Authors: Siyi Zhong, Xin Tang, Xiaojun Cheng, Yafeng Pan
Categories: Research, Bodily maps, Embodied cognition, Academic emotions, Subject-specific feelings, Gender differences, Academic achievement
Source: BMC Psychology
Authors: Siyi Zhong, Xin Tang, Xiaojun Cheng, Yafeng Pan
Learning is an embodied process in which emotion and cognition converge through the body’s expressive patterns. However, the bodily manifestation of learning-related emotions, potential gender differences in these patterns, and their impact on academic performance remain unclear. We examined high school students’ bodily sensation maps (BSMs) during learning across different school subject contexts within the Chinese high-school curriculum, and assessed their associations with gender and academic achievement.
This study mapped high school students’ BSMs across subject-specific feelings and academic emotions, examining gender differences and links with academic achievement. A total of 588 students marked body regions of increased or decreased activity on two-dimensional silhouettes in response to nine subjects and five academic emotions. Statistical analyses examined learning-related BSM patterns, gender differences, and relationships with academic performance.
Distinct embodied profiles humanities subjects primarily activated the head and distal upper limbs, whereas science subjects engaged the head, chest, and proximal upper limbs. Positive emotions elicited widespread bodily activation, while negative emotions induced localized or global deactivation. Gender differences were minimal in learning contexts but evident for anxiety, with females showing stronger head and torso activation. BSMs were positively correlated with academic achievement (particularly in English, Physics, Chemistry, Biology, and History), suggesting that bodily responses mirror the interplay of cognitive engagement and emotional arousal.
These findings reveal systematic embodied signatures of subject-specific feelings and academic emotions, elucidating how cognitive and emotional processes are integrated in the body. They provide actionable insights for personalized, emotion-sensitive, and domain-specific educational practices.
Learning is the core mechanism through which individuals adapt to academic and environmental demands, shaping both cognitive development and academic success [1]. In high school, students progressively enhance their disciplinary understanding and cognitive competence through structured instruction and cumulative experience. Different academic domains engage distinct cognitive science subjects such as mathematics and physics depend on logical reasoning and analytical problem-solving, whereas humanities subjects like Chinese and history emphasize comprehension, interpretation, and expression [2–4]. Persistent gender disparities have been documented across these domains, with boys often excelling in science and girls performing better in humanities [5–11]. Exploring students’ conscious emotional experiences during learning, and how these experiences differ by gender, may therefore yield critical insight into the mechanisms that underlie academic development.
Learning evokes a spectrum of conscious emotional experiences that shape attention, motivation, and achievement [12–14]. These emotions can be broadly divided into subject-specific feelings (emotion embedded in a subject) and academic emotions (emotion toward studying and achievement in general). This distinction is introduced as a working conceptual differentiation grounded in existing emotion theories, specifically Pekrun’s work on academic emotions [12], and is used to refine and capture domain-specific emotional experiences related to different academic contexts. Subject-specific feelings refer to domain-specific emotional experiences that arise when students engage with particular academic subjects. They are tightly linked to the content, demands, and cognitive style of a discipline. For example, solving a challenging physics problem may evoke feelings of tension or excitement, while reading literature may elicit curiosity or empathy. These feelings are thus context-bound, reflecting how learners emotionally experience specific subject matter. Academic emotions, in contrast, are domain-general emotional states that occur within academic contexts regardless of subject. They include emotions such as enjoyment, hope, anxiety, boredom, or pride [12], which are tied to learning, achievement, and classroom experiences rather than any particular discipline. Academic emotions influence motivation, attention, and performance across subjects [12, 15–17].
