Introduction

Through interviews, single-administration questionnaires, and repeated experience sampling procedures, individuals described flow or optimal experience as an extremely positive, complex and gratifying experience characterized by deep concentration, absorption, enjoyment, control of the situation, clear feedback on the course of the activity, clear goals, intrinsic reward, and perceived high opportunities for action (challenges) balanced with high personal skills (Csikszentmihalyi 1975, 1990; Csikszentmihalyi and Csikszentmihalyi 1988; Massimini and Delle Fave 2000; Massimini et al. 1987). In particular, flow was defined as an autotelic experience (Csikszentmihalyi 1975, 1990), that is having a purpose in its own right: Individuals perform associated activities (labelled optimal activities) based on the satisfaction of behaving “for its own sake” (Deci and Ryan 2000).

A privileged area of flow investigation has been the leisure domain (Abuhamdeh and Csikszentmihalyi 2009; Delle Fave et al. 2003; Jackson and Csikszentmihalyi 1999), as optimal activities are often “those that have been designed for the purpose of enriching life with enjoyable complex experiences, such as games, sports, or artistic and literary forms” (Csikszentmihalyi 1990, p. 51). These activities are freely chosen, allowing for creative self-expression and cultivation of personal skills. However, optimal experiences were also detected in productive life areas, such as the educational setting (Asakawa 2009; Bassi et al. 2007; Delle Fave and Bassi 2000; Csikszentmihalyi and Larson 1984; Hektner 2001), and the work domain (Cskiszentmihalyi and LeFevre 1989; Delle Fave and Massimini 2003; Haworth and Hill 1992; Rheinberg et al. 2007; Salanova et al. 2006), which are compulsory by definition.

While studies in the leisure domain substantially confirmed the characteristics of optimal experience across various samples and activities, investigations in the productive life areas have repeatedly identified some peculiarities (Delle Fave 2007; Delle Fave and Massimini 2005). In particular, they showed that optimal experience is characterized by a stable and positive cognitive core—including components such as high concentration and control of the situation—which does not show remarkable variations across domains. On the contrary, its affective (happiness), motivational (wish doing the activity, free, involvement), and volitional (perceived goals and stakes) components take on different values according to activities. More specifically, optimal experience in productive activities is characterized by average or low levels of happiness, the perception of long-term goals, and low short-term desirability and perceived freedom (Delle Fave and Massimini 2005).

To date few investigations have delved deeper into the motivational characteristics of optimal experience in the productive area. In this study, in particular, we addressed this topic, analyzing schoolwork activities in a group of adolescents. The investigation of the motivational characteristics of optimal experience during schoolwork can advance our knowledge on the flow construct and provide practical information on the promotion of well-being at school. In particular, we attempted to build a bridge between flow theory (Csikszentmihalyi 1975) and self-determination theory (SDT; Deci and Ryan 1985a). The reasons for our choice were twofold: First, SDT has had extensive application in the educational setting, yielding relevant findings concerning both academic achievement and individual well-being; second, in their early works, both Csikszentmihalyi (1975) and Deci and Ryan (1985a) acknowledged flow as an intrinsically motivated state contributing to individuals’ personal growth.

Self-determination and learning

SDT distinguishes different types of motivation on the basis of the perceived locus of causality and degree of autonomy in behavior regulation (Deci and Ryan 2000). Autonomous motivation represents the highest level of self-determination and involves the experience of volition and choice; controlled motivation represents the lowest level of self-determination and involves the experience of being pressured and coerced (Rigby et al. 1992; Vansteenkiste et al. 2006); amotivation is a state in which individuals lack the intention to behave, or the ability to regulate themselves with respect to a given behavior. Intrinsic motivation is autonomous by definition, and an intrinsically-motivated activity is undertaken for its inherent interest and enjoyment. In addition, intrinsic motivation is strictly connected with the satisfaction of the basic psychological needs for competence, relatedness, and autonomy, referring to innate and life-span tendencies toward achieving effectiveness, connectedness, and coherence (Deci and Ryan 2000). On the contrary, extrinsic motivation characterizes activities performed to attain an outcome that is separable from the activities themselves. However, extrinsic motivation is not invariantly controlled, as the initially external regulation of behaviors can be internalized to various degrees. Three levels of internalization have been identified: introjection, identification, and integration. In particular, identified and integrated motivations, even though still extrinsic in nature, approximate intrinsic motivation in terms of extent of autonomy, and can have similar beneficial effects on performance and well-being (Deci and Ryan 2000).

