The integration of psychological education and moral dilemmas from a value perspective

During the contemporary social transition period, the contradictions between psychological adaptation and moral choices triggered by value pluralization are becoming increasingly prominent1. Psychological education emphasizes satisfying individual needs, whereas moral education focuses on the internalization of social norms, often revealing a theoretical schism in the tension between “individual well-being” and “collective good”. Existing research indicates that 17.3% of university students in China exhibit a dissociation between moral cognition and psychological behavior, a rift urgently requiring resolution through educational coordination mechanisms. In recent years, deep learning has won numerous competitions in the fields of pattern creation and machine learning. The distinction between shallow learning and deep learning lies in the depth and pathways of learning, which represent causal relationships in learning actions and depth [1].

Research in deep learning, unsupervised learning, reinforcement learning, and evolutionary computation are all applicable to addressing the contradictions between psychological and moral education. Against the backdrop of the digital era, the exponential development of internet technology is reshaping value transmission paradigms, exacerbating the binary division between traditional psychological and moral education. Studies show that in online environments, the correlation coefficient between individual moral judgment and psychological adaptability drops to 0.32, reflecting deep-seated contradictions between value cognition and emotional experience. Current research faces three major limitations:

1. 1.

   The internalization mechanisms of moral norms and satisfaction processes of psychological needs are often discussed in isolation;

2. 2.

   Traditional survey methods struggle to capture dynamically evolving features of value conflicts;

3. 3.

   There is a lack of quantitative tools to model the complex interactive network of “cognition-affection-behavior”.

In the process of integrating psychological and moral education, deep learning provides quantitative analytical frameworks and practical tools through technological pathways such as data modeling, feature fusion, and dynamic prediction. The following elaborates on its contributions from three dimensions: technical principles, application scenarios, and implementation methods.

Core technical support from deep learning

Feature fusion modeling of Multi-Source data

Cross-modal feature extraction

CNN processes textual moral case descriptions, while RNN analyzes psychological state time-series data, with attention mechanisms establishing inter-feature mappings (e.g., linking “Prisoner’s Dilemma” choices to anxiety levels via BiLSTM).

Latent feature mining

Autoencoders quantify abstract constructs (e.g., empathy) by extracting high-dimensional features from unstructured data (e.g., facial expressions via CNN correlated with moral judgment surveys).

Dynamic relationship prediction

Causal simulation

Transformers model how psychological interventions (e.g., mindfulness training) affect moral behaviors by analyzing weekly counseling records and campus behavior data.

Riskearly-warning model

XGBoost-MLP hybrids predict extreme moral deviations by integrating depression scores and online aggression metrics.

Application scenarios

Personalized education design

Feature Tagging and Clustering: Using deep learning for unsupervised clustering (e.g., DBSCAN algorithm) of student data, classifying students into types such as “high empathy - low moral cognition” and “high moral judgment - high anxiety”, and designing integrated programs for different groups. For example, for the “cognitive-emotional imbalance” group, psychological training games integrated with reinforcement learning are designed, incorporating moral dilemma scenarios into tasks, with the model adjusting training difficulty in real time.

Adaptive Learning Path Generation: Constructing an educational decision-making model using Deep Q-Network (DQN), which automatically recommends integrated courses based on students’ performance in psychological tests (e.g., Big Five Personality Inventory) and moral scenario tests. For instance, if a student performs poorly in “honesty tests” but has high anxiety, the model suggests a combined training of “emotional management + integrity case analysis”.

Quantified outcome evaluation

Multi-Metric Fusion Evaluation System: Traditional evaluations struggle to quantify the cross-impact of psychology and morality, while deep learning can optimize multiple objectives simultaneously through multi-task learning. For example, using CNN to analyze moral emotional expressions in students’ compositions and LSTM to analyze emotional fluctuations in psychological logs, evaluating the collaborative improvement of “moral cognition - psychological state” through a joint loss function.

Counterfactual Reasoning Verification: Simulating scenarios without integrated education using Generative Adversarial Networks (GAN) to compare with actual educational effects. For instance, using GAN to generate behavioral data of students who “only received psychological education”, and inputting both real “psychological + moral integrated education” data and simulated data into a contrastive learning model to quantify the contribution of integrated education to moral behavior improvement.