Despite extensive research on the cognitive and emotional determinants of academic achievement, the embodied dimension of learning remains largely overlooked [18]. Students often exhibit salient bodily responses during learning and emotional experiences (responses that may index cognitive load, attentional engagement, and affective arousal), yet their patterns and underlying mechanisms are poorly understood. Theories of embodied cognition propose that bodily engagement externalizes cognitive processes during learning [19–24], while embodied emotion frameworks posit that emotional experiences arise not only from cognitive appraisal but also from bodily responses to emotional stimuli and their interpretation [25–28]. Within this framework, bodily sensation maps (BSMs) provide a quantifiable means to visualize and examine the embodied signatures of learning-related emotions. Prior studies have demonstrated that BSMs are stable and distinct across basic or situational emotions [29, 30]. For instance, anger is typically associated with heightened bodily sensations in the upper limbs and chest, reflecting increased physiological arousal and readiness for action. Similarly, different cognitive processes manifest distinct embodied patterns. Memorizing tends to produce localized sensations in the head region, reflecting the activation of working memory and information retrieval processes that demand focused yet sustained mental effort. In contrast, thinking evokes broader and stronger activations across the head, extending beyond those seen during memorizing, and indicating deeper engagement of cognitive control, reasoning, and attentional processes [30, 31]. However, their patterns across academic domains and emotions remains largely uncharted. Furthermore, potential gender modulation of these embodied responses, and their links to academic achievement, has yet to be systematically investigated.
To address these gaps, we examined high school students’ BSMs during learning in different school subject contexts, and assessed their associations with gender and academic achievement. Our aims were first, to characterize BSM patterns across subject domains and academic emotions, and to test for gender differences; second, to determine whether BSM features predict academic performance, thereby elucidating how embodied experiences are associated with learning outcomes. We hypothesized that BSMs would exhibit distinct and stable configurations across subjects and academic emotional contexts; that gender would modulate these embodied patterns; and that BSM intensity or topology would be associated with academic achievement. By integrating subject-specific learning, academic emotion, and embodiment, this study advances a cognitive-emotional framework linking emotional and bodily processes to educational performance, offering a theoretical and empirical basis for personalized learning strategies.
A total of 588 first-year high school students (341 females; M ± SD = 15.90 ± 0.50 years) from a mid-level school in Southwest China participated. The sample size was determined by the full availability of the target cohort at the participating school and is comparable to, or larger than, samples used in previous studies employing the emBODY paradigm [29, 30]. Most were right-handed (n = 564), with the reminder left-handed (n = 24). All participants had normal or corrected-to-normal vision and no history of motor or neurological disorders, and none had previously taken part in similar experiments. Informed consent was obtained from all participants after explanation of the study aims and procedures. The study adhered to the ethical principles of the Declaration of Helsinki and was approved the ethics committee of Zhejiang University (No. 2022009).
The experimental program was adapted and optimized from previous work, with the front-end implemented in Hyper Text Markup Language (HTML) and the back-end logic in Hypertext Preprocessor (PHP) [30]. It was deployed on Alibaba Cloud servers to provide public access to the Chinese version of the experiment, with all data transmitted and stored using encrypted, isolated processes to ensure privacy. Experiments were conducted with identical computer models and standardized display settings to ensure uniform stimulus presentation.
After providing informed consent, participants’ demographic and academic information was collected, including gender, age, weight, height, medical history, handedness, and recent midterm grades and class rankings. On-screen instructions guided participants through the procedures to minimize potential misunderstandings.
Fifteen verbal cues served as stimuli to probe participants’ bodily representations of emotions and academic five emotions (enjoyment, hope, anger, anxiety, and boredom), nine subjects (Chinese, Mathematics, English, Physics, Chemistry, Biology, Politics, History, and Geography), and a neutral state. Each stimulus was presented once in a randomized order to control for sequence effects.
