According to SDT, humans have three basic psychological needs: autonomy (experiencing one’s actions as volitional and self-endorsed), competence (feeling effective and capable), and relatedness (feeling connected and cared for in important relationships) (Deci and Ryan, 2000; Ryan and Deci, 2000, 2020; Rosli et al., 2022). When these needs are satisfied, learners tend to display autonomous motivation, deeper engagement, and better psychological functioning (Müller et al., 2021; Shelton-Strong, 2022). A systematic review by Hu and Xiao (2025) found that fulfilling students’ autonomy, competence, and relatedness needs significantly boosts their engagement in online courses: learners who feel in control of their learning (autonomy), capable of mastering the material (competence), and connected to instructors or peers (relatedness) tend to participate more actively and persist in e-learning (Hu and Xiao, 2025).

However, besides need satisfaction, other internal motives (e.g., curiosity, interest) and external incentives both play a role in online learner engagement, underscoring the importance of addressing why students learn online (their regulated motives) in addition to how their needs are met (Hu and Xiao, 2025). Rosli et al. (2022) reviewed SDT studies in university online learning and noted that online learners may also be driven by identified goals or external rewards, or experience amotivation if needs are thwarted (Rosli et al., 2022). In a comprehensive review of Internet-supported learning, Bekele (2010) found that technology attributes (e.g., user-friendly platforms, reliable tools) and support services (technical help, tutoring) were consistently linked to higher student motivation and satisfaction (Bekele, 2010). A literature review by Nortvig et al. (2018) found that instructor presence and rich interactions (student–instructor, student–student, and student–content interaction) are dominant factors affecting learning outcomes and student satisfaction in e-learning (Nortvig et al., 2018). Therefore, previous studies suggest that an online course that provides timely instructor feedback, peer discussion opportunities, and engaging content interactions is more likely to satisfy students’ needs and keep them motivated: such interactions can fulfill relatedness (through teacher and peer support) and even competence (through interactive content and feedback).

Conversely, the frustration of basic psychological needs of autonomy, competence, and relatedness, and feeling controlled, incompetent, or rejected, constitutes a qualitatively negative experience associated with ill-being, disengagement, and maladaptive outcomes (De Naeghel et al., 2012; Kiltz et al., 2024; Krijgsman et al., 2025; Liu et al., 2024; Vansteenkiste et al., 2004). Recent work has emphasized that need frustration is not simply the absence of need satisfaction, but a distinct construct with its own nomological network. High levels of autonomy frustration (e.g., feeling coerced or controlled), competence frustration (e.g., feeling like a failure), and relatedness frustration (e.g., feeling rejected) are associated with controlled forms of motivation, defensive patterns of self-regulation, a heightened risk for depression and anxiety, a tendency to adopt extrinsic goals (such as grade-focused striving) rather than more self-endorsed, mastery-oriented goals (Krijgsman et al., 2025).

Thus, need frustration is not merely the absence of need satisfaction, but a distinct experience with unique negative consequences (Chen et al., 2015; Liu et al., 2024). High autonomy frustration (e.g., feeling pressured), competence frustration (feeling incapable), and relatedness frustration (feeling excluded) have been linked to decline in motivation and self-efficacy (Janke, 2022; Müller et al., 2021; Wang and Tsai, 2020), and increased risk for psychopathology (Liu et al., 2024; Tindall and Curtis, 2019; van der Kaap-Deeder, 2023; Van der Kaap-Deeder et al., 2020). Moreover, prolonged need frustration can give rise to controlled motivation (e.g., external pressure or obligations) and even amotivation and is also associated with lower engagement or higher anxiety (Krijgsman et al., 2025). E-learning can easily become a breeding ground for need frustration when it is poorly designed or unsupported: for example, a controlling online course that offers no choices can frustrate autonomy; poorly designed tasks that lead to repeated failures can frustrate competence; and isolation in a virtual class can frustrate relatedness (Bowers and Kumar, 2015; Kiltz et al., 2024; Piccoli et al., 2001).