Characteristics of psychological education

All forms of education inherently involve value issues, and psychological education is no exception. Rooted in the principles of psychology, the feasibility and effectiveness of psychological education are grounded in its focus on individual psychological development. This approach addresses evolving psychological issues as individuals grow, including self-cognition, social cognition, and the processes of recognition or rejection during socialization. As individuals become more integrated into society, their subjective freedom increasingly encounters constraints imposed by societal norms. This dynamic leads to the accumulation and formation of conflicts between personal desires and societal expectations. To navigate these challenges, individuals require a rational new identity that can evaluate the relationship between themselves and society, making informed judgments about various interpersonal relationships. This need reflects the individual’s subjectivity and offers an effective pathway for conflict resolution. The inherent contradiction between the individual and society is evident: any external entity poses a counterpoint to individual subjectivity. However, the collective and relational nature of society, coupled with individuals’ social psychological needs, means that they cannot exist in isolation from others and groups. Consequently, the conflict between the individual and others is both eternal and absolute. Nonetheless, the malleability and social psychological needs of individuals provide opportunities for addressing and resolving these conflicts. Encourage individuals to develop a deep understanding of their own identities, strengths, and weaknesses. This awareness fosters a stronger sense of self and enhances the ability to navigate social interactions. Equip individuals with the skills necessary to understand societal norms, cultural contexts, and interpersonal dynamics. This understanding helps bridge the gap between personal desires and societal expectations. Teach techniques for managing and resolving conflicts between personal freedoms and societal constraints. These skills empower individuals to find balanced solutions that respect both personal and communal values. Foster environments where individuals feel supported and valued, reducing feelings of alienation and promoting a sense of belonging. Strong support systems contribute to the resolution of conflicts by providing emotional and practical assistance. By focusing on the individual’s subjectivity, psychological education can play a pivotal role in helping individuals navigate the complexities of personal and social life. This approach not only supports personal growth but also contributes to the development of well-adjusted, socially responsible citizens who are capable of contributing positively to their communities.

Psychological education is fundamentally grounded in attention to the individual’s psychological state and their developmental process. Originating from Wilhelm Wundt’s laboratory as a neutral discipline, psychology is rooted in direct experience, encompassing social influences encountered during socialization. Consequently, the subjectivity addressed by psychology involves the continuous updating and decision-making processes of individuals within the context of their social environments. Some argue that focusing on subjectivity within psychological education inherently aims at promoting socialization, potentially diminishing the ontological significance of individual subjectivity. While this perspective holds some validity, it requires further exploration. The interplay between the individual and society—comprising both conflict and integration—rests upon the existence of the individual’s subjectivity and its ongoing renewal and resilience amidst complex interpersonal relationships. Socialization is an inevitable process; however, the autonomous existence of subjectivity is not guaranteed. For instance, negative emotional states such as apathy, depression, and despair signify a loss of subjectivity. In severe cases, these feelings may culminate in life-threatening behaviors or a profound loss of spiritual drive and aspirations, effectively representing another form of existential demise. Therefore, nurturing subjectivity remains one of the most critical aspects and needs within psychological education. To illustrate the classification of psychological characteristics relevant to this discussion, refer to Fig. 1 (Note: As there is no actual figure provided, you should insert the appropriate image or diagram here according to your research).

Classification of relevant conditions for psychological characteristics

Moral dilemmas

Moral norms are characterized by their social orientation, necessitating conformity to collective expectations and imposing significant constraints on individual behavior. Traditional Chinese culture places a strong emphasis on ethical conduct, encapsulated in the maxim “do not look at what is improper, do not act improperly.” This principle mandates adherence to moral and ethical standards, prohibiting behaviors or speech that deviate from these norms. Such strict social regulation can lead to the suppression of individual needs, desires, and emotions. When these aspects remain unexpressed, they often manifest as anxiety, excessive caution, suspicion, and even obsessive-compulsive personality traits. In this cultural context, virtues such as “endurance” and “modesty” are highly valued but can sometimes contribute to psychological disorders. Endurance involves self-control and self-repression, which can inhibit normal biological processes and the expression of instinctual desires. This repression may result in anxiety, depression, and other neurotic tendencies. Modesty, while promoting humility and respect for others, can also foster low self-esteem and hypersensitivity to external judgments, further exacerbating psychological distress. The interplay between moral norms and psychological health reveals that moral conflicts often underlie various psychological conflicts. For instance, individuals with obsessive personalities frequently experience moral pain without deriving the pleasure of moral satisfaction. They seek external validation to bolster their self-esteem, leading to low confidence, hypersensitivity, and a constant state of insecurity. These factors can eventually culminate in psychological disorders. Encourage individuals to recognize and express their emotions in a healthy manner. This practice helps mitigate the negative effects of emotional suppression and fosters better mental health. Foster an environment where individuals feel accepted and valued for who they are, reducing the need for external validation and enhancing self-esteem. Establish support systems that offer counseling and therapeutic interventions to help individuals navigate moral conflicts and manage associated psychological symptoms. Ensure that psychological interventions are culturally sensitive, acknowledging the influence of traditional values on mental health and tailoring treatments accordingly. By addressing the psychological impacts of moral norms and cultural values, it is possible to promote healthier coping mechanisms and improve overall well-being. This approach supports personal growth and contributes to the development of resilient, socially responsible individuals capable of balancing personal needs with societal expectations.

