The study, published in Natural Resources Research, was led by Changkui Lei and Qi Qiao of Taiyuan University of Technology, together with Jun Deng and Jingyu Zhao of Xi’an University of Science and Technology, Chuanbo Cui, and Weigang Wang of the Gansu Bureau of the National Mine Safety Administration. Their central innovation is a hybrid model that pairs support vector regression, a powerful but notoriously parameter-sensitive prediction method, with the grey wolf optimization algorithm, a metaheuristic that mimics the leadership hierarchy and hunting strategy of wolf packs. The result, dubbed GWO-SVR, achieved a coefficient of determination of 0.9902 on test samples, meaning the model explained more than 99 percent of the variance in coal temperature during oxidation experiments.

To understand why this matters, it helps to grasp the physics of spontaneous combustion. When coal is exposed to oxygen at ambient temperatures, it oxidizes slowly, releasing heat. In loose coal piles, goaf areas behind longwall mining faces, and stockpiles, that heat can accumulate faster than it dissipates. As temperature rises, oxidation accelerates exponentially, creating a runaway feedback loop. The researchers’ large-scale coal oxidation experiment, using samples from the Paner Coal Mine, traced this process in detail. They observed that the high-temperature point inside their experimental furnace did not stay put: it migrated dynamically from the middle-upper section toward the lower section, eventually settling at the air inlet position, where fresh oxygen continuously fed the reaction.

The experimental campaign also quantified the characteristic indicators that signal a developing fire. The oxygen consumption rate, the generation rates of indicator gases, and the exothermic strength of the oxidation reaction all followed exponential growth trends as temperature climbed. These gas signatures, including carbon monoxide and other products of low-temperature oxidation, are what mine safety engineers monitor in the field. The challenge has always been translating gas concentrations measured in a mine roadway back into an accurate estimate of the coal temperature deep inside a goaf, where no thermometer can reach. That inverse problem is exactly where machine learning excels, provided the model’s internal parameters are tuned correctly.

This is where the grey wolf enters the story. Support vector regression depends critically on hyperparameters, such as the penalty factor and kernel settings, that govern how it fits nonlinear relationships in data. Set poorly, the model underfits or overfits; set well, it generalizes beautifully. Traditionally, engineers have relied on trial and error or grid searches. The grey wolf optimization algorithm instead treats the parameter search as a simulated hunt: candidate solutions are ranked as alpha, beta, delta, and omega wolves, and the pack iteratively encircles and converges on the optimal parameter combination. The researchers found that this optimization step was decisive. On the experimental test set, the root mean square error of the GWO-SVR model dropped to 2.4725, while a grey wolf-optimized back propagation neural network achieved 2.9024, both substantially better than standalone SVR and BPNN models.

Laboratory results alone rarely convince mining engineers, so the team took a further step that distinguishes this work: field verification. They validated the GWO-SVR model against in situ monitoring data from the Sanhejian Coal Mine, a real underground environment with all the messiness that entails, including variable airflow, moisture, and heterogeneous coal distributions. Under these genuine field conditions, the GWO-SVR model posted a root mean square error of just 0.9310 degrees in its temperature predictions, compared with 1.7778 for standalone SVR, 2.5469 for the BPNN, and 1.2791 for the GWO-BPNN. The comparison underscores that the optimization algorithm, not merely the choice of base model, drives the performance gain.

The implications for mine safety are significant. Spontaneous combustion fires in goaf areas are extraordinarily difficult to detect early because they develop out of sight, behind sealed zones, and by the time smoke or elevated carbon monoxide reaches sensors, the fire may already be well established. A model that can convert routinely measured indicator gas concentrations into a reliable temperature estimate gives fire prevention teams a quantitative gauge of how close the coal is to critical stages, allowing targeted interventions such as nitrogen injection, grouting, or adjusted ventilation before conditions become dangerous. The authors’ earlier work, including comparisons of random forest and support vector machine approaches and studies of high-temperature point migration, laid the groundwork for this refined approach.