In the educational setting, autonomy-promoting learning conditions were positively associated with students’ deeper active learning versus rote learning (Grolnick and Ryan 1987), lower drop-out rates (Vallerand and Bissonnette 1992), and well-being at school (Deci and Ryan 2000) than controlling conditions. Additionally, internalized extrinsic motivation and intrinsic motivation were shown to have differential effects on students’ achievement and well-being (Rigby et al. 1992). In particular, while intrinsic motivation impacts on well-being—in terms of positive affect and satisfaction with life—independently of academic performance, internalized extrinsic motivation is predictive of academic performance, and its effect on well-being is contingent on performance (Burton et al. 2006). In the same vein, a study on representative adolescent samples from 26 countries showed that interest in reading mostly had direct effects on students’ reading performance, and that instrumentally-motivated students performed better because of their frequent use of self-regulated learning strategies (Artelt 2005).

Optimal experience and learning

Since Csikszentmihalyi’s early works (1975), a wide range of studies have explored the phenomenology and the antecedents of optimal experience. Both single-administration questionnaires and repeated real-time assessments (Experience Sampling Method; Csikszentmihalyi et al. 1977) have been adopted.

In line with the theory (Csikszentmihalyi 1975), findings identified the balance between perceived high challenges and high skills in the activity as a powerful flow predictor (Chen et al. 1999; Guo and Poole 2009; Mesurado 2009; Moneta and Csikszentmihalyi 1996). Experimental designs confirmed the causal impact of this balance on the emergence of flow, by manipulating challenges/skills relationship in performing physics exercises (Pearce et al. 2005) and computer games (Keller and Bless 2008; Keller and Blomann 2008), and then assessing matching scores of some of its cognitive, motivational, volitional, and affective components.

Concerning flow phenomenology, studies identified experiential differences across activities, supporting Delle Fave and Massimini’s suggestion that flow experience may not be characterized by a rigidly recurrent structure (2005); rather, there could be a family of optimal experiences related to the features of associated activity domains (see Delle Fave et al. 2011 for a review). As reported above, high cognitive investment characterizes all optimal activities, whereas affective, motivational and volitional variables vary across them. In particular, during productive tasks (work and schoolwork) optimal experience is characterized by low levels of happiness, freedom and wish to do the activity (Cskiszentmihalyi and LeFevre 1989; Haworth and Hill 1992).

The same results were obtained across cultures (1) when optimal experience was measured indirectly by means of challenges/skills balance (Delle Fave and Massimini 2005), and (2) when it was measured directly through questionnaires in which participants were asked to read quotations describing optimal experience, identify activities in which they have had such experience and to rate on Likert-type scales the levels of cognitive, affective, motivational and volitional variables, as well as the levels of challenges and skills perceived in the situation (Delle Fave 2007; Delle Fave et al. 2011).

Concerning productive activities, work represents the primary source of flow among adults (Cskiszentmihalyi and LeFevre 1989; Delle Fave and Massimini 2005; Haworth and Hill 1992), and schoolwork represents a potential source of flow among students (Bassi and Delle Fave 2004), posing highly challenging tasks in which individuals can invest personal abilities. In particular, studies have shown that the association of learning activities with optimal experience has both short-term consequences in terms of rewarding engagement in learning (Bassi et al. 2007), and far reaching implications in promoting longitudinal coherence in the amount of time devoted to study (Hektner 1996), in predicting the level of academic career students are willing to pursue (Hektner 1996; Wong and Csikszentmihalyi 1991), and in shaping individual long-term goals (Asakawa and Csikszentmihalyi 1998; Delle Fave and Massimini 2005).

Joining perspectives

As stated by Deci and Ryan (2000), there are both points of convergence and points of divergence between flow theory and SDT. Both theories emphasize the importance of optimal challenge. The postulate of optimal challenge is consistent with SDT specification of the competence need as a basis for intrinsic motivation: It is success at optimally challenging tasks that allows people to feel a true sense of competence. In addition, both theories stress the importance of phenomenology. Instead of focusing on distal causes of motivation, Csikszentmihalyi (1990) focused on the proximal causes, with the idea of a phenomenal experience being a sufficient reason for action. This is consistent with SDT emphasis on the functional significance of events as a determinant of motivation.

According to Deci and Ryan (2000) the most important point of divergence is that flow theory does not have a formal concept of autonomy. The authors maintain that flow cannot be engendered unless people experience themselves as autonomous in performing an activity. An exclusive focus on optimal challenge cannot address the dimension of perceived locus of causality. Additionally, flow theory does not deal with more versus less volitional forms of extrinsic motivation that result from the degree to which external regulations have been internalized and integrated with the self.