During the task, participants viewed two abstract, two-dimensional human silhouettes accompanied by stimulus words in randomized order. They were instructed to reflect on their bodily sensations during specific subject-domain learning or emotional experiences. Specifically, participants were instructed to color body regions where they subjectively experienced internal bodily activity as “increasing or getting stronger” or “decreasing or getting weaker” (Fig. 1), with explicit emphasis on internally felt sensations rather than overt movements, task performance, or objective physiological signals, following established body sensation mapping protocols. Participants were not required to provide responses for each of stimulus. Stimulus words were embedded within a standardized instruction format (e.g., “For the pictures below, evaluate how the activity of your body changes when you learn/feel X”), where the target word varied depending on the condition. For subject-domain learning, “X” was replaced with academic subjects such as “Physics”, “Mathematics”. For emotional experiences, it was replaced with academic emotions such as “Anxiety”, “Enjoyment”. Areas of increased activity were colored on the left silhouette, and areas of decreased activity on the right silhouette using the emBODY tool [30] (Fig. 1A). Trials were self-paced, with an option to reset in case of coloring errors. Coloring was performed by dragging the mouse over the body template (12-pixel tool diameter), with repeated strokes increasing opacity (Fig. 1B). Each silhouette contained 50,364 pixels, and resulting images were stored as matrices with intensity values from 0 to 100. Activation and deactivation maps were then combined by adding corresponding pixel values, following previous emBODY pipelines [30], to create a single body sensation map for each stimulus (Fig. 1C).
Fig. 1The emBODY tool. **A **Participants indicated bodily sensations by coloring blank areas of increased activity on the left and decreased activity on the right. **B **Each body was represented by 50,364 data points, with activation and deactivation recorded as integer values. **C **The resulting maps were combined to create a single body sensation map for subsequent statistical analyses
Data preprocessing followed established procedures [30]. Participants who left more than the mean + 2.5 SDs of body areas uncolored, or who completed fewer than 14 of 15 stimuli, were excluded. Activation and deactivation maps were combined into a single body sensation map (BSM) representing the spatial distribution of bodily activity **(**Fig. 1C). Manual inspection removed artifacts such as symbolic drawings or random scribbles. After quality control, 576 valid datasets (334 females) were retained for further analyses.
Statistical analyses were performed in MATLAB R2024a (Fig. 2). First, BSMs for subject-specific feelings and academic emotions were constructed using one-sample t-tests, with false-discovery-rate (FDR) correction to account for multiple comparisons across the whole-body template (threshold at q < 0.05). Statistically significant regions of increased or decreased activation were visualized to represent the bodily correlates of each learning experience and emotional state.
Second, sex differences were evaluated by performing one-sample t-tests on male and female BSMs separately, with FDR correction applied (threshold at q < 0.05). Independent-samples t-tests were then used to identify regions showing significant male-female differences, generating difference maps in which positive and negative values indicate greater activation in males and females, respectively. Spearman correlation coefficients were calculated between male and female difference maps under each academic subject condition to quantify the consistency and divergence of bodily response patterns. Furthermore, to examine sex-related differences across and within conditions, the number of activated and deactivated pixels was also calculated separately for each sex and condition; within-group differences were tested with Friedman’s ANOVA, and between-group differences with the Mann–Whitney U test.
Finally, associations between bodily sensation and academic achievement were assessed using pixelwise generalized linear models (GLM) with FDR correction (threshold at q < 0.05). Regions with positive values indicate a significant positive correlation between bodily sensation and academic achievement, whereas regions with negative values indicate a negative correlation.
Fig. 2Statistical analysis workflow. Flowchart summarizing the main analyses of BSM one-sample t-tests to identify bodily activation and deactivation associated with subject-specific feelings and academic emotions; independent-samples t-tests to compare male and female patterns; and pixelwise generalized linear models (GLMs) to examine associations with academic achievement. False discovery rate (FDR) correction was applied in all analyses
After preprocessing, the final sample included 242 male participants (age 15.97 ± 0.51 years; 94% right-handed) and 334 female participants (age 15.85 ± 0.49 years; 98% right-handed). Independent-samples t-tests revealed that female participants scored significantly higher than males in Chinese, English, and Politics (ps < 0.02), whereas males outperformed females in Mathematics, Physics, Chemistry, Geography, and overall academic achievement (ps < 0.05). No significant gender differences were observed in Biology or History (ps > 0.06) (Table 1).