The study also situates itself within a broader scientific effort to tame coal’s reactivity. Recent research has explored biomass aerogels that inhibit combustion at the microstructural level, shape memory hydrogels and plastogels for fire prevention, and thermokinetic analyses of low-rank coals during low-temperature oxidation. Other teams have applied genetic algorithm-optimized SVR to predict coal temperature from carbon monoxide and used back propagation networks for spatio-temporal temperature prediction. What the new study adds is a rigorous, experimentally grounded pipeline: a large-scale oxidation experiment to characterize the indicators, a hybrid model whose hyperparameters are intelligently optimized, and validation at both laboratory and field scales, a combination that few previous studies have achieved end to end.

From a technical standpoint, the exponential growth patterns observed in oxygen consumption, gas generation, and exothermic strength explain why linear or simple empirical models have historically struggled. The relationship between indicator gases and coal temperature is strongly nonlinear, with different gases dominating different temperature ranges. Machine learning models can capture these nonlinearities, but only if trained on representative data and tuned with care. The GWO-SVR framework addresses both requirements, and its strong generalization performance on unseen test samples suggests it is not merely memorizing the experimental data but learning the underlying physics of coal oxidation.

For an industry still central to global energy supply, and for the hundreds of thousands of miners who work underground every day, tools like this represent a quiet but meaningful advance in the long fight against one of mining’s oldest hazards. The research was supported by the National Natural Science Foundation of China and the Basic Research Program of Shanxi Province. As mines push deeper and coal seams become more prone to self-heating, the ability to forecast the invisible heat before it becomes an inferno may prove not just scientifically elegant but genuinely lifesaving.

Subject of Research: Machine learning prediction of coal spontaneous combustion temperature using grey wolf optimization and support vector regression

Article Title: Enhanced Prediction of Coal Spontaneous Combustion Temperature via GWO-SVR with Experimental and Field Verification

Article References: Lei, C., Qiao, Q., Deng, J., Zhao, J., Cui, C., & Wang, W. (2026). Enhanced Prediction of Coal Spontaneous Combustion Temperature via GWO-SVR with Experimental and Field Verification. Natural Resources Research. https://doi.org/10.1007/s11053-026-10779-9

Image Credits: AI Generated

DOI: 10.1007/s11053-026-10779-9

Keywords: coal spontaneous combustion, grey wolf optimization, support vector regression, machine learning, mine safety, indicator gases, temperature prediction, coal oxidation, nonlinear prediction, field verification, Enhanced, Prediction

The research, whose canonical record is available at https://www.nature.com/articles/s41366-026-02191-3, addresses one of the most persistent gaps in obesity medicine. For decades, the field has relied on demographic and physical baselines — age, sex, starting body mass index, metabolic markers — to anticipate outcomes, yet these variables explain only a modest fraction of the differences observed between participants. The remainder has been attributed loosely to motivation, adherence or circumstance, categories too vague to guide clinical decision-making. By systematically identifying cognitive predictors, the study positions itself within a growing movement to bring the tools of psychological science and cognitive assessment into the routine design of weight-management care.

The logic behind the approach is grounded in well-established models of health behaviour. Contemporary theories of self-regulation describe eating and activity as behaviours governed by an interplay of executive functions — the suite of mental processes that includes working memory, inhibitory control, cognitive flexibility and planning. Inhibitory control, for example, determines how effectively a person can suppress an automatic impulse to eat in the presence of palatable food cues, while working memory capacity influences the ability to hold long-term goals in mind when short-term temptations arise. Cognitive flexibility shapes how readily individuals adapt strategies when a chosen plan collides with real-world obstacles such as travel, stress or social eating occasions.

Each of these capacities varies considerably across individuals, and that variation is precisely what makes them attractive as predictive candidates. If a clinician could estimate, at intake, the strength of a patient’s executive functions, food-related attentional bias, or delay discounting — the tendency to devalue rewards that lie in the future — the argument runs, then treatment could be matched to the person rather than delivered as a one-size-fits-all protocol. A patient with weak inhibitory control might benefit from environmental restructuring that minimises exposure to food cues, whereas a patient with strong planning abilities but poor coping under stress might need a different emphasis entirely. Prediction, in this framing, is the first step toward personalisation.

The study’s central contribution is its effort to move beyond anecdote and small-scale correlational work. Previous investigations have linked individual cognitive measures to weight outcomes in isolation: impulsivity has been associated with poorer adherence to dietary prescriptions, attentional bias toward food cues with greater susceptibility to overeating, and self-regulatory capacity with better maintenance of lost weight. But single-variable studies have often produced inconsistent findings across samples, partly because cognitive traits are correlated with one another and with socioeconomic and emotional factors. A multivariate identification strategy — one that tests a panel of cognitive candidates together against measured intervention outcomes — offers a more rigorous route to knowing which signals genuinely carry predictive weight and which are statistical echoes of other influences.