A privileged area to investigate autonomy and self-determination in relation to optimal experience is namely the school domain. While leisure activities are freely chosen (they are also referred to as free time from duties; Dumazedier 1999), academic tasks are compulsory by definition. In addition, school-related optimal activities were shown to depart from theoretical expectations in their motivational dimension (Delle Fave and Massimini 2005). To the best of our knowledge, only two studies on quality of experience and self-determination were performed in the academic setting, both involving talented high-school students. One of them (Wong and Csikszentmihalyi 1991) investigated intrinsic motivation as enjoyment and motivation directed by long-term goals (defined work orientation), without specifying whether goals were intrinsic or extrinsic (Kasser and Ryan 1996; Sheldon et al. 2004). This study revealed that work orientation was significantly associated with the amount of time participants spent studying, but was unrelated to their experience while studying. By contrast, intrinsic motivation was related to enjoyable studying experiences, but not to the amount of time spent studying (Wong and Csikszentmihalyi 1991). The other study (Wong 2000) focused on causality orientations and showed that autonomy, relative to control, was positively related to academic experience and to flow.

The present study

Given Deci and Ryan’s (2000) claim of the importance of taking into account autonomy and self-determination in flow theory, and the dearth of empirical findings on this issue, the present paper aimed at exploring a series of pertinent questions: What levels of self-determination most frequently characterize optimal experience in school activities? Are there differences in the experiential profile of school-related optimal activities based on locus of causality? What are the implications for the quality of students’ learning?

Participants in our study were Italian high-school students followed for 1 week with Experience Sampling Method (ESM; Csikszentmihalyi et al. 1977), a tool providing repeated real-time assessments of daily activities, the associated quality of experience, as well as perceived self-determination levels. The use of ESM rested on the premise that one-time assessment measures rely too much on people’s memory and judgment and thus may not accurately represent what happens in everyday life (Hektner et al. 2007). The Experience Fluctuation Model (Massimini et al. 1987) allowed us to calculate the proximal conditions of optimal experience, namely the balance between high perceived challenges and matching personal skills.

We preliminarily aimed at comparing the quality of experience associated with schoolwork perceived as optimal activity (OA)—that is when a balance between high challenges and high skills was reported—and schoolwork perceived as not optimal activity (NOA), when this balance was not reported. We expected to replicate extant findings, in particular to detect a more positive experiential profile during schoolwork as OA, but also low levels of freedom and wish to do the activity.

We next addressed our major research questions regarding self-determination levels and quality of experience during schoolwork as OA. Three degrees of self-determination were identified based on the perceived level of autonomy in behavior regulation: high, moderate, and low, with the high degree representing autonomous regulation, the moderate degree representing mixed autonomous and controlled regulation, and the low one representing controlled regulation. First, we compared the degrees of self-determination during schoolwork as OA and as NOA. Based on the review of the literature, our research hypothesis was that during schoolwork as OA participants would more frequently report a high degree of self-determination. Our second hypothesis concerned the quality of experience. In particular, we expected that the global experience during schoolwork as OA would be more positive when participants reported high versus low self-determination. No specific hypothesis was formulated for single experiential variables and for moderate self-determination levels on the grounds that literature provided mixed results, if any, on these issues.

Methods

Participants

The participants in this study were 268 Italian adolescents, 152 girls and 116 boys, ranging in age from 15 to 19 years. They were all secondary school students equally drawn from two metropolitan areas, one in Northern and one in Central Italy. Adolescents came from middle class families. Permission for conducting the study was obtained from participants, their parents, and school councils.

Instruments and procedure

Participants were administered Experience Sampling Method (ESM; Csikszentmihalyi et al. 1977), a procedure providing information on contextual and experiential aspects of daily life through real-time repeated self-reports completed during the unfolding of daily events and situations. For 1 week, each participant was given an electronic agenda and a booklet of experience sampling forms (ESFs).