Table 1Gender differences in academic performanceMean academic scoresMale (N = 242)Female (N = 334)t p Chinese82.42 ± 7.9384.12 ± 7.66-2.550.011Mathematics78.98 ± 22.3771.33 ± 20.214.16< 0.001English79.79 ± 19.2885.82 ± 19.52-3.64< 0.001Physics58.88 ± 15.0450.19 ± 15.566.66< 0.001Chemistry59.71 ± 15.6854.06 ± 16.284.15< 0.001Biology63.91 ± 17.8563.37 ± 15.490.380.707Politics53.00 ± 9.9055.50 ± 9.41-3.000.003History72.03 ± 9.9673.57 ± 9.45-1.850.065Geography69.58 ± 11.9566.66 ± 10.922.960.003Total score617.99 ± 82.58603.97 ± 84.361.970.049p* < .05 ; p < .01
High school students’ bodily sensation maps revealed subject- and emotion-specific patterns (Fig. 3A). Humanities subjects (Chinese, English, Politics, History) were primarily associated with increased activation in the head and distal upper limbs, accompanied by deactivations in the lower limbs. Geography, though classified as a humanities subject, exhibited a pattern more akin to the sciences (Biology). Science subjects (Physics, Chemistry, Biology) generally showed stronger activation in the head, chest, and proximal upper limbs, whereas Mathematics displayed a distinct pattern within the sciences, with increased activity in the head, upper limbs, and upper torso.
Academic emotions also exhibited distinct bodily signatures. Positive emotions (hope, enjoyment) were characterized by increased activity in the head, upper limbs, and upper torso. Negative emotions differed by anger elicited widespread activation, anxiety selectively engaged the head and chest, boredom induced broad deactivation, and the neutral state decreased activity in the lower limbs.
Fig. 3Bodily sensation maps of subject-specific feelings and academic emotions among high school students. **A **Pixelwise t-statistics showing regions of increased (warm colors) or decreased (cool colors) activity across subject-specific feelings and academic emotions (q < 0.05). **B **Maps for male (top) and female (bottom) participants. **C **Sex-difference t-test warm colors indicate greater activity in males, cool colors in females (q < 0.05), N.S., nonsignificant. A vertical dashed line separates subject-specific feelings (left) from academic emotions (right)
Sex-specific analyses revealed largely similar bodily responses across academic subjects (Fig. 3B). Independent-samples t-tests identified a significant sex difference only for anxiety, with females showing stronger head and torso activation than males (ps < 0.05, qs < 0.05; Fig. 3C).
Spearman correlation coefficients between male and female BSMs (Fig. 4) showed the highest consistency for science subjects (Physics: Spearman r = .938 [0.937, 0.939], p < .001; Mathematics: Spearman r = .920 [0.919, 0.921], p < .001; Biology: Spearman r = .910 [0.908, 0.911], p < .001), whereas correlations were lower for negative and neutral emotions (Neutral: Spearman r = .644 [0.639, 0.649], p < .001; Anxiety: Spearman r = .607 [0.601, 0.612], p < .001).
Fig. 4Spearman correlations of bodily sensations by sex. Spearman correlation coefficients between male and female BSMs are shown for each subject-specific feeling and academic emotion. Example maps depict male (left) and female (right) responses for the most similar (Physics) and least similar (Anxiety) experiences. Error bars for 95% confidence intervals are omitted due to negligible size. A horizontal dashed line separates subject-specific feelings (left) from academic emotions (right)
Analysis of the number of activated and deactivated pixels per stimulus by sex (Fig. 5) indicated significant effects of both subject-specific feelings and academic emotions in males (Friedman ANOVA, χ² = 438.1026 and = 139.4389, ps < 0.001) and females (χ² = 643.1968 and 176.3039, ps < 0.001). Mann–Whitney U tests showed that, compared with males, females marked significantly more activation pixels for anxiety (U = 26683.00, Z = -6.230, p < .001, q < 0.05) and significantly fewer deactivation pixels for neutral emotion (U = 32082.00, Z = -3.393, p < .001, q < 0.05), with no significant differences for other subjects or emotions (all U < 38509.50, Z < − 0.016, q > 0.05).