Methodologically, this kind of research demands careful design. Cognitive factors must be measured with validated tasks or instruments before the intervention begins, so that prediction is genuinely prospective rather than retrospective. Outcomes must then be tracked with standard metrics used across the obesity field, typically percentage change in body weight over defined follow-up periods, alongside secondary indicators such as adherence, attrition and maintenance. Statistical models must account for the established baseline predictors — starting weight, age, sex — so that any additional explanatory power attributable to cognition can be isolated. The strength of the resulting evidence depends on how well these steps are executed, and the field has repeatedly seen promising psychological predictors fade when subjected to this level of scrutiny.

The implications, should cognitive predictors prove robust, extend well beyond the clinic. Public health programmes spend enormous resources on behavioural weight-loss interventions, and the returns are notoriously uneven. Population-level trials often report average weight changes of a few percentage points, figures that conceal a wide distribution in which some participants achieve transformative results while others benefit minimally. Identifying who is likely to respond — and why — would allow scarce clinical resources to be allocated more efficiently, would spare low-likelihood responders from programmes poorly suited to them, and could redirect those individuals toward alternative approaches, whether pharmacological, surgical or differently structured behavioural support.

There is also a scientific payoff. Obesity is increasingly understood as a condition in which neurocognitive processes interact with a food environment engineered to exploit them. Ultra-processed, energy-dense foods are deliberately designed to be hyperpalatable, and the cognitive machinery of inhibition and attention evolved for scarcity is frequently outmatched by abundance. Research that quantifies which cognitive capacities buffer people against this environment — and which leave them vulnerable — feeds directly into theories of why obesity prevalence varies so widely among people exposed to similar surroundings. It also connects the obesity literature to adjacent fields, including addiction science, where cue reactivity, impulsivity and executive dysfunction have been studied for decades as predictors of treatment response.

Cautious interpretation remains essential. Cognitive measures are not destiny: they capture tendencies, not certainties, and they interact with context, motivation and life circumstances in ways that no baseline assessment can fully anticipate. Predictive models built in one population may not generalise to another, particularly across differences in culture, socioeconomic status and the specific design of the intervention. There are also ethical considerations: cognitive profiling of patients raises questions about stigma, consent and the risk of lowering expectations for individuals labelled as poor responders. Researchers in this area generally emphasise that the goal is to tailor support, not to ration it, and that cognitive data should inform the design of better-matched interventions rather than justify withholding care.

Even with those caveats, the study marks a meaningful step in a direction the field has been edging toward for years. The era of treating behavioural weight loss as a uniform prescription is giving way to an era of stratified, psychologically informed care, in which the starting point is a fuller picture of the individual — not just their metabolism and history, but the cognitive architecture they bring to the struggle with food. If cognitive factors identified in this research hold up under replication and validation in independent cohorts, clinicians may one day open a weight-management consultation with a brief cognitive assessment the way they currently open with a blood panel, using the results to choose the intervention most likely to work. For the millions of people who have cycled through programmes that failed them, that prospect — prediction as the foundation of personalisation — is what makes this line of research worth watching closely.

Subject of Research: Cognitive predictors of outcomes in behavioural weight-loss interventions

Article Title: Identification of cognitive factors that predict behavioural weight-loss intervention outcomes

Article References: Arjmand, G., Morys, F. M., Sung, J. J., Duncan, C. C., Davis, X. S., Heshmati, S., Fang, X., White, M. A., Grilo, C. M., & Small, D. M. (2026). Identification of cognitive factors that predict behavioural weight-loss intervention outcomes. International Journal of Obesity. https://doi.org/10.1038/s41366-026-02191-3

Image Credits: AI Generated

DOI: 10.1038/s41366-026-02191-3

Keywords: obesity, weight loss, cognitive factors, behavioural intervention, executive function, self-regulation, inhibitory control, prediction, personalised medicine, International Journal of Obesity, Identification, cognitive