Agendas were programmed to send random acoustic signals 8 times a day from 8.00 am to 10.00 pm. When beeped, participants were asked to fill in a form, containing a standard set of open-ended questions and Likert-type 0–12 scales. The open-ended questions investigated thoughts, activities, locations, and social context, as well as short- and long-term goals of the activities: for example, when beeped, “what were you doing?”. A multiple-choice question investigated the perceived level of self-determination in performing the ongoing activity (“Why were you doing it?”). Answers comprised: I wanted to do it, I had to do it, I had nothing else to do. Participants could select more than one answer. The quality of experience perceived at signal receipt was assessed through Likert-type scales ranging from 0 (“not at all”) to 12 (“to the maximum”). These scales measured the levels of affective (e.g. happy), cognitive (e.g. concentrated), and motivational variables (e.g. free), as well as the level of perceived challenges and skills. “Appendix” shows the English version of the ESF that was used in this study. The validity and reliability of the ESM method and assessment have been described in Hektner et al. (2007).

At the beginning of the sampling week, participants were briefed about the use of ESM in small groups at their schools, after classes. They were invited to fill in a trial ESF which was then discarded from analysis. They were also asked to raise questions or doubts they envisaged in the ESM procedure. After 1 week, the participants handed in their ESM agendas and booklets and were debriefed.

Data analysis

In a preliminary phase, we discarded ESFs that were completed more than 20 min after signal receipt, in order to avoid distortions due to retrospective recall (Larson and Delespaul 1992). Altogether, participants filled out 10,326 valid ESFs, 68.8% of the programmed beep schedule (on average 38.5 sheets per participant for the testing week).

Answers to open-ended questions were treated according to well-established procedures (Csikszentmihalyi 1997; Delle Fave and Bassi 2000; Delle Fave et al. 2003; Hektner et al. 2007). Three researchers coded and grouped each answer into broader functional categories using extant manuals. In order to increase coding homogeneity, calibration meetings were organized to discuss controversial items and to reach consensus on common rules for coding. Interrater reliability for coding daily activities amounted to 96%. In this study, we focused on schoolwork activities comprising tasks such as attending classes, listening to the teacher, taking a written/oral exam, taking notes. Concerning the self-determination levels perceived by participants in performing a given activity, we distinguished three categories: “I wanted to do it”, “I wanted and had to do it” and “I had to do it”, respectively representing high, moderate and low levels of self-determination. We further assimilated the answer “I had nothing else to do” with lack of self-determination (corresponding to lack of regulation). Other combinations of answers were kept in a separate category.

In the analysis of the quality of experience, we focused on the following variables: concentrated, in control, happy, involved, free, wish doing the activity, stakes (short-term importance of the activity), and goals. As shown above, these represent key dimensions of optimal experience (Csikszentmihalyi and Larson 1984; Delle Fave and Massimini 2005; Hektner and Asakawa 2000; Mesurado 2009). Correlations between the variables are reported in Table 1.

Table 1 Pearson’s correlations between experiential variables
Full size table

In line with the literature (Delle Fave et al. 2011), we calculated the balance between high challenges and high skills associated with optimal experience, using the Experience Fluctuation Model (Massimini et al. 1987). This model of analysis has been widely used in flow research (Clarke and Haworth 1994; Csikszentmihalyi and Csikszentmihalyi 1988; Csikszentmihalyi et al. 1993; Delle Fave and Bassi 2000; Delle Fave and Massimini 2005; Haworth and Evans 1995), and is based on the arctangent function of challenges and skills. Through this function the circumference is partitioned into sectors, whose limits (angles) represent different ratios of challenges and skills. For each participant, values of challenges and skills were first transformed into z-scores by subtracting each value from the item mean and dividing it by its standard deviation. They were then entered into the trigonometric function to identify the condition of high challenges and high skills, versus other challenges/skills combinations. In particular, the sector of the circumference corresponding to high challenges and high skills is found in the upper right quadrant included between 22.5° and 67.5°. A total of 1,710 ESFs fell into the condition high challenges and high skills (for challenges: M = 1.00, SD = .55, range .06/3.76; for skills: M = .84, SD = .47, range .05/3.38), and 8,616 ESFs fell into other challenges/skills combinations (for challenges: M = −.20, SD = .93, range −4.72/6.78; for skills: M = −.17, SD = .98, range −6.48/4.22).

In order to assess the associated quality of experience, also the values of the experiential variables were standardized. Aggregated values (mean z-scores) were calculated for each individual, and subsequently averaged on the number of participants (subject-level analysis; Hektner et al. 2007; Larson and Delespaul 1992). Based on ESM literature, t tests were then performed to assess whether experience mean z-scores significantly differed from zero.