Comparisons of online and face-to-face learning suggest that online learners often report higher levels of basic need frustration and lower need satisfaction (Müller et al., 2021; Wang et al., 2019; Wong, 2020). For instance, Wang et al. (2019) found that learners in online courses reported greater frustration of all three needs relative to those in face-to-face classes, likely reflecting reduced instructor support and peer interaction. Need frustration is not just unpleasant; it predicts lower intrinsic motivation, reduced vitality, and higher intentions to drop out (Müller et al., 2021; Rovai and Wighting, 2005).

At the same time, online learning can support needs when educational and technological design are thoughtful, and systematic reviews emphasize that instructor presence, meaningful interaction, and well-designed balance between online and offline activities are key to student satisfaction and motivation (Bekele, 2010; Fuentes and Labad, 2025; Nortvig et al., 2018). Furthermore, need-supportive practices, including autonomy support, competence-supportive feedback, and social relatedness, reliably predict engagement and persistence (Hu and Xiao, 2025). Thus, online environments can either thwart or support needs, depending on how they are designed and enacted.

However, research shows there is heterogeneity in need frustration experiences. Rather than assuming all online students suffer equally, recent studies have used Latent Profile Analysis (LPA) to uncover subpopulations with different need satisfaction/frustration patterns. For example, Krijgsman et al. (2025) identified multiple need profiles among students, including groups with highly fulfilled needs and others with unfulfilled or frustrated needs, each associated with distinct motivational outcomessciencedirect.comsciencedirect.com. However, such person-centered analyses have mostly focused on face-to-face education settings (Krijgsman et al., 2025) and less so on fully online learning contexts. There remains a gap in understanding need frustration profiles in e-learning: do some online learners experience universally high frustration across all needs, while others feel relatively little frustration? How prevalent are these profiles, and what characterizes the learners within them? Addressing this gap, our first aim is to conduct an LPA on e-learners’ autonomy, competence, and relatedness frustration to determine whether distinct profiles (e.g., “high frustration” vs. “low frustration” groups) exist in an online learning sample. Therefore, it can be assumed that at least two distinct profiles (“high frustration” vs. “low frustration” groups) would emerge in an online learning sample.

EVT proposes that learners’ achievement behaviors are jointly determined by their expectancy beliefs (e.g., self-efficacy) and the value they place on tasks (Eccles and Wigfield, 2002, 2023; Wigfield and Eccles, 2000). Learners are more likely to engage and persist when they believe they can succeed and see tasks as valuable, interesting, or important (Lee and Song, 2022; Wang and Xue, 2022). In other words, EVT suggests that students’ achievement behaviors are driven by their expectancy of success (self-efficacy beliefs about their capabilities) and the value they attach to the task. Therefore, key constructs include self-efficacy, which can include confidence in one’s ability to learn or perform well, to master online course demands, and task value, which can encompass the perceived importance, usefulness, or interest of the learning task, course content and activities. Previous studies found that students who strongly believe they can succeed (high expectancy) and who find the learning task valuable or interesting (high value) are more likely to engage persistently and perform better (Eccles and Wigfield, 2020, 2023; Lee and Song, 2022; Rachmatullah et al., 2021; Wang and Xue, 2022; Wigfield and Eccles, 2000). Yet online environments can undermine these beliefs: reduced feedback, ambiguous expectations, and technological hurdles may erode efficacy and perceived value over time (Xu, 2022). The absence of immediate feedback or richer communication cues may undermine students’ sense of efficacy (if they are unsure how to gauge their progress) and may diminish perceived task value or interest over time (Hartnett, 2015; Hartnett et al., 2023). At the same time, there is considerable individual variability: some e-learners maintain high self-efficacy and value for their online coursework, whereas others struggle (Hartnett et al., 2023; Noour and Hubbard, 2015). Theoretically, learners with high self-efficacy and interest might better cope with challenges, potentially buffering against subjectively perceived incompetence or disconnection (i.e., less need frustration), and learners with low confidence or low task value might be more prone to frustration, as they struggle to see progress or purpose in the online tasks, but these questions still are under-researched.