Total soluble solids, typically expressed in degrees Brix, constitute the standard industry metric for sweetness and ripeness in mango. The measurement integrates the concentration of sugars, primarily sucrose, glucose, and fructose, along with smaller contributions from organic acids, amino acids, and other dissolved compounds. Conventional determination requires homogenizing flesh samples and reading refractometry values, a process that destroys the fruit and provides information only about the sampled tissue. Because mangoes display considerable spatial heterogeneity in soluble solids, with gradients from the stem end to the blossom end and from the peel inward toward the stone, destructive sampling introduces uncertainty about whether a single measurement represents the whole fruit. A color-based predictive model circumvents this limitation by estimating quality from the intact exterior, enabling repeated assessment of the same fruit across time.

Vitamin C presents an even greater analytical challenge than soluble solids. Ascorbic acid is labile, oxidizing readily upon exposure to oxygen, light, heat, and enzymes released during tissue disruption. Accurate quantification demands rapid extraction into stabilizing media such as metaphosphoric acid, followed by titration with 2,6-dichlorophenolindophenol or separation by high-performance liquid chromatography. These procedures are time-consuming, reagent-intensive, and subject to artifacts if samples are not handled immediately. The finding that peel reflectance characteristics can predict flesh ascorbic acid content therefore offers substantial practical value, particularly for breeding programs and quality assurance workflows where hundreds or thousands of fruit must be screened rapidly without access to full analytical laboratories.

The scientific rationale linking external color to internal vitamin C rests on shared biosynthetic and catabolic pathways. In climacteric fruit such as mango, the respiratory burst accompanying ripening accelerates reactive oxygen species production, and ascorbic acid functions as a principal antioxidant defense. As ripening proceeds, the balance between ascorbate synthesis, recycling through the glutathione-ascorbate cycle, and irreversible oxidation shifts, producing characteristic declines or plateaus in vitamin C content that coincide temporally with pigment changes in the peel. Both chlorophyll catabolism and ascorbate turnover are modulated by ethylene signaling, harvest maturity, and postharvest storage conditions, creating the statistical covariance that predictive models exploit. This coupling is cultivar-dependent, however, since varieties differ in their carotenoid profiles, ascorbate retention, and the degree to which peel coloration tracks flesh maturity.

Reflectance color measurement itself relies on well-established colorimetric principles, most commonly the CIELAB system, in which L* describes lightness, a* the green-to-red axis, and b* the blue-to-yellow axis. Portable colorimeters or spectrophotometers illuminate a small area of the peel with a standardized light source and record the spectrum or tristimulus values of reflected light. These coordinates can be used directly as predictor variables or transformed into indices such as hue angle and chroma, which often correlate more intuitively with human perception of ripeness. Compared with hyperspectral imaging or near-infrared spectroscopy, simple reflectance colorimetry requires inexpensive instrumentation, minimal training, and no complex spectral preprocessing, making it attractive for deployment in packinghouses, wholesale markets, and even field conditions in producing regions.

Statistical modeling of the relationship between color coordinates and quality attributes typically employs regression frameworks ranging from simple linear models to machine learning approaches such as support vector regression, random forests, and artificial neural networks. Model performance is conventionally evaluated through the coefficient of determination and the root mean square error of prediction on independent validation sets. A recurring theme in the literature is that prediction accuracy for soluble solids generally exceeds that for vitamin C, reflecting the tighter biochemical linkage between pigment development and sugar accumulation than between pigments and ascorbate dynamics. Preharvest factors, including orchard location, canopy position, irrigation regime, and maturity at harvest, introduce variability that models trained on one population may not generalize to another, underscoring the importance of cultivar-specific and season-specific calibration.

The practical implications of validated color-based prediction extend across the mango supply chain. Growers can time harvests more precisely, reducing the incidence of fruit picked too early, which never develops full flavor, or too late, which deteriorates rapidly in transit. Packinghouse operators could sort fruit into ripeness classes non-destructively, enabling targeted distribution so that riper lots reach nearby markets while greener fruit is reserved for long-distance shipping. Retailers might monitor displayed inventory and adjust pricing or discounting based on predicted remaining shelf life. For consumers, the approach underpins the growing interest in smartphone-based applications that estimate fruit quality from photographs, democratizing access to quality information that was previously confined to laboratory settings.