In line with the purposes of our study, through Experience Fluctuation Model we identified those students who reported schoolwork both as optimal activity (OA—balance between high challenges and high skills) and not as optimal activity (NOA—other combinations of challenges and skills) over the testing week. These were 155 adolescents, 88 girls and 67 boys, amounting to 57.8% of the initial sample. They filled in 5,985 valid ESFs, of which 340 regarded schoolwork as OA, and 896 schoolwork as NOA (total N = 1,236). The remaining participants (N = 113) never reported schoolwork as OA, and were thus not analyzed further. No significant differences in age, gender, and frequency of schoolwork activities were detected between the two groups of students. For the targeted participants, we first investigated and compared the experience associated with schoolwork as OA and as NOA through paired t tests.

To test our first hypothesis, we analyzed the mean frequency distribution of self-determination levels reported during schoolwork as OA and as NOA through paired t tests. The following categories were identified, as reported above: high, moderate, low and lack of self-determination, and other combinations. Five participants did not report their level of self-determination (either during schoolwork as OA or during schoolwork as NOA), and were thus discarded from this comparative analysis. Mean percentages were thus calculated on 150 participants. To test our second hypothesis, we analyzed the quality of experience during schoolwork as OA when the 3 degrees of self-determination were reported: high (“I wanted to do it”), moderate (“I wanted and had to do it”), and low (“I had to do it”). For this analysis, we discarded other five participants who reported either lack of self-determination or other combinations which were not the object of our investigation. Altogether this dataset on schoolwork as OA comprised 145 students and 300 ESFs. The most suitable way to address this analysis is to adopt a multilevel modeling approach. ESM data present a typically hierarchical structure, with repeated measures (or beeps) nested within individuals (Bryk and Raudenbush 1992). Contrary to standard statistical tests, multilevel models are designed to analyze variables from different levels simultaneously, using a statistical model that properly includes dependence of observations (Hox 2002).

We designed a model with the raw-scored experiential variables (e.g. concentrated) as dependent variables and self-determination as predictor at Level 1. Since self-determination was measured as a categorical variable, it was transformed into dummy variables with “I wanted to do it” or high self-determination as reference category. We thus generated two dummies: WHY1 with 1 for “I had to do it” or low self-determination and 0 otherwise; and WHY2 with 1 for “I wanted and had to do it” or moderate self-determination and 0 otherwise. The obtained model was:

$$ {\text{Y}}_{ij} = \beta_{00} + \beta_{10} *{\text{WHY}}1 + \beta_{20} *{\text{WHY}}2 + {\text{e}}_{ij} + {\text{u}}_{0j} $$
(1)

where the subscript i stands for beeps (i = 1…n j ) and the subscript j stands for individuals (j = 1…J); Y ij is one dependent experiential variable (ex. concentrated); β00 is the intercept, that is the grand mean value of Y ij when the predictors equal 0, in other words when self-determination was “I wanted to do it”; β10 and β20 are the regression coefficients for the explanatory variable self-determination; e ij is the Level 1 residual error term; u0j is the Level 2 residual error term.

In order to control for gender and age differences, we initially added the Level 2 predictors gender and age in the model. Since no significant effects were detected, we did not consider them in our final model (1).

Analyses were performed by means of software HLM 6.07 (Bryk and Raudenbush 1992). First we calculated an intercept-only model for each experiential variable; then we added the self-determination predictor to the model. We used Full Maximum Likelihood (FML) as the estimation procedure of the parameters, as it has the advantage of providing an overall Chi-square test based on the likelihood (the deviance) that could be used to compare our nested model with the intercept-only model (Hox 2002).

Results

Schoolwork and associated quality of experience

We initially calculated participants’ activity distribution over the tested week (total ESFs = 5,985). Schoolwork was the most frequently reported activity category, amounting to 20.7% of the ESFs. Other frequent categories were interactions (14.8%), maintenance (13.8%, including personal hygiene, eating, resting), studying at home (12.2%), and watching TV (10.2%). Schoolwork was associated in 27.5% of the self-reports with high challenges and high skills (the condition of optimal experience).

Table 2 illustrates the quality of experience associated with schoolwork as OA and as NOA. The scores of the experiential variables were standardized at the subject level, as reported in the methods section. T tests were performed to assess whether mean z-scores significantly differed from zero within each condition. We next compared the values in the two conditions (OA vs. NOA) with paired t tests. As we performed a large number of t tests on the same dataset, we corrected significance levels by means of the conservative Bonferroni approach (Abdi 2007; Bassi and Delle Fave 2004; Delle Fave et al. 2003): Obtained p values were multiplied by 16 for the one-sample t tests and by 8 for the paired t tests.