Fig. 5Number of activated and deactivated pixels by sex. Box-and-whisker plots show medians (horizontal line) and interquartile ranges; whiskers extend to 1.5 × Interquartile Range, with all individual data points (including outliers) shown. * p < .05, FDR corrected
GLM analyses further revealed modest positive associations between body sensations in the head and torso regions and academic performance in the corresponding subject domains (in particular, English, Physics, Chemistry, Biology, and History). In contrast, for academic emotions, only head activity during boredom showed a weak negative correlation with overall academic achievement (Fig. 6).
Fig. 6GLM regression maps of bodily activity and academic achievement. GLM regression maps linking bodily activity to academic activity for subject-specific feelings corresponds to subject-specific grades, and activity for academic emotions corresponds to overall performance; warm = positive, cool = negative correlations (q < 0.05). Beta-values are depicted, with warm colors indicating positive correlations and cool colors indicating negative correlations (q < 0.05). N.S., nonsignificant, indicating no significant correlation
The present study investigated bodily sensation maps (BSMs) in high school students across different subject-specific learning contexts and academic emotional experiences, revealing both domain-specific and domain-general patterns. The findings demonstrate that the embodiment of cognitive and emotional processes conforms to the principles of embodied cognition and embodied emotion theories, which posit that cognition and affect are inherently linked with bodily states [19, 21, 22, 24, 28].
In subject-domain learning, humanities subjects such as Chinese and history primarily engaged linguistic comprehension and memory processing, as reflected in participants’ reported bodily sensation maps by activations in the head and distal upper limbs. By contrast, science subjects such as physics and chemistry, which rely more on logical reasoning and problem solving, elicited activations in the head, torso, and proximal upper limbs. Consistent with previous research, activations in the head region are commonly interpreted as subjective bodily correlates of high-level cognitive engagement, as a variety of cognitive functions such as thinking, attention, memory, and reasoning are closely associated with head sensations [31, 32]. Thus, the widespread head activation observed across both humanities and science learning contexts may indicate a heightened sense of cognitive involvement during learning, as perceived by participants. Furthermore, activations in the chest region have been linked in prior BSM research to perceived physiological arousal, such as changes in heart rate or breathing, which are commonly reported across both positive and negative emotional experiences [33]. These subject-dependent topographies likely mirror differential cognitive demands of language-based versus analytic reasoning processes and suggest that variations in reported bodily activity may index differences in perceived cognitive investment [34–36]. This interpretation accords with the embodied language comprehension framework, which emphasizes that meaning is grounded in sensorimotor representations and that cognitive processing is intrinsically linked to bodily states [37–42]. Importantly, readers should bear in mind that BSMs capture subjective bodily representations rather than direct measures of cognitive or physiological processes.
Distinct embodied profiles also emerged across academic emotions. Positive emotions such as hope and enjoyment were associated with widespread activation in the head, upper limbs, and torso, as reflected in participants’ bodily sensation maps, consistent with heightened motivational engagement and goal-directed arousal at the level of subjective bodily experience. In contrast, negative states such as anxiety and boredom produced localized or global deactivations, which may be interpreted as reflecting participants’ subjective bodily experiences associated with diminished attention, avoidance tendencies, and reduced cognitive resource allocation [30, 32, 43, 44]. Friedman analyses confirmed significant within-subject variability across both learning and emotional conditions, suggesting that BSMs are context-sensitive yet exhibit systematic, group-level patterns. Together, these findings reinforce the view that bodily responses as captured by BSMs were associated with perceived cognitive–emotional load during learning.