Food loss and waste provide an additional motivation for this line of research. Mangoes are climacteric and highly perishable, with postharvest losses in some producing regions estimated at a substantial fraction of total production. A significant portion of these losses stems from mismatches between fruit maturity and market timing: fruit that appears acceptable externally may be internally underripe or overripe when it reaches the consumer. Objective, non-destructive quality assessment allows interventions such as modified atmosphere packaging, controlled temperature regimes, or accelerated marketing to be applied selectively to fruit predicted to be at risk, rather than uniformly to entire lots. This targeted approach conserves resources and reduces the environmental footprint associated with wasted production inputs.

From a breeding perspective, rapid phenotyping of vitamin C content addresses a persistent bottleneck in developing nutritionally enhanced cultivars. Biofortification efforts aimed at increasing micronutrient content in staple and horticultural crops require screening large segregating populations across multiple seasons and environments. Destructive vitamin C assays limit throughput and consume valuable fruit that breeders may wish to retain for seed or further evaluation. If reflectance color measurements can reliably predict ascorbic acid concentration, breeders could screen far more individuals at earlier stages, accelerating genetic gain. Similar logic applies to soluble solids, a heritable trait that directly influences consumer acceptance and market price, and for which high-throughput indirect phenotyping has long been sought.

Several methodological considerations temper enthusiasm and define the agenda for future work. Color measurements capture only the superficial few hundred micrometers of the peel, so their predictive power depends entirely on statistical association rather than direct sensing of flesh composition. This association can be disrupted by treatments that decouple peel color from flesh maturity, such as ethylene degreening, hot water treatment, controlled atmosphere storage, or the application of skin coatings. Pathogen damage, sap burn, lenticel discoloration, and sunburn alter surface optics without proportional changes in internal quality, potentially biasing predictions. Robust deployment therefore requires either careful fruit selection and cleaning protocols or models that incorporate additional spectral bands beyond the visible range to distinguish genuine ripeness signals from surface defects.

Instrument standardization presents a further challenge. Different colorimeters vary in illuminant geometry, aperture size, and calibration, and ambient lighting conditions influence measurements taken with consumer devices. Efforts to harmonize protocols, publish open calibration datasets, and report colorimetric conditions alongside model coefficients would facilitate comparison across studies and support the development of transferable models. The growing adoption of standardized reporting in food research journals reflects recognition that reproducibility is essential if color-based prediction is to move from academic demonstration to industrial practice. Cultivar-specific calibration databases, updated across seasons and growing regions, would constitute valuable shared infrastructure for the mango industry.

The broader scientific context situates this work within the field of non-destructive food quality evaluation, which encompasses hyperspectral imaging, near-infrared spectroscopy, Raman spectroscopy, acoustic and vibration methods, computer vision, and electronic noses. Each technique occupies a niche defined by cost, speed, penetration depth, and the specific quality attributes it senses most effectively. Visible reflectance colorimetry sits at the accessible end of this spectrum, trading depth of information for simplicity and affordability. Hybrid systems that combine color coordinates with a small number of near-infrared wavelengths, or that fuse color imaging with mass estimation and shape analysis, represent a promising middle ground that could improve prediction of attributes like vitamin C while retaining practical deployability.

Looking forward, the integration of color-based prediction models with digital supply chain infrastructure offers transformative potential. When paired with lot-level tracking, temperature logging, and ripening models, per-fruit color measurements taken at packing could feed dynamic shelf-life forecasts that inform logistics decisions in near real time. Machine learning models retrained continuously on incoming measurement-outcome pairs could adapt to seasonal drift and regional variation. In producing countries where laboratory capacity is limited, validated color-based methods could extend quality assessment capabilities to cooperatives and smallholder aggregation centers, improving bargaining position and reducing losses at the point closest to production. The convergence of inexpensive optical sensing, robust statistical modeling, and mobile computing thus positions external color as a durable and scalable window into the internal quality of mangoes and, by extension, other climacteric horticultural commodities.