Moral education aspires to cultivate an ideal personality, yet the discrepancy between this ideal and reality can lead to psychological challenges. Consequently, moral imperatives and psychological well-being sometimes come into conflict in daily life. Moral education, grounded in the inevitable subjectivity required for individual development, seeks to resolve conflicts between the individual and the collective, as well as between individuals themselves, to ensure both continuity and harmonious development. The public nature of moral education inherently creates tension with the individual’s pursuit of freedom. Morality often manifests as a set of agreed-upon or enforced rules, representing a form of public power that individuals alone cannot easily resist. This dynamic raises important questions about the balance between personal autonomy and social conformity. From a relativistic perspective, social rules are time-bound and conditional, evolving with societal conditions. While the limited lifespan of individuals may prevent them from witnessing these changes firsthand, the formation and evolution of morality are long-term processes. Thus, the power and relevance of moral norms are inherently mutable, adapting to new circumstances and shifting societal values. The long-term evolution of morality underscores its adaptive nature. As societies change, so too do the moral frameworks that guide behavior. Understanding this mutability can empower individuals to navigate the complexities of modern life more effectively. By acknowledging that moral norms are not static but evolve over time, individuals can better reconcile their personal values with societal expectations. In conclusion, moral education must strike a delicate balance between instilling idealized moral principles and addressing the psychological realities of individual lives. By promoting self-awareness, critical thinking, dialogue, and personal growth, moral education can foster a more harmonious integration of personal and collective values. This approach supports the development of resilient, socially responsible individuals capable of navigating the ever-changing landscape of moral norms and societal expectations.

Harmonious development of psychological and moral education from a value perspective

The integration of psychological and moral education calls for the synthesis of two distinct yet complementary research perspectives, framed within the broader context of school education. From a disciplinary standpoint, both psychological and moral education serve critical roles in the educational system, fulfilling ontological and social functions. Both are educational disciplines that must adhere to core educational principles. However, they each have unique focuses and methodologies. Psychological education is a specialized field dedicated to enhancing psychological functions through an understanding of psychological processes and their development. It involves not only the dissemination of knowledge but also the transformation of psychological structures and functions. The challenge lies in utilizing “psychological content”—the information conveyed through education—to promote changes in “psychological form,” thereby optimizing this content to enhance and transform psychological functions. This approach emphasizes fostering self-awareness, emotional regulation, and cognitive skills that support mental health and well-being. Moral education, on the other hand, centers on instilling ethical values and guiding behavior through the study of moral norms and principles. It aims to cultivate virtues such as empathy, respect, and responsibility, which contribute to the formation of an ideal personality. Moral education seeks to resolve conflicts between individual desires and collective expectations, promoting harmonious coexistence within communities. While psychological and moral education share some commonalities in content and methods—such as promoting critical thinking, reflection, and dialogue—their primary objectives differ. Psychological education focuses on internal processes and personal growth, whereas moral education emphasizes external behaviors and societal norms. Integrating these two fields can create a new, unified discipline that leverages the strengths of both. By integrating the value interventions and orientations of moral education with the structural and functional changes in psychology, we can create a new, unified discipline of psychological-moral education. This discipline would focus on nurturing individuals who are not only psychologically resilient but also ethically grounded. It would emphasize the importance of aligning personal values with societal expectations, fostering a generation capable of navigating the complexities of modern life with integrity and compassion. In conclusion, the integration of psychological and moral education offers a promising avenue for enhancing educational practices. By synthesizing these two disciplines, we can develop more effective strategies for promoting holistic development, supporting mental health, and cultivating ethical behavior among students. This interdisciplinary approach holds the potential to equip future generations with the tools they need to thrive in an increasingly complex world.

