning speed and natural foraging success in bumble-bees. Proceedings of the Royal Society B: Biological Sciences, 275(1636), 803–808. https://doi.org/10.1098/rspb.2007.1652
- Sheehan, M. J., & Miller, S. E. (2025). Linking individual fitness to the evolution of cognition. Philosophical Transactions of the Royal Society B: Biological Sciences, 380(1929), 20240122. https://doi.org/10.1098/rstb.2024.0122
- Sih, A., & Del Giudice, M. (2012). Linking behavioural syndromes and cognition: A behavioural ecology perspective. Philosophical Transactions of the Royal Society B: Biological Sciences, 367(1603), 2762–2772. https://doi.org/10.1098/rstb.2012.0216
- Sol, D., Bateman-Neubert, A., Noguer, L., & Taylor, A. H. (2025). The evolutionary puzzle of cognition: Challenges and insights from individual-based studies. Philosophical Transactions of the Royal Society B: Biological Sciences, 380(1929), 20240123. https://doi.org/10.1098/rstb.2024.0123
- Sol, D., Duncan, R. P., Blackburn, T. M., Cassey, P., & Lefebvre, L. (2005). Big brains, enhanced cognition, and response of birds to novel environments. Proceedings of the National Academy of Sciences, 102(15), 5460–5465. https://doi.org/10.1073/pnas.0408145102
- Sonnenberg, B. R., Branch, C. L., Pitera, A. M., Bridge, E. S., & Pravosudov, V. V. (2019). Natural selection and spatial cognition in wild food-caching mountain chickadees. Current Biology, 29(4), 670–676.e3. https://doi.org/10.1016/j.cub.2019.01.006
- Stephens, D. W. (1991). Change, regularity, and value in the evolution of animal learning. Behavioral Ecology, 2(1), 77–89. https://doi.org/10.1093/beheco/2.1.77
- Tebbich, S., Griffin, A. S., Peschl, M. F., & Sterelny, K. (2016). From mechanisms to function: An integrated framework of animal innovation. Philosophical Transactions of the Royal Society B: Biological Sciences, 371(1690), 20150195. https://doi.org/10.1098/rstb.2015.0195
- Tinbergen, N. (1963). On aims and methods of ethology. Zeitschrift für Tierpsychologie, 20(4), 410–433. https://doi.org/10.1111/j.1439-0310.1963.tb01161.x
Social Ecology and Cognitive Performance in Magpies
The relationship between social complexity and cognitive evolution is often tested through species-level comparisons, but social organization can also vary within a species. Australian magpies provide an opportunity to investigate whether individuals living in differently sized groups exhibit measurable differences in cognitive task performance and whether those differences are associated with reproductive outcomes. A multivariate battery can address whether several task scores share common variance, while longitudinal demographic observations connect the cognitive measures to ecological significance. Ashton et al. (2018).
Ashton et al. (2018) reported that individuals from larger social groups tended to show better performance on several cognitive tasks, and that cognitive performance was associated with female reproductive success. The design offered a meaningful within-species complement to broad comparative correlations between sociality and brain measures. It also raised a causal question: does a larger group enrich learning opportunities and thereby improve cognition, or do certain cognitive phenotypes enable individuals to enter and persist in particular social groups?
These alternatives are not exhausted by a two-way choice. Habitat quality, parental environment, age, dominance, and repeated exposure to tasks might affect both social group structure and measured cognitive performance. Group-level data also require analytic attention to nonindependence: individuals sharing a group share contexts, companions, and potentially relatedness. If measured scores are correlated within groups, treating all birds as independent replicates would exaggerate precision. The study's importance lies in opening a tractable, field-based connection among social ecology, cognitive phenotype, and reproduction.
The magpie evidence is especially useful when interpreted alongside food-caching and problem-solving studies. It shows that the ecological route from cognition to fitness may operate through relationships and cooperative organization as well as private information about food. Selection on cognitive performance can be contingent on social context and on what a particular task actually measures.