Table 2 Quality of experience associated with schoolwork
Full size table

Schoolwork as OA was associated with significantly high values of the variables concentrated (t(154) = 9.56, p < .002), in control (t(154) = 6.54, p < .002), involved (t(154) = 6.48, p < .002), stakes (t(154) = 10.61, p < .002) and goals (t(154) = 7.99, p < .002). When schoolwork was reported as NOA, the scores of stakes and goals were significantly above average (t(154) = 7.58, p < .002 and t(154) = 6.35, p < .002, respectively), and significantly below average scores were detected for the variables in control (t(154) = −7.48, p < .002), happy (t(154) = −7.26, p < .002), free (t(154) = −10.71, p < .002), and wish doing the activity (t(154) = −11.36, p < .002).

The experience associated with schoolwork as OA was much more positive in its cognitive, affective, motivational and volitional dimensions compared to schoolwork as NOA. Paired t tests highlighted significant differences for all the variables: concentrated (t(154) = 6.83, p < .001), in control (t(154) = 9.49, p < .001), happy (t(154) = 4.47, p < .001), involved (t(154) = 6.27, p < .001), free (t(154) = 4.92, p < .001), wish doing the activity (t(154) = 5.28, p < .001), stakes (t(154) = 4.98, p < .001), and goals (t(154) = 3.48, p < .006).

The self-determination levels during schoolwork

We next analyzed the mean percentage distribution of reported self-determination levels in performing schoolwork as OA and as NOA, calculated on 150 participants. As shown in Table 3, contrary to expectations, participants primarily perceived schoolwork as a compulsory activity in both conditions, mainly reporting “I had to do it” (low self-determination). However, students more frequently reported “I wanted and had to do it” (moderate self-determination) during schoolwork as OA than during schoolwork as NOA (t(149) = 2.30, p < .03), followed by “I wanted to do it” (high self-determination).

Table 3 Mean percentage distribution of answers to the question “why were you doing it?”
Full size table

Table 4 summarizes the results of the HLM analyses in which the experiential variables were the dependent variables and the self-determination dummies “I had to do it” and “I wanted and had to do it” were Level 1 predictor. As described in the methods, the answer “I wanted to do it” was the reference category.

Table 4 Coefficients (and standard errors) of hierarchical linear models
Full size table

Overall, the Level 2 variance component revealed that there was significant variation across individuals for all outcome variables, thus attesting to the suitable use of multilevel modeling in data analysis. In addition, the deviance Chi-square statistics showed that the inclusion of the self-determination predictor in the model significantly improved the model fit, except for the outcome variables stakes and goals.

The analysis of Level 1 fixed coefficients showed that for most experiential variables the “I wanted and had to do it” coefficients were not significant, and the “I had to do it” coefficients were significant and negative: concentrated (t(295) = −3.16, p < .001), in control (t(293) = −2.38, p < .02), happy (t(297) = −2.98, p < .004), free (t(294) = −3.61, p < .001), and wish doing the activity (t(297) = −2.84, p < .005). Results thus showed that the quality of experience associated with schoolwork as OA was significantly more negative when individuals reported low self-determination than when they reported high self-determination. In addition, the quality of experience did not differ when individuals reported high self-determination and moderate self-determination.

For the variable involved, the “I had to do it” coefficient was not significant, and the “I wanted and had to do it” coefficient was significant and positive (t(295) = 2.32, p < .03). Individuals thus felt more involved in association with moderate self-determination than with high self-determination, and there were no differences between the conditions high and low self-determination. Finally, no relationship was detected between the self-determination levels and the outcome variables stakes and goals.

Discussion

Starting from Deci and Ryan’s (2000) claim on the importance of investigating autonomy in people’s perception of flow, this study examined the levels of self-determination adolescents reported during schoolwork activities. Based on the review of the literature, we expected that schoolwork as OA would be more frequently associated with a high degree of self-determination, compared to schoolwork as NOA. In addition, we expected to detect a more positive experience when schoolwork as OA was associated with high versus low self-determination. Our analyses revealed three major findings.

First, analyzing the cognitive, affective, motivational and volitional components of experience we replicated evidence—obtained with both ESM and single-administration questionnaires—that schoolwork as OA (high challenges and high skills) was associated with an optimal experiential profile compared to schoolwork as NOA. Participants reported high concentration and control, high short-term importance and relevant long-term goals, as well as high involvement. As expected, however, low even though not significant levels of the variables happy, free and wish doing the activity were also detected.