Sex-specific analyses revealed largely convergent embodied profiles across male and female students. Spearman correlations indicated particularly strong cross-gender similarity in science subjects, suggesting that both sexes recruit comparable perception-action systems under high cognitive load [23, 45, 46]. However, gender similarity decreased in emotional correlations were lower for neutral and negative emotions, and females showed stronger head and torso activations under anxiety—a pattern that may reflect differences in subjective bodily experiences associated with anxiety, consistent with higher self-reported anxiety in females [47, 48]. This pattern aligns with the pixel-based analyses presented, which further suggest that females may experience relatively stronger bodily sensations under anxiety while exhibiting somewhat more stable responses than males in neutral states, in accordance with previous reports showing that adolescent females tend to report higher anxiety levels than males [49, 50]. These results imply that, while the embodiment of domain-specific learning experiences is largely gender-invariant, sex-dependent modulation is primarily observed in emotional contexts, particularly anxiety.
Associations between bodily activity and academic performance further revealed that activation in the head and torso correlated positively, albeit modestly, with subject-specific achievement, suggesting that embodied engagement may be associated with cognitive control and sustained attention [51–53]. Conversely, head deactivation during boredom was negatively associated with overall academic performance, implying that attenuated embodied engagement could reflect reduced motivation or attentional withdrawal [54]. These results highlight the potential of BSMs as nonverbal indicators of learning engagement and as a window into the cognitive–emotional mechanisms underpinning academic success. While modest associations were also observed in upper-limb regions for specific subjects (e.g., English and Chemistry; Fig. 6), these effects were less consistent across domains than those involving head and torso regions.
Despite yielding important findings, the present study has several limitations. First, although it addressed both subject-specific learning and academic emotions, the measurement of academic emotions was limited to Pekrun’s classical five categories and did not encompass additional dimensions developed in subsequent research [12–14, 55–59]. This may underestimate the complexity of emotional experiences and their embodied manifestations. Future studies could explore a broader range of academic emotional dimensions, potentially incorporating qualitative methods, such as interviews, to capture richer, more nuanced emotional experiences. Second, the cross-sectional design precludes conclusions regarding the dynamic changes of bodily responses during learning. Future research could adopt longitudinal designs combined with multimodal approaches, integrating physiological, behavioral, and classroom data, to provide a more comprehensive characterization of the dynamic embodiment of subject-specific learning and academic emotions. Furthermore, the sample was limited to high school students, so the generalizability of the findings requires verification across other educational stages. Future studies could extend the investigation to elementary and middle school students to examine the stability and variation of embodied patterns across developmental stages. Finally, the bodily maps maybe shaped to some extent by response strategies or task demand characteristics. Although participants were not forced to mark a location on every trial and individuals with insufficient valid responses were excluded, such influences cannot be fully eliminated.
In sum, high school students exhibit systematic, interpretable bodily sensation patterns across learning domains, with gender selectively modulating embodied responses under emotional conditions. Moreover, these bodily signatures relate to academic achievement, bridging the gap between cognitive–emotional processes and bodily expression. The results provide empirical support for embodied frameworks of learning and suggest practical avenues for developing responsive pedagogical strategies that integrate emotional regulation, domain-specific engagement, and individualized learning support. For example, for male students, who show more deactivation during neutral emotional states, teachers could engage students by incorporating more interactive and dynamic activities, such as group discussions or competitive tasks, to stimulate emotional engagement and increase classroom participation. For female students, whose bodily sensations are more activated under anxiety, teachers might focus on integrating emotional regulation strategies, such as incorporating relaxation exercises or providing emotional support, to help manage anxiety and enhance focus. These adjustments would be designed to create a more tailored and emotionally supportive learning environment. Future research should employ longitudinal and multimodal designs, combining physiological, behavioral, and classroom data, to elucidate the dynamic interplay between embodiment, emotion, and cognition in real-world educational settings [60, 61]. In addition, studies could examine a broader range of academic emotional dimensions beyond Pekrun’s classical five categories and extend investigations to younger students, in order to explore developmental variations and the stability of embodied patterns across educational stages.