To transcend the limitations of current research and more effectively integrate psychological and moral education, it is essential to consider the characteristics of their subjects at a higher level. Both fields deal with living individuals rather than natural phenomena, making them fundamentally humanistic sciences. All work in these areas revolves around different understandings of what it means to be human, with the question of “what is a human being?” serving as the logical starting point for the formation and development of these disciplines. Psychology, as an empirical science, focuses on factual issues, aiming to discover the truth about psychological phenomena and the laws governing their development. This pursuit is guided by epistemological principles, which seek to uncover objective truths through observation and experimentation. Psychology employs methodologies rooted in scientific inquiry, emphasizing data collection, analysis, and interpretation to understand cognitive, emotional, and behavioral processes. In contrast, moral education, while also addressing factual issues, primarily explores values—questions of right and wrong, good and bad—that cannot be analyzed using natural science methods. Instead, they require a value-philosophical methodology that considers ethical norms, virtues, and moral reasoning. Moral education seeks to cultivate virtues such as empathy, respect, and responsibility, fostering individuals who can navigate complex ethical dilemmas and contribute positively to society. The main distinction between psychological and moral education lies in their differing emphases on value issues. While psychology focuses on empirical facts and the mechanisms underlying psychological processes, moral education emphasizes the normative dimensions of human behavior. Integrating these two fields involves clarifying the value dimensions within psychological education, recognizing that the fusion of facts and values is an inevitable choice for the development of psychological and moral education. Constructing a theoretical and practical framework for psychological-moral education under this scientific and methodological perspective remains a significant challenge. Here are some strategies to address this challenge: Encourage collaboration between experts in psychology and ethics to develop comprehensive curricula that address both psychological and moral development. This interdisciplinary approach can help bridge the gap between empirical and value-based inquiries; Design educational programs that promote holistic development by integrating psychological well-being and moral character. Such programs should include activities that encourage self-reflection, ethical reasoning, and empathy-building; Adopt pedagogical approaches that facilitate the synthesis of psychological and moral education. For example, using case studies, role-playing, and group discussions can help students explore complex issues from multiple perspectives, enhancing their critical thinking skills; Provide ongoing training for educators to ensure they are equipped with the latest knowledge and skills in both psychological and moral education. This will enable them to guide students through their developmental journey effectively; Conduct empirical research to understand the intersection of psychological and moral development, while also engaging in philosophical inquiry to clarify the value dimensions inherent in this integration. This dual approach can provide a robust foundation for developing effective educational practices. The integration of psychological and moral education offers a promising avenue for enhancing educational practices. By synthesizing these two disciplines, we can develop more effective strategies for promoting holistic development, supporting mental health, and cultivating ethical behavior among students. This interdisciplinary approach holds the potential to equip future generations with the tools they need to thrive in an increasingly complex world. Constructing a unified theoretical and practical framework for psychological-moral education remains a challenging but crucial endeavor that requires further exploration and innovation.

Algorithm of deep learning neural network basic model

This paper delves into a specialized branch of machine learning that encompasses several nonlinear transformation structures, particularly focusing on convolutional neural networks (CNNs) and motor neural networks (MNNs). The study leverages insights derived from the analysis of human brain distribution information to simulate and investigate human data processing mechanisms. Despite the diverse frameworks currently employed in this domain, these techniques have been widely adopted across various fields including pattern recognition, data analysis, and machine translation. In this section, we explore the models and fundamental theories underlying CNNs and MNNs, emphasizing their application in audio systems. We analyze the distinct advantages each model offers in predicting time series data, such as speech or music signals. By comparing these methodologies, we aim to highlight how they complement one another in enhancing predictive accuracy and robustness. To further enhance the performance of neural network models, we introduce optimization techniques relevant to both CNNs and MNNs. These strategies include advanced regularization methods, dynamic learning rate adjustments, and novel activation functions designed to mitigate common issues such as overfitting and vanishing gradients. Implementing these optimizations lays a solid theoretical foundation for subsequent research and practical applications. This paper provides a comprehensive analysis of the strengths of different neural network architectures in the context of time series prediction, especially focusing on audio systems. Through detailed comparisons and the introduction of advanced optimization techniques, we establish a strong theoretical framework for future investigations. This foundational work not only underscores the importance of selecting appropriate models but also highlights the necessity of employing effective optimization strategies to achieve superior performance in complex data-driven tasks. The formula for max-pooling is as follows:

Deep learning causal convolution significantly differs from traditional full convolution networks by enforcing strict adherence to temporal causality. This means that causal convolutions are designed to operate in such a way that the computation of an output at any given time point relies exclusively on the historical and current data points, without revealing or utilizing subsequent information. This approach is particularly beneficial for tasks involving time series analysis where maintaining the integrity of temporal sequences is critical.Causal convolution networks are uniquely suited for applications requiring strict temporal causality, such as time series forecasting, speech recognition, and natural language processing. By ensuring that outputs at any time step t depend only on the data available up to that point, these networks prevent the leakage of future information, thereby preserving the chronological integrity of the data. This makes them invaluable tools for accurately modeling and predicting sequential data across various domains. The output value formula of a causal convolution network everywhere is:

Deep learning algorithm is applied to deal with timing problems. The traditional neural network in deep learning has no full connection between neurons in each layer, so it can’t deal with the time context of time series data. Deep learning cyclic neural network is different from traditional neural network. Each layer of RNN shares weight parameters, that is, matrices U, V and W are the same, which can reduce the parameters of the model. The inverse process includes gradient back propagation algorithm across time, which aims to update and iterate parameters such as weights and offsets in the model. The specific calculation formula is:

When deep learning cyclic neural network predicts time series, the gradient of subsequent nodes of back propagation algorithm will gradually decrease with the increase of time steps, so it is difficult to update early nodes, and the learning speed is very slow, resulting in the problem of “gradient disappearance”. RNN is difficult to store earlier historical information, which changes the hidden layer structure of traditional RNN and can solve the gradient disappearance problem well. In traditional RNN, the neural unit may be a simple tanh function. In order to solve these problems, LSTM adds a unit state to preserve the long-term memory state, which can greatly improve the prediction accuracy. The gated loop unit was proposed in 2014 to improve the structure based on LSTM, which reduces the number of gates and has a simple structure. It is widely used in sequence data processing and reduces the risk of over-fitting. In GRU network, the function of “gate” is to control the cell state at each time point. Like RNN, GRU model is a chain structure composed of multiple neural units. The neural units in GRU network are complex, and the calculation formula of GRU neural unit structure is as follows:

In deep learning, attention mechanism pays more attention to important information when using limited resources to process data. Different weights are assigned to each area of information to make it pay more attention to useful information. Attention mechanism is a mathematical transformation process such as similarity calculation and normalization. The calculation formula is:

After the weight coefficient is calculated by attention mechanism, the larger the value, the more important the information is, which can effectively screen useless information and is suitable for data with strong nonlinear relationship. The generation of deep learning countermeasure neural network can optimize two objectives at the same time, and the optimization objective function expression is:

Deep learning optimization function

The activation function of deep learning neural network plays an important role in the learning of network model, which adds nonlinear factors to neural network and makes neurons output nonlinear results. For a neuron node in a hidden layer, the calculation process of its activation function is generally as follows: assuming that the input value of the node first undergoes linear transformation after entering the hidden layer node, the formula is:

Then, the nonlinear transformation of deep learning is carried out to calculate the output value, and the calculation formula is:

The above subexpression is the activation function of deep learning. The common activation function of deep learning is suitable for input values with small feature difference. The formula is:

Tanh function in deep learning is more common than sigmoid function. The output value mapping range of tanh function is (-1, 1), and it can be regarded as a linear relationship when the input value is near 0. But tanh function solves the drawback of non-zero mean value of sigmoid function, and has better application effect than sigmoid function. Its calculation formula is:

The Re LU activation function of deep learning is between the output range (0, ∞). When the input value is greater than 0, the derivative of the output value is always 1, so the long-term dependence problem is alleviated to some extent. In addition, Re-LU function can make some neurons activation value to 0, which reduces over-fitting. The application range expression of its Re LU activation function is:

In describing deep learning algorithms, common loss functions can directly minimize the distance and maximize the direct distance of mismatched samples. Where the distance expression is:

The depth of the project loss did not match the matching distance of the samples, the quality supervisor said. When the sample is 0, this leads to over-matching of deep training. Relaxation matching value can make the matching value of samples greater than 0, so as to obtain better matching effect, but the main problem is that the loss value of samples is difficult to identify. A group of three samples is composed of matching distance. The definition formula of loss is:

The first section of the deep learning loss function is consistent with the loss, and the second section introduces a new boundary to consider the change of class. The increased restriction makes the minimum distance between classes greater than the maximum internal distance between classes. Loss introduces negative samples into the loss function and pushes these negative samples from their respective samples to update. The formula is as follows:

Research design scheme

Participants and samples are shown in Table 1:

The design process of deep learning experiment is shown in Fig. 2:

Deep learning experiment design and comparison process

The key elements of the controlled experimental design were compared as follows:

Baseline comparison group: traditional machine learning methods (such as SVM and random forest) are used to process the same features to verify the advantages of deep learning.

Ablation experiment: gradually remove physiological signals, psychological characteristics and other inputs, and analyze the contribution of each modality to the performance of the model.

Clinical control: Compare the changes of SCL-90 scores between the deep learning intervention group and the traditional psychological education group after 3 months to verify the educational effect.

Analysis of psychological education characteristics using deep learning

In recent years, there has been increasing attention towards integrating psychological education with deep learning techniques. As an advanced algorithmic approach, deep learning excels at extracting meaningful patterns from vast datasets, thereby making it exceptionally suitable for applications in both psychological education and information processing. The rapid development of educational technology in China has facilitated the deployment of deep learning across diverse media platforms, including film, video, news, and music. The enhancement of psychological education through deep learning primarily focuses on two aspects: first, imparting cultural knowledge grounded in widely accepted moral standards; and second, developing innovative higher education materials. Empirical studies have demonstrated that deep learning models achieve superior accuracy and performance in recognizing various psychological traits. Specifically, an enhanced deep learning methodology achieved 98.4% accuracy in assessing psychological expressiveness, surpassing traditional teaching approaches. Research indicates that deep learning offers distinct advantages over alternative methods such as transfer learning, machine learning, and shallow learning, especially concerning the analysis of psychological features associated with moral dilemmas and the resolution of moral conflicts. These benefits are particularly pronounced in the context of psychological education. A comparative analysis of deep learning’s performance relative to other methodologies is provided in Fig. 3.