Second, contrary to our expectations, the analysis of the self-determination levels during schoolwork as OA and as NOA revealed that students mostly reported “I had to do it” in both conditions. This finding was quite surprising, considering that high challenges and high skills are antecedents of flow and that flow is defined as an intrinsically-motivated state characterized by autonomous regulation (Csikszentmihalyi 1975; Deci and Ryan 1985a). Even though we did not directly measure intrinsic motivation in terms of enjoyment and interest (Deci and Ryan 1985a), previous studies with both ESM and single-administration questionnaires highlighted activity-related variation in the features of optimal experience, with productive activities being characterized by low levels of intrinsic motivation compared to leisure activities such as sports and hobbies or socializing (Cskiszentmihalyi and LeFevre 1989; Delle Fave 2007; Delle Fave and Massimini 2005). In our study, we focused on a positive predictor of intrinsic motivation, namely perceived autonomous regulation: Findings pointed to the fact that autonomy does not seem to be an essential aspect of schoolwork as OA, and that compulsory activities can be associated with the experience of flow. In particular, controlled regulation does not seem to preclude opportunities for optimal experience at school, at least not in our study. These results can be related to the level of analysis applied in this study which focused on real-time contingences rather than trait characteristics (Delle Fave et al. 2011). When motivation and flow are assessed as situational constructs (such as through ESM), extrinsic contingent factors have a relevant influence on individuals’ ongoing experience. In addition, as highlighted by Hektner and Asakawa (2000), students reporting flow in learning cannot sustain their efforts through intrinsic motivation alone, but also need extrinsic rewards and social recognition for their accomplishment.

Our findings however indicated a possible trend toward internalization of regulation, on the grounds that schoolwork as OA was more frequently associated with “I wanted and had to do it” than schoolwork as NOA. Individuals may engage in an activity because they are initially forced to do so by the surrounding environment (family members, teachers). People often need external incentives to take the first steps in an activity that requires effort and attention (Csikszentmihalyi 1990). In the course of time, the activity can be perceived as an opportunity for optimal experience; thus a virtuous circle can be triggered in which challenges and skills are increased, on the one hand, and locus of causality is gradually internalized.

The third finding of our study is strictly related to this latter point; it regards the features of optimal experience during schoolwork when different degrees of self-determination were perceived. HLM analyses allowed us to inspect whether there were differences in the experiential profile of school-related optimal activities based on locus of causality, by contrasting high self-determination (“I wanted to do it”; autonomous regulation) with low self-determination (“I had to do it”; controlled regulation) and moderate self-determination (“I wanted and had to do it”; mixed autonomous and controlled regulation). Results substantially supported our hypothesis that when participants reported autonomous regulation during schoolwork as OA, their quality of experience would be more positive than in conditions of controlled regulation. This was true of most experiential variables—concentrated, in control, happy, free, and wish doing the activity—for which item scores were higher with “I wanted to do it” than “I had to do it”. No specific prediction was made in relation to the moderate self-determination level “I wanted and had to do it”. Findings largely showed no significant difference in the experiential variables when either “I wanted and had to do it” or “I wanted to do it” were reported. This could be due to the fact that the moderate self-determination level is perceived as both controlled and autonomous; thus it can have a positive effect on the quality of experience.

Unexpected results were obtained for the variable involved. Students perceived themselves as more involved when reporting a moderate level relative to a high level of self-determination. Additionally, no significant difference in involvement was detected between high and low self-determination levels. A vast literature attests to the positive relationship between autonomy and task engagement (Amabile et al. 1994; Deci and Ryan 2000). However, in line with the studies highlighting the positive relationship between internalized extrinsic motivation and academic performance (Artelt 2005; Burton et al. 2006), our results suggested that the presence of moderate levels of self-determination (mixed autonomous and controlled regulation) can enhance students’ involvement in academic activities.

Results contrary to our expectations were obtained also for stakes and goals, in that self-determination levels did not have predictive effect on these variables. A possible interpretation could rest on previous ESM studies showing that high levels of stakes and goals are stable characteristics of the schoolwork experience (Delle Fave and Massimini 2005). In addition, SDT research has highlighted that goal contents (the “what” of goal pursuit) and goal motives (the “why” of goal pursuit) are different concepts and can predict independent variance in well-being and adjustment (Sheldon et al. 2004).

To sum up our major findings, we observed that optimal experience during schoolwork can be perceived also when participants report controlled regulation, which was actually the most frequent condition in our sample. In spite of its predominant frequency, however, it was not associated with the best quality of experience. As expected, this was primarily reported in relation to high or moderate self-determination levels.