Goal orientation, especially extrinsic goal orientation, is another motivational dimension linked to EVT and SDT. Extrinsic goal orientation refers to the extent to which learners engage in learning for external reasons, primarily to obtain external outcomes (grades, rewards, approval, recognition) rather than for inherent interest or mastery (Patrick et al., 1999; Wolters et al., 1996; Zhou and Zhang, 2024). In the context of EVT, an extrinsically oriented student may value a task mainly as a means to an end (e.g., a high grade), not for the task itself. This construct also intersects with SDT: extrinsic goal orientation can be seen as a form of controlled motivation, often associated with external regulation or introjected pressure that tend to undermine autonomy satisfaction (Ryan and Deci, 2000; Zhou and Zhang, 2024). Students with more intrinsic or mastery-oriented goals (focusing on learning and personal growth) might better fulfill their needs even online, as they engage out of genuine interest and seek understanding. In contrast, extrinsic goal orientation is related to surface learning strategies, anxiety, and lower engagement: e-learners who are strongly extrinsically oriented could experience greater frustration of autonomy and competence, for example, feeling pressured to achieve (undermining autonomy) and fearing failure (undermining competence), especially in an online setting where they have less guidance (Abdolrezapour et al., 2023; Wolters et al., 1996; Xu, 2022; Zhou and Zhang, 2024).

Contemporary views of learning emphasize online learners’ active role in directing and co-constructing knowledge (Bayly-Castaneda et al., 2024; Chou and Liu, 2005; Forbes et al., 2023; Huang and Wang, 2023; Salikhova et al., 2020; Schneider, 2024). Self-directed learning involves setting one’s own goals, monitoring progress, and regulating strategies (Geng et al., 2019; Karatas and Arpaci, 2021; Rini et al., 2022; Tekkol and Demirel, 2018; Timothy et al., 2010). Collaborative learning involves working with peers to share perspectives and solve problems, and in ICT-rich environments, these processes may occur via digital tools (e.g., forums, collaborative documents, video conferencing) (Loes, 2022; Makransky and Petersen, 2023; Qureshi et al., 2023; Yang, 2023). Engaging in self-directed and collaborative learning may help learners constructively navigate online tasks, potentially buffering against need frustration (Geng et al., 2019; Karatas and Arpaci, 2021; Wang et al., 2023). However, despite research on intrinsic and extrinsic motivation in online education (Zhou and Zhang, 2024), it is unclear how task value and goal orientation are related to basic psychological needs. Based on previous research, it can be presumed that basic needs’ frustration is related to lower task value and self-efficacy, higher extrinsic goal orientation, lower self-directed and collaborative learning tendencies.

E-learning environments overrun learners with potential distractions: multiple browser tabs, notifications, and home-based interruptions. In this context, attentional control is an important self-regulatory resource (Derryberry and Reed, 2002; Eysenck et al., 2007; Eysenck and Derakshan, 2011; Green and Bavelier, 2012; Blanco et al., 2023). In addition to motivational beliefs and goals, attentional control is a self-regulatory capacity to focus and sustain attention on relevant tasks and to shift attention flexibly that may impact e-learning experiences (Clement and Anderson, 2023; Green and Bavelier, 2012). Attentional control refers to an individual’s ability to focus and sustain attention on relevant tasks and to shift attention effectively when needed, and in a traditional classroom, external structure and cues from instructors can help students maintain attention. Online, however, learners often face a multitude of distractions (e.g., other browser tabs, notifications, home environment disturbances) and must exert greater conscious control to stay focused (Bavelier and Green, 2019; Khasawneh et al., 2025). Learners with strong attentional control can maintain focus, resist distractions, and keep up with online demands, and those with poor attentional control are more prone to mind-wandering, multitasking, and procrastination (Blanco et al., 2023; Daniel et al., 2020; du Rocher, 2020; Filičková et al., 2015; Green and Bavelier, 2012; İnal et al., 2023; Meinhardt and Grabbe, 2002; Turoman et al., 2021).