Subject of Research: Color-based prediction of mango total soluble solids and vitamin C using reflectance color measurement

Article Title: Color-based prediction of mango total soluble solids and vitamin C using reflectance color measurement

Article References: Kusumiyati, K., Sutari, W., Supratman, U., & Munawar, A. A. (2026). Color-based prediction of mango total soluble solids and vitamin C using reflectance color measurement. npj Science of Food. https://doi.org/10.1038/s41538-026-01120-y

Image Credits: AI Generated

DOI: 10.1038/s41538-026-01120-y

Keywords: Color-based, prediction, mango, total, soluble, solids, vitamin, reflectance, color, measurement, scientific research

Prosocial behavior refers to intentional actions that benefit someone else, whether or not the person helping receives an external reward. In school settings, it can include cooperating with classmates, offering support, sharing resources, defending someone who is being mistreated, or performing small favors for friends and family. Empathy is not a single ability. It includes cognitive processes, such as taking another person’s perspective and understanding what they may be feeling, as well as affective processes involving emotional responsiveness to another person’s condition. The distinction matters because recognizing distress does not automatically lead to assistance. In some circumstances, another person’s suffering can generate self-focused personal distress and avoidance instead of constructive action. Gratitude may help redirect attention outward by highlighting the benefits people receive from others and the value of their relationships. In that sense, the emotion could provide a motivational bridge between noticing another person’s needs and choosing to respond.

To investigate this proposed pathway, the researchers used a quantitative, cross-sectional, correlational design. The sample included 159 students from public schools and 159 from private schools, all between 12 and 16 years old. The average age was 13.2 years, with a standard deviation of 1.5 years. The public-school participants were classified within a low socioeconomic group, while the private-school participants were classified within a middle socioeconomic group using information that included residential stratum, family income, parental education, and occupational stability. Students completed questionnaires in their classrooms during four consecutive weeks of the 2024 academic calendar. Trained research assistants administered the instruments to groups of approximately 20 to 30 students, using standardized instructions and emphasizing voluntary participation, anonymity, and honest responses. The complete session lasted about 25 to 30 minutes. Participation required student assent and written consent from parents or legal guardians, and the fieldwork was reviewed as a no-risk project by Corporación Universitaria Rafael Núñez in Colombia.

The measures were designed to capture different parts of the proposed model. Prosocial behavior was assessed with a 27-item scale covering actions directed toward strangers, friends, and family members. Each category contained nine items rated from one, meaning “not at all like me,” to five, meaning “very much like me.” Empathy was measured with a 15-item Spanish questionnaire for children and early adolescents. Its dimensions included empathic action, perspective taking, emotional contagion, awareness of others, and emotional regulation, with responses recorded on a four-point scale from never to always. Gratitude was assessed with a six-item scale measuring a general tendency to recognize and feel thankful for benefits received from others, using responses from strongly disagree to strongly agree on a seven-point scale. Earlier validation work cited in the study reported acceptable reliability for these instruments in Latin American and Colombian youth, including McDonald’s omega values of .70 to .78 for the empathy dimensions and .75 to .91 for the prosocial behavior dimensions.

The central analysis used structural equation modeling, a statistical technique that allows researchers to estimate relationships among latent constructs that cannot be observed directly. Instead of treating empathy, gratitude, or prosociality as single raw scores, the approach uses patterns across questionnaire items to estimate the underlying variables and then tests pathways between them. The researchers examined whether empathy predicted gratitude, whether both were associated with prosocial behavior, and whether gratitude mediated the relationship between empathy and helping actions. They also conducted multigroup analyses to determine whether the measurement instruments and proposed relationships operated similarly in public and private schools. Measurement invariance is important in this context: before comparing groups, researchers must establish that students in each setting interpret the scales in sufficiently comparable ways. The study reported configural, metric, and scalar invariance for the instruments, indicating that the basic factor structure, factor loadings, and item intercepts were comparable across the two school groups.

The results supported the broad proposed relationship, but its strength varied according to whom students were helping. The model for prosocial behavior toward family members explained 12 percent of the variance, while the model for behavior toward friends explained 42 percent and the model for behavior toward strangers explained 23 percent. These values indicate that empathy and gratitude accounted for a meaningful, but incomplete, share of differences in students’ reported behavior. The model involving friends showed acceptable fit for both groups, with a comparative fit index of .93 in private schools and .84 in public schools, although the root mean square error of approximation was higher for public-school students. The family model showed weaker fit in the public-school group, including a comparative fit index of .73 and an RMSEA of .11. The stranger model also showed weaker fit for public-school students, with a comparative fit index of .77 and an RMSEA of .13. Such values caution against treating the proposed model as a complete explanation of adolescent prosociality.