Findings have both relevant theoretical and practical implications. At the theoretical level, some researchers would object that optimal experience should not only be identified by the balance between perceived challenges and skills—as was the case in the present study—and that other components such as goals, feedback, autonomy, and focused attention should be taken into consideration (Kawabata and Mallett 2011; Schmidt et al. 2007). In other words, only those instances in which autonomous regulation was reported could really be defined as optimal experience. However, plenty of studies using experience sampling as well as experimental methods have attested to the central role of challenges and skills in determining optimal experience (Chen et al. 1999; Guo and Poole 2009; Keller and Bless 2008; Keller and Blomann 2008; Massimini et al. 1987; Moneta and Csikszentmihalyi 1996; Pearce et al. 2005). Further experimental studies—though limited in capturing as dynamic a conscious state as flow—are needed to tackle this issue, particularly to understand what components can be predictors and what components can be correlates of optimal experience. In the light of current knowledge, however, our data suggest that optimal experience can be perceived when individuals report controlled regulation, as long as they find the ongoing activity challenging enough in the face of personal abilities. In addition, findings point to different features of optimal experience based on the self-determination levels: The more the locus of causality is internalized the better is the overall quality of experience.

At the practical level, these findings can have relevant implications in education. On the one hand, Deci and Ryan (2000) have stressed that children often take spontaneous interest in and assimilate new material, so that learning can take place without external direction. On the other hand, most current school systems—especially in secondary education—are indeed based on external direction, and learning is contingent on rewards (Shernoff and Csikszentmihalyi 2009). Even though school is a highly controlled setting, our data showed that individuals can retrieve in it opportunities for optimal experience. This suggests the importance of paying attention to the relationship between perceived challenges and skills in school activities, thus promoting opportunities for complex experiences in line with students’ professional and personal goals (Hektner and Asakawa 2000; Massimini and Delle Fave 2000). A controversial issue in current formal education is the simplification of learning tasks—providing low levels of challenges—in order to make them more appealing and to facilitate experiences of fun. In this condition, however, students mostly associate academic activities with experiences of apathy and disengagement (Bassi and Delle Fave 2004). Also in the present study, 42.2% of participants in the initial sample did not report flow during schoolwork. To deal with this problem, educators should offer students complex challenges and, at the same time, have a scaffolding role in students’ competence building. In line with the vast SDT literature, educators should also promote the process of internalization of behavior regulation through autonomy support (Niemiec and Ryan 2009). In this way, they would facilitate students’ adjustment in terms of deeper active learning and academic performance, as well as optimal experiences and well-being at school.

Limitations and future research

The present study provided original information on the characteristics of optimal experience at school. However, it has limitations that should be addressed in future research. The first one regards the measurement of self-determination. By distinguishing three self-determination levels we broadly identified 3 degrees of behavior regulation—including autonomy, mixed autonomy and control, and external control—and their relationship with schoolwork as OA. Additional information could be gathered by refining this operationalization, further distinguishing between external, introjected, identified, integrated, and autonomous regulation. No study on flow has thus far adopted such a refined focus of analysis, even though research contrasting intrinsic, introjected and identified regulation has detected different effects on academic performance and well-being (Assor et al. 2009; Burton et al. 2006). Such a refined analysis could also bring about useful insight into our finding that higher levels of involvement were reported at a moderate level of self-determination relative to high self-determination.

A second limitation regards the simplicity of the multilevel analysis. In our investigation we primarily focused on the relationship between one Level-1 variable, self-determination, and the quality of experience associated with schoolwork as OA. HML results pointed to a significant between-individual variability, which was not taken into consideration in the present study. In a preliminary model, gender and age were shown to have no influence. Future studies should explore other individual and personality factors that can affect experience directly and indirectly through the moderation of self-determination levels. Profitable information could be gained, for example, from the administration of the General Causality Orientations Scale (Deci and Ryan 1985b), which identifies enduring personality aspects about the perceived locus of behavior causality.

Finally, and partially connected to the second limitation, our sample size was rather small and did not allow for more sophisticated multilevel analyses that would otherwise have had numerosity problems. Larger samples are thus needed, along with the analysis of students samples belonging to different age groups and attending different school types. In addition, an even higher level of analysis could be included in the equation, accounting for the different degrees of self-determination that teachers promote at different school levels (ex. elementary vs. secondary school vs. college).

In spite of these limitations, our study identified novel findings and a series of challenging questions that future research should tackle in order to shed light on both the characteristics of optimal experience and their implications for fostering long-term commitment and enjoyment in learning activities.