Presumably, attentional control has an indirect but significant relationship with need frustration: low attentional control could intensify feelings of incompetence (as the learner struggles to absorb content or finish tasks, undermining competence need) and even relatedness (if distraction leads to missed communications, deadlines or a sense of disconnection). Additionally, an unfocused learner might feel less autonomy, as they cannot manage their learning time effectively and feel controlled by distractions or deadlines. Empirical evidence in online education suggests that students who report difficulty concentrating indeed have lower satisfaction and higher stress in virtual courses (Staller et al., 2021; Wong, 2020; Blanco et al., 2023). Conversely, those with good attentional regulation skills tend to achieve better learning outcomes online (Hodis et al., 2023) and may navigate the online environment with less frustration (Bavelier and Green, 2019; Eysenck et al., 2007; Green and Bavelier, 2012; Henriksen et al., 2023). Therefore, it can be assumed that difficulties in attentional control are linked to basic psychological needs frustration.

Based on SDT and prior research (Krijgsman et al., 2025; Xu, 2022), we presumed (H1) there are at least two groups of online learners: a “High Need Frustration” profile and a “Low Need Frustration” profile. We also hypothesized that (H2) basic needs’ frustration is related to lower task value and self-efficacy, higher extrinsic goal orientation, lower self-directed and collaborative learning tendencies, and that (H3) difficulties in attentional control are linked to basic psychological needs frustration. In the binominal logistic regression model, we expected extrinsic goal orientation and attentional control to emerge as especially strong predictors of profile membership.

The present study integrates SDT and EVT perspectives with the purpose to identify latent profiles of basic psychological need frustration (autonomy, competence, relatedness) in online learning contexts using latent profile analysis and to examine which motivational beliefs and self-regulatory variables, including task value, self-efficacy, extrinsic goal orientation, self-directed learning (with and without ICT), collaborative learning (with and without ICT), and attentional control, predict profile membership when considered simultaneously in a binomial logistic regression model. The overall aim of the study is to contribute to better understanding of online learners’ psychological need frustration in learning settings, by integrating the self-determination theory, expectancy - value theory, contemporary perspectives on self-directed and collaborative learning, and research on attentional control into a unified framework.

Participants were university learners enrolled in online or blended courses at large higher education institutions in Vilnius, Lithuania. Initially, 749 learners completed at least part of the survey. After screening careless and invalid responses (e.g., excessive missing data, invariant responding), cases were removed, resulting in an analytic sample of n = 541. Age of the learners ranged from 18 to 64 years (M = 26.37, SD = 8.61), with a positively skewed distribution (skewness = 1.19, kurtosis = 0.72), reflecting a predominance of younger adults but with some older learners as well. Students were drawn from a variety of academic majors and were taking at least one course delivered fully online at the time of the study. The ethnic composition was not formally recorded, but the sample was predominantly European (Baltic) with no representation of other groups.

The study was approved by the Institutional Review Board of Klaipeda University (Approval No. STIMC-BMTEK-P03, April 12, 2021). Data collection ended in November 2024. Participants were recruited via announcements in online class forums and email invitations and voluntarily provided their consent. Following informed consent, interested learners were directed to a web-based survey. To encourage honest responding, participants were assured that their answers were confidential, that participation was voluntary, and that data would be used only for research. Questionnaire sections were presented in randomized order (except for demographics, which appeared last) to reduce order effects.

Basic psychological need frustration, motivated approaches to learning, and attentional control were assessed with established self-report questionnaires.