Performance Comparison of Deep Learning and Other Methods in Analyzing Moral Dilemmas

In the domain of psychological feature recognition, we evaluate the performance of advanced deep learning algorithms, particularly Convolutional Neural Networks (CNNs), against that of traditional neural networks. CNNs are especially proficient at recognizing and extracting features from structured data such as vectors, making them highly effective for tasks involving complex pattern recognition. By integrating Deep Belief Network (DBN) feature extraction techniques, the efficacy of feature extraction can be significantly enhanced. DBNs, which consist of multiple layers of stochastic, latent variables, offer an unsupervised approach to pre-training that can improve the initialization of weights in subsequent supervised learning stages. This combination leverages the strengths of both architectures: the robust feature extraction capabilities of CNNs and the deep hierarchical representation learning of DBNs. The synergy between CNNs and DBNs allows for more accurate and nuanced feature extraction, leading to improved performance in psychological feature recognition. Specifically, this hybrid approach has demonstrated superior results compared to traditional neural networks, which often struggle with capturing intricate patterns within high-dimensional data. The significance of deep learning algorithm comparison is mainly reflected in the following aspects:

Performance Verification: By comparing the performance of deep learning with transfer learning, machine learning and other methods in the field of music composition (such as 98.4% recognition accuracy), the superiority of deep learning in feature extraction and pattern recognition is confirmed³⁵. This quantitative comparison provides an empirical basis for technical selection in the field of psychological education.

Technical Adaptability: The combination of CNN and DBN verifies the feasibility of processing multimodal physiological signals (electrocardiogram/skin electrical signals), indicating that deep learning can effectively integrate cross-modal psychological data, which is superior to the ability of traditional neural networks to process single signals.

Application Innovation: Algorithm comparison reveals the unique advantages of deep learning in creative tasks (such as music generation), which not only provides analytical tools for psychological education, but also expands the possibilities of intervention methods such as art therapy.

After preprocessing the psychological feature dataset, physiological signal data, including electrocardiogram (ECG) signals, skin conductance, and 3D signal models, are utilized as inputs to the Convolutional Neural Network (CNN). The preprocessing step ensures that the data is in an optimal format for subsequent analysis, enhancing the effectiveness of feature extraction. The CNN architecture employed consists of four convolutional layers designed to extract key physiological features from the input data. These extracted features are then processed through two fully connected layers before being passed to a classifier for final output and classification. The system leverages a multimodal physiological signal dataset, integrating various types of physiological data. This approach allows for a comprehensive representation of the subjects’ responses, which is crucial for accurately analyzing complex phenomena such as moral dilemmas. Two distinct neural data processing methods are applied to the preprocessed dataset. These methods enhance the robustness of the feature extraction process, ensuring that the model can effectively capture intricate patterns within the data. The experimental setup employs a 4-layer CNN model with the following configuration:

Fully Connected Layers: Two layers responsible for learning and expressing the extracted features.

Activation Function: The Exponential Linear Unit (ELU) is used as the activation function across the network to introduce non-linearity, facilitating more effective learning. The combination of convolutional and fully connected layers enables the model to handle both feature extraction and learning tasks efficiently. The parameters and details of the CNN model configured for the analysis of moral dilemmas are summarized in Table 2.

The model adopts a typical CNN architecture combined with fully connected layers and a softmax classifier, specifically designed for feature extraction in music composition. The overall structure exhibits the following characteristics:

Hierarchical Distribution: A total of 11 layers, including 1 input layer, 4 convolutional layers, 4 pooling layers, 1 fully connected layer, and 1 softmax output layer.

Feature Processing Path: Input data → convolutional feature extraction → pooling for dimensionality reduction → multi-layer feature abstraction → fully connected integration → classification output.

Music Composition Focus: Multi-layer convolutional kernel design captures the spatiotemporal features of musical signals (such as melodic timing and harmonic dimensions), and pooling operations adapt to the long-term dependency characteristics of music data.

An analysis of the psychological characteristics of music creation

When confronted with large volumes of sample data, relying solely on a uniform machine learning approach proves inadequate for enhancing learning efficiency. Human feature extraction, which heavily depends on prior experience, necessitates more sophisticated methodologies to fully capture the nuances within complex datasets. In response to these challenges, scholars are increasingly focusing on developing educational models through in-depth research methods. Specifically, there is a notable shift from speculative and philosophical approaches to empirical and psychological ones in the study of moral dilemmas. This transition underscores the need for integrating psychological findings into the analysis of moral issues, thereby enriching our understanding of how individuals navigate ethical quandaries. At the core of this research lies the integration of moral dilemmas with psychological features. These features, characterized by modern experimental science, fall under the natural sciences rather than the humanities or social sciences. Nevertheless, certain branches of psychology adeptly employ experimental and quantitative methods to elucidate complex and subtle moral psychological activities. Consequently, empirical analysis remains a cornerstone in the study of moral dilemmas, providing valuable insights into related phenomena. By leveraging empirical analysis, researchers can delve deeper into the psychological mechanisms underlying moral decision-making. This method not only offers a systematic way to explore intricate moral psychological activities but also provides a robust framework for addressing related issues. Therefore, revisiting and expanding upon previous work within the context of psychological feature analysis is essential. This reorganization allows for a more comprehensive examination of the subject matter, facilitating a richer understanding of the interplay between moral dilemmas and psychological features. In summary, the move towards empirical and psychological approaches in studying moral dilemmas represents a significant advancement in the field. By incorporating psychological findings and utilizing empirical analysis, researchers can gain deeper insights into the complexities of moral decision-making. This approach not only enhances the rigor of moral dilemma studies but also paves the way for more effective educational models that can adapt to the evolving landscape of ethical inquiry.