One of the clearest group differences concerned behavior toward unfamiliar people. Students in public schools scored lower than private-school students on perspective taking, with a small effect size of Cohen’s d = 0.32, and on prosocial behavior toward friends, with a moderate effect size of d = 0.55. The study reported that the empathy–gratitude relationship did not predict helping strangers in the same way in the two groups, and that the public-school model produced a negative predictive value for this outcome. The researchers interpret this pattern in light of the social environments represented in the sample, including socioeconomic vulnerability, insecurity, distrust, and stronger orientation toward close relational networks. In circumstances where social risk is perceived as high, helping a stranger may carry different meanings or costs than helping a family member or friend. Empathy and gratitude may still encourage solidarity, but that solidarity could be directed primarily toward people regarded as trusted or familiar. The finding therefore does not show that public-school students lack empathy or gratitude, nor that they are generally less prosocial.

The study’s cross-sectional design also limits what can be concluded. Because all measures were collected at one point in time, the results demonstrate associations rather than causation. They cannot establish that empathy produces gratitude, that gratitude produces helping, or that a school program designed around either quality would necessarily increase prosocial behavior. It is also possible that students who regularly help others develop greater empathy and gratitude through those experiences, creating a reciprocal developmental process. The authors note that some structural models had fit indices below conventional recommendations and call for more flexible models and additional contextual variables. Future studies could examine school climate, perceived safety, parenting practices, social support, sense of belonging, emotion regulation, and community conditions. Longitudinal research would be particularly valuable because it could track how emotional skills and prosocial actions change across adolescence and whether improvements in one precede changes in the other.

Despite these qualifications, the findings offer a practical direction for education systems seeking to promote supportive school climates. Programs could combine perspective-taking exercises with opportunities for students to recognize assistance, express appreciation, and connect gratitude with concrete cooperative actions. The results suggest that such efforts may be relevant in both public and private schools, while also indicating that the social context must shape how activities are designed. In schools serving communities affected by insecurity or economic strain, building trust and safety may be necessary before encouraging students to extend helping behavior beyond immediate relationships. The researchers conclude that empathy and gratitude are consistent predictors of prosociality across the school contexts studied, especially toward family and friends. Their work does not reduce kindness to two psychological traits, but it identifies a potentially important mechanism through which emotional understanding can become social action—and shows why that mechanism must be examined within the real environments where young people live and learn.

The study’s comparison across school types is scientifically important because socioeconomic context can affect not only behavior itself but also how students interpret questionnaire items. Configural invariance indicates that the same broad pattern of dimensions was present in both groups; metric invariance supports comparing associations among those dimensions; and scalar invariance permits comparisons of average latent scores. Together, these tests strengthen the basis for group comparisons, although they do not eliminate differences in lived experience or guarantee that every item carries precisely the same practical meaning for every student.

The mediation result is best understood as a proposed statistical pathway rather than a confirmed sequence of psychological events. In a cross-sectional model, students who report greater empathy may also report greater gratitude and prosociality, but the data cannot determine whether gratitude develops afterward, whether helping experiences cultivate both qualities, or whether another factor influences all three. This distinction matters for intervention design. School programs can reasonably use the findings to test activities that connect perspective-taking and appreciation with cooperative behavior, while evaluating outcomes prospectively and monitoring whether effects differ by target of assistance, classroom climate, and perceived safety. Such testing could clarify whether gratitude is a modifiable mechanism or primarily a marker of broader supportive relationships.

Subject of Research: Empathy, gratitude, and prosocial behavior among Colombian school students

Article Title: Study on the prediction of prosociality through empathy and gratitude in the school context of Cartagena de Indias, Colombia

Article References: Tezón, M. I. (2026). Study on the prediction of prosociality through empathy and gratitude in the school context of Cartagena de Indias, Colombia. Journal of New Approaches in Educational Research, 15(1), Article 24. https://doi.org/10.1007/s44322-026-00054-3

Image Credits: AI Generated

DOI: 10.1007/s44322-026-00054-3

Keywords: empathy, gratitude, prosocial behavior, adolescent development, school psychology, Colombia, structural equation modeling, social-emotional learning, prediction, prosociality, school, context