Need frustration was assessed using the BPNSFS (Basic Psychological Need Satisfaction and Frustration Scale (BPNSNF), 2022; Chen et al., 2015), developed within self-determination theory. We utilized the frustration subscales of the BPNSFS, which contain 4 items per need (12 items total) that capture experiences of need thwarting: autonomy frustration—experiencing pressure, coercion, or a sense of being forced to behave in certain ways; competence frustration—feeling ineffective, incapable, or overwhelmed by demands; relatedness frustration—feeling rejected, excluded, or disconnected from significant others. Items are answered on a 5-point agreement scale (1–5), with higher scores indicating stronger agreement with the statement, or greater frustration of the respective need. Previous research reported good internal consistencies for subscales (αs commonly around 0.70–0.88). In this study, each subscale showed good internal consistency (α ≈ 0.80–0.85).

Motivational beliefs were measured using Motivated Self-Directed Learning and Collaborative Learning Questionnaire (MSDLCL) (Choy et al., 2016; Wang et al., 2023), which measures task value (perceived importance and usefulness of the learning activities), self-efficacy (confidence in one’s ability to succeed in the learning tasks), extrinsic goal orientation (emphasis on external outcomes such as grades, rewards, or approval), self-directed learning (SDL) (tendency to plan, monitor, and regulate one’s learning), self-directed learning with ICT (SDL-ICT) (using digital tools, e.g., online resources, learning platforms, to self-direct learning), collaborative learning (CL) (working with peers to understand course content), and collaborative learning with ICT (CL-ICT) (using ICT, e.g., discussion boards, video calls, collaborative documents for teamwork). The MSDLCL is a 28-item questionnaire organized into seven subscales, each comprising four items, rated on a 5-point Likert scale (1–5), with higher scores indicating stronger endorsement of the relevant motivational belief or learning behavior. Reliability estimates for the subscales in previous research typically ranged from α ≈ 0.78 to 0.85, indicating adequate internal consistency across the dimensions (Choy et al., 2016; Wang et al., 2023). In this study, each subscale also showed good reliability (α ≈ 0.75–0.91).

Attentional control in the academic context was measured with items adapted from attentional control frameworks (Derryberry and Reed, 2002; Hodis and Hodis, 2025). The attentional control scale (ACS) comprises 20 items, rated on a 4-point Likert-type scale ranging from 1 to 4, with the original anchors typically described as almost never (1) to always (4). The scale assesses two components of attentional control: focusing and shifting. The ACS is most often used as a single composite score, representing overall self-regulatory control of attention. In their original work, Derryberry and Reed (2002) reported satisfactory internal consistency for the total scale and demonstrated that ACS scores moderated anxiety-related attentional biases in experimental cueing paradigms, supporting the construct and predictive validity of the measure. In the present study, the ACS was used as an indicator of students’ perceived capacity to maintain and flexibly shift attention during academic activities, with higher scores interpreted as reflecting more effective voluntary attentional control. Internal consistency was good (α = 0.88). Reliability results for the scales/subscales are presented in Table 1.

Data was screened for outliers and missingness and were handled using listwise deletion. For the LPA, the three need frustration variables were z-standardized to place them on a common metric. For logistic regression, continuous predictors (task value, self-efficacy, extrinsic goal orientation, SDL, SDL-ICT, CL, CL-ICT, attentional control) were standardized to aid interpretation of coefficients.

Statistical analyses were conducted in JAMOVI 2.6. LPA used the tidyLPA package, with some model comparisons aided by the snowRMM module for JAMOVI (Bauer, 2022; Sequeira and Borges, 2024; Rosenberg et al., 2021). We conducted LPA on the three need frustration indicators (autonomy, competence, relatedness) to identify latent subgroups of learners with similar frustration patterns. Models specifying 1 to 4 latent classes were estimated, and we compared models using multiple indices: Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), sample-size adjusted BIC (SABIC), entropy, and the Bootstrap Likelihood Ratio Test (BLRT). Lower AIC/BIC and significant BLRT p-values indicate better fit.