In our analysis of moral dilemmas, we utilize the Symptom Checklist-90 (SCL-90) as a survey tool. The SCL-90 assesses various aspects of mental health through 10 factors: somatization, obsessive-compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, and paranoid ideation, psychoticism, and sleep disturbances. The scale consists of 90 items, each rated on a “none, mild, moderate, severe” scale. Higher individual scores indicate a more detailed description of mental health factors and symptoms, suggesting poorer mental health. The scoring and interpretation of the SCL-90 are detailed in Table 3.

The total distribution and severity of the scores across the 10 factors of the SCL-90 scale in our study on moral conflicts were generally consistent, with mean values ranging from 1.23 to 1.66. These scores indicate that participants’ psychological distress levels had not yet reached moderate severity. Initially, we utilized the SCL-90 scale to examine the psychological characteristics of each group involved in moral conflicts. Subsequently, we analyzed the psychological features associated with these conflicts by focusing on the 10 dimensions assessed by the SCL-90. The following table (Table 4) illustrates the distribution and severity of the scores for each of the 10 factors among individuals experiencing moral conflicts:

Integration analysis of moral dilemmas and psychological features

Our analysis reveals that the majority of individuals experiencing moral dilemmas exhibit either no symptoms or only mild symptoms across the 10 psychological factors measured by the SCL-90 scale. Only a small number of participants reported severe scores in these dimensions. To better visualize the distribution of symptom severity within this population, we utilized a radar chart to represent the proportion of individuals with varying levels of psychological distress (no symptoms, mild, moderate, and severe). The radar chart clearly illustrates that the proportion of individuals with no symptoms and mild symptoms is significantly higher than those with moderate and severe symptoms. Specifically, the proportion of individuals with moderate symptoms ranges between 10% and 30%, representing a minority of the overall population. The distribution of the severity of psychological features in moral dilemmas is illustrated in Fig. 4.

Comparative Analysis of the Severity of Psychological Feature Factors in Moral Dilemmas

Analysis of the impact of moral dilemmas on psychological education

Recent advancements have seen the application of deep learning techniques to analyze emotional expressions in the context of moral dilemmas. This study reviews current applications of deep learning in understanding emotional expression within psychological education. Emotional expression, as used here, refers to the emotions experienced by individuals during moral dilemmas. An emotional scale was employed to assess these expressions, and a one-sample t-test was conducted to compare the mean scores of individual items against the scale’s average score of 3. The results indicated that the total score for emotional expression flexibility (t = 29.56, p < 0.001) and the mean scores of individual items across its two dimensions were significantly higher than the scale’s average, indicating a high level of emotional expression flexibility among participants. Additionally, self-esteem, anxiety, and depression were assessed using scales scored on a 4-point Likert scale, with a theoretical mean of 2.5 per item. The findings are summarized below: The mean score was 2.70 (t = 12.81, p < 0.001), significantly higher than the theoretical mean, suggesting elevated levels of self-esteem. The mean score was 2.00, significantly lower than the theoretical mean (p < 0.001). The mean score was 2.16, also significantly lower than the theoretical mean (p < 0.001). These results are summarized in Table 4 and provide insights into the emotional states and psychological well-being of individuals experiencing moral dilemmas. The significant differences observed highlight the importance of considering emotional flexibility and psychological factors when examining responses to moral dilemmas. Please ensure that Table 4 is accurately labeled and includes all relevant data points discussed. If you need assistance creating or refining this table, feel free to let me know.

Table 5 presents the descriptive statistics (mean M, standard deviation SD, single-item mean) and t-test results for 8 psychological and emotional variables, reflecting the psychological characteristics of the music composition group. The data show that variables related to emotional expression generally have high scores, while negative psychological indicators such as anxiety and depression show moderately high levels, and life satisfaction scores are low with non-significant t-values. The larger the absolute value of the t-value (e.g., anxiety t=-32.93), the more significant the difference from the norm or control group.

Emotional Expression Flexibility: Scores were significantly higher than the norm (t = 29.56), indicating that creators have strong ability to switch emotional expressions in music. The single-item mean was 3.58 (moderately high), but the standard deviation was large (7.85), suggesting obvious individual differences.

Emotional Expression Inhibition: Inhibition refers to the ability to control emotional expression, with high scores and significant t-values, possibly reflecting creators’ regulatory capacity between rationality and sensibility. The standard deviation was small (4.16), indicating consistent group-level performance.