After selecting the optimal profile solution and assigning each student to the profile with highest posterior probability, we used binomial logistic regression to predict profile membership (High vs. Low Need Frustration). The dependent variable was coded such that 1 = High Need Frustration and 2 = Low Need Frustration (for interpretability, we report results where positive coefficients indicate greater odds of being in the low-frustration profile). Predictors were task value, self-efficacy, extrinsic goal orientation, SDL, SDL-ICT, CL, CL-ICT, and attentional control. All predictors were entered simultaneously to estimate unique effects. Multicollinearity was evaluated via VIFs and tolerance; all VIFs were below 3 (tolerances ≥ 0.34), indicating acceptable levels of predictor overlap. Residual and influence diagnostics showed no problematic outliers or high-leverage cases.

While this study offers some insights, it has several limitations. First, the use of cross-sectional self-report data means that a snapshot of students’ perceptions and motivations were captured at one point in time. Causal interpretations should therefore be made with caution. Longitudinal designs would help determine how frustration profiles evolve and whether, for instance, prolonged frustration leads to increasing amotivation or dropout. Experimental studies could also test the effectiveness of the recommended interventions (e.g., introducing an autonomy-supportive module and observing if previously frustrated learners shift to a more positive profile). Besides, the sample consisted of e-learners in a particular context; generalizability to other populations (such as K-12 online learners or corporate e-training participants) may be limited. Different age groups or cultural contexts might exhibit different profile structures or predictor effects. Future research should replicate the profile structure in diverse educational systems and course types, including fully asynchronous and synchronous formats. Future research could also compare profiles across diverse e-learning settings or examine additional predictors (e.g., perceived technological support, learning environment usability) which might also impact need frustration.

In this study, we focused on need frustration rather than need satisfaction, but the frustration is not merely the opposite of satisfaction but a separate construct with its own outcomes (Bartholomew et al., 2011; Liu et al., 2024). It would be enlightening to investigate if a parallel person-centered approach on need satisfaction yields complementary profiles (perhaps mapping onto high vs. low thriving), and how those relate to our frustration-based profiles. This could deepen our understanding of the full spectrum of learner experiences.

Although the logistic model identified significant predictors, the explained variance was modest. This is common in educational and motivational research, but it shows that many other factors likely contribute to whether learners experience high or low need frustration. Integrating contextual measures (e.g., perceived instructor autonomy support, platform usability, gamification features) and personality traits (e.g., conscientiousness, neuroticism) may yield richer models.

In addition, in this study, we examined self-directed and collaborative learning tendencies, and did not directly measure actual behaviors (e.g., logged time on task, forum participation). Combining self-report with behavioral metrics from learning management systems could reveal how motivational profiles translate into concrete engagement patterns.

Despite limitations, this study contributes novel evidence to the literature on online learner motivation. By integrating SDT and EVT perspectives in a latent profile analysis, it provides a differentiated view of e-learners: not all online students are alike – some are navigating their courses with relative fulfillment and self-regulation, while others are caught in a cycle of thwarted needs and controlled motivation.

Future studies could incorporate measures of motivational quality (autonomous vs. controlled motivation) alongside attentional control to see how they interact in predicting need satisfaction or frustration. Longitudinal research can also test whether high-attentional-control, high-frustration students are at risk of burnout or academic maladjustment over time. Additionally, it would be important to investigate contextual moderators, such as the role of teacher or peer to peer support, learning environment variables, or parental pressure. Finally, research could explore the efficacy of interventions (e.g., self-compassion) that address the psychological need deficits of need-frustrated but focused students. Recognizing that strong attentional control can coexist with need frustration opens up new questions about individualized learning (Dietrich et al., 2021; Bayly-Castaneda et al., 2024) and possible contribution of educators and psychologists to supporting students in the online learning environments.

Based on the findings of this study, the following practical recommendations can be suggested to educators, instructional designers, and policymakers.