Emotional Expression Catharsis: Catharsis ability had the highest score (single-item mean 3.74) and an extremely significant t-value (33.16), showing that creators excel in releasing emotions through music—possibly related to the emotional counseling function of music composition.

Self-Esteem: Self-esteem levels were moderate (single-item mean 2.71) with significant t-values, slightly higher than the general population. The standard deviation was 4.85, indicating large fluctuations in self-esteem among some creators.

Anxiety: Anxiety scores were significantly lower than the norm (negative t-value), but the single-item mean was 2.07 (close to the “mild anxiety” critical value of 2.5). The standard deviation was 9.42, indicating that high-anxiety individuals exist in the group (e.g., scores exceeding 39.91 + 9.42 = 49.33).

Depression: Depression levels were similar to anxiety, with a mean close to “mild depression” (2.16), but t-values showed they were lower than the norm. Caution is needed for individuals with high depression scores (e.g., 43.28 + 8.73 = 52.01), who may be at emotional risk.

Life Satisfaction: The single-item mean was 3.96 (out of a possible 5), seemingly high, but the t-value was non-significant (absolute value < 1.96), indicating no significant difference from the norm. The standard deviation was 6.61, suggesting low life satisfaction among some creators.

The statistical results reveal the advantages of music composition groups in emotional expression abilities and the critical status of psychological indicators such as anxiety and depression. The data provide preliminary evidence for “music composition as a psychological regulation tool,” but more rigorous research designs (e.g., longitudinal tracking, control group matching) are needed for further validation. Future research could focus on exploring the interaction between the three dimensions of emotional expression and psychological characteristics to provide quantitative basis for music-based psychological education interventions.

Analysis of the impact of moral dilemmas on psychological education

Emotional expression within the context of moral dilemmas can manifest through various mediums, including vocal, facial, and bodily expressions. Among these, the voice serves as a primary channel for conveying an individual’s thoughts and feelings. Emotions are deeply intertwined with personal thoughts and are expressed through individual emotional responses, with both aspects mutually reinforcing one another. This interplay enhances the effectiveness of emotional and psychological feature expression, particularly in moral dilemmas, which are often driven by underlying emotions. These emotions not only influence behavior but also manifest prominently in work, learning, and other contexts. For example, an individual might adopt a cheerful and smooth tone when experiencing happiness or use a more intense and exaggerated vocal style when expressing anger. To systematically investigate the nature of emotional expressions in moral dilemmas compared to those in everyday life, we conducted a survey using a randomly selected sample to ensure the accuracy and representativeness of our findings. The analysis of emotional expressions in moral dilemmas, relative to their manifestation in work and learning contexts, is illustrated in Fig. 5. This figure highlights the distinct patterns and intensities of emotional expressions across different scenarios, providing valuable insights into how emotions shape and are shaped by moral decision-making processes.

Comparison of Emotional Expression in Moral Dilemmas and Work/Learning Contexts

The comparison between emotional expressions in work and learning contexts versus moral dilemmas, as illustrated in Fig. 4, reveals significant differences in how emotions are expressed and experienced. In moral dilemmas, individuals often find a more conducive environment for the direct and unfiltered expression of their emotions. This setting allows for a cathartic release, enabling deeper emotional processing and understanding. In contrast, work and learning contexts tend to offer visual and experiential cues that indirectly reflect the inner emotions of individuals. These environments typically impose certain behavioral norms and expectations, which can constrain spontaneous emotional displays. Consequently, while emotions may still be present and influential, they are often expressed through subtler, non-verbal means such as body language or facial expressions.　To further strengthen your analysis, consider integrating specific examples or case studies that demonstrate these differences. Additionally, ensure that Fig. 4 clearly illustrates the distinctions you describe, using appropriate labels and annotations to highlight key findings.

Emotional expression in moral dilemmas can be categorized into three forms: verbal, facial, and bodily expressions. Compared to traditional methods, deep learning offers significant analytical advantages in the study of these expressions. To evaluate the performance of deep learning in analyzing the degree of emotional expression in moral dilemmas, we conducted a survey using a random sample of individuals. The data showed that deep learning has a relatively high performance in this context. The survey also revealed variations in the types of emotional expressions. The results are illustrated in Fig. 6.

Analysis of the influence of moral conflict on emotional expression

Authors and Affiliations

1. School of philosophy, Shanxi University, Taiyuan, 030000, China

   XiaoFen Jia & WenQing Wu

Contributions

XiaoFen Jia: Conceptualization, Methodology, Software, investigation, Formal Analysis, Writing - Original Draft, Visualization; WenQing Wu: Conceptualization, Resources, Supervision, Writing - Review & Editing, Investigation, Data Curation.

Corresponding author

Correspondence to XiaoFen Jia.

Ethics approval and consent to participate

All procedures performed in studies involving human participants were in accordance with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This study was approved by the Research Ethics Review Committee of Shanxi University. This study obtained informed consent from all participants.

Consent for publication

No.

Competing interests

The authors declare no competing interests.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions