Converging evidence suggests that providing learners with a certain degree of self-control, or autonomy, benefits motor learning. Advantages of self-control relative to yoked control groups have been demonstrated for the provision of feedback (e.g., Janelle et al., 1997; Chiviacowsky and Wulf, 2002; Huet et al., 2009; Patterson and Carter, 2010), order of trials during multi-task practice (Keetch and Lee, 2007; Wu and Magill, 2011), model observation (Wulf et al., 2005), use of assistive devices (Wulf and Toole, 1999; Wulf et al., 2001; Hartman, 2007), amount of practice (Post et al., 2011), and task difficulty (Andrieux et al., 2012). Self-controlled practice has been shown to benefit motor learning not only in adults but also in children (e.g., Chiviacowsky et al., 2008), older adults (Alcântara et al., 2007), and people with disabilities (Chiviacowsky et al., 2012).

Providing participants with some freedom of choice has been shown to benefit performance and learning in other domains as well (Cordova and Lepper, 1996; Tafarodi et al., 1999). For example, adults working in a high task-autonomy condition, in which participants were given considerable freedom and independence in deciding how to carry out a task (Hacken and Oldham, 1976), found more creative solutions to administrative problems than in a low task-autonomy condition (Zhou, 1998). Children provided with a higher level of autonomy on an arithmetic task showed increases in intrinsic motivation, depth of engagement, perceived competence, and learning (Cordova and Lepper, 1996). Allowing individuals to exercise control over the environment may satisfy a basic psychological need (White, 1959; Deci and Ryan, 2000, 2008), and/or biological necessity (Leotti et al., 2010; Leotti and Delgado, 2011). Studies with both humans (Tiger et al., 2006) and animals (Voss and Homzie, 1970; Catania, 1975; Catania and Sagvolden, 1980) have shown that both prefer an option leading to a choice than an option that does not, even if this option results in greater effort or work. These results suggest the existence of an inherent reward with the exercise of control. Choice, or the anticipation of the opportunity for choice, is associated with increased activity of brain regions directly involved in reward processing, and is greater than it is for anticipation of reward in the absence of choice (Leotti and Delgado, 2011).

Only a few studies have attempted to examine the underlying reasons for the benefits of self-controlled practice in motor learning. Some theoretical explanations have been proposed – borrowing from the verbal or cognitive learning domain where self-regulated learning has been more widely discussed – but they are relatively vague when applied to motor learning. For example, it has been suggested that more active involvement of the learner in the learning process promotes deeper processing of relevant information (Watkins, 1984; McCombs, 1989; Chen and Singer, 1992); that it encourages the use of self-regulation strategies (Kirschenbaum, 1984); or that giving the learner control over the practice conditions might be, in a general sense, more motivating (Bandura, 1993; Boekaerts, 1996). Motor learning researchers have largely adopted the explanations, although, empirical tests of these hypotheses still appear to be lacking.

One line of evidence points to the role of motivational factors associated with self-control as a possible explanation for the observed learning benefits. In a study by Chiviacowsky and Wulf (2002), participant interviews showed that (a) self-controlled learners asked for feedback predominantly after they believed they had a “good” trial, (b) errors on feedback trials were indeed smaller than those on no-feedback trials, and (c) yoked learners indicated that they would have preferred feedback after successful trials. These findings have since been replicated (e.g., Chiviacowsky et al., 2008; Patterson and Carter, 2010; Patterson et al., 2011). This suggests that the opportunity to request information after successful performance may play an important role in self-controlled learning. Moreover, follow-up studies have shown that providing learners with feedback after “good” relative to “poor” trials can enhance learning (e.g., Chiviacowsky and Wulf, 2007; Chiviacowsky et al., 2009; Saemi et al., 2011; Badami et al., 2012). It is possible that increased perceptions of competence and/or heightened self-efficacy – accruing from the opportunity to confirm successful movement outcomes, while “ignoring” less successful attempts – play a mediating role in this process. Higher levels of self-efficacy, defined as an individual’s belief regarding her or his ability to produce a desired result (Bandura, 1977), can lead to enhanced performance in various domains (Bandura, 1993; Feltz et al., 2008; Hutchinson et al., 2008). Similarly, the desire for competence is conceptualized as a drive to feel effective in interactions with the environment (White, 1959; Deci and Ryan, 1985, 2008; Conroy et al., 2007). A desire for competence is considered a basic psychological need, along with the needs for autonomy and social relatedness (i.e., the desire to feel connected to, and accepted by, others; Deci and Ryan, 2000). The satisfaction of all three psychological needs is believed essential for ongoing psychological growth and well-being – leading to more effective learning and maintained engagement with the task (Deci and Ryan, 2008; Sheldon and Filak, 2008).

The purpose of the present study was to examine more directly if the opportunity to select feedback after good trials – with potential concomitant effects on perceived competence and self-efficacy – is critical for the typically seen benefits of self-controlled feedback. If this were the case, depriving performers of the opportunity to feel competent under self-controlled practice conditions would be expected to have detrimental effects on learning. We sought to manipulate participants’ perceived competence in the present study by providing them with a standard or reference for “good performance.” We assumed that a high performance criterion that was difficult to meet would result in few ostensibly successful trials and thus would degrade perceptions of competence, self-efficacy and, consequently, motor learning. We also asked whether a criterion that was relatively easy to reach or surpass would provide learners with a boost in perceived competence and self-efficacy that could in turn enhance learning. Three groups of participants practiced a novel anticipation timing task and were able to request feedback about their temporal accuracy after 3 trials in each of three 10-trial blocks. Two of the groups were provided different criteria regarding what would be considered good performance (i.e., errors of 30 vs. 4 ms or less, respectively), thereby influencing their opportunities to conclude that they had a successful trial. The third, a self-control only group was not given a criterion for success. Questionnaires completed by participants at the end of the practice phase were used to determine participants’ perceived competence, self-efficacy, and task interest and enjoyment and their possible role in mediating learning. In particular, we were interested in whether self-efficacy ratings would predict learning (i.e., retention and transfer performance; Stevens et al., 2012).

The purpose of the present study was to shed some light on the reasons underlying the benefits of self-controlled feedback. In previous studies (e.g., Chiviacowsky and Wulf, 2002; Patterson and Carter, 2010), self-control learners preferred and chose feedback mostly after successful trials. Therefore, we asked whether this possibility might result in heightened perceptions of competence and self-efficacy that could contribute to the observed learning advantages in self-controlled conditions. We succeeded in manipulating learners’ perceptions of competence and related self-efficacy through information about what was considered successful performance. Relative to the Self-4 group, the other groups experienced higher self-efficacy (Self-30 and Self groups) and competence (Self-30 group). Importantly, learning (i.e., retention and transfer performance) was more effective in these two groups compared with Self-4 group participants who experienced lower perceived competence and self-efficacy. Interestingly, the Self-30 group that received both the opportunity for self-controlled feedback and the criterion-defined experience of successful performance on a majority of their feedback trials, also reported heightened task interest and enjoyment relative to both the Self and Self-4 groups. However, this affective augmentation in intrinsic task motivation was not associated with more effective motor learning in the present study.

Our findings confirm those of previous studies (e.g., Chiviacowsky and Wulf, 2002, 2005; Chiviacowsky et al., 2008; Patterson and Carter, 2010; Patterson et al., 2011) in that participants generally preferred and asked for feedback mainly after successful trials. This preference may reflect the expression of a basic psychological need for competence or participants’ insights that their learning is better when success or a correlate such as self-efficacy is experienced. Yet, while opportunities for self-controlled feedback were available for all groups, this self-control was not met with the validation of good performance for those in the Self-4 condition, in which only 6% of the practice trials were considered “good” according to the provided criterion. The lack of positive feedback or receipt of essentially negative performance information resulted in reduced perceptions of competence and self-efficacy, degrading learning. This finding of impaired learning with indications of poor performance is in line with the findings of several recent studies which demonstrated that feedback provided after “poor” trials is not as effective for learning as feedback provided after “good” trials (e.g., Chiviacowsky and Wulf, 2007; Saemi et al., 2012) and that negative normative feedback degrades learning relative to positive social-comparative feedback (Lewthwaite and Wulf, 2010; Wulf et al., 2010, 2012). Interestingly, in one of those studies, which included a control group with veridical but no social-comparative information (Lewthwaite and Wulf, 2010), negative feedback also hindered learning relative to the control condition. This was not the case in the present study in which self-control without competence augmentation served as a control condition.

The similarity of both the perceived competence/self-efficacy and learning findings in the Self and Self-30 groups is notable. It may suggest that satisfaction of more than one psychological need is superfluous for learning (but see Sheldon and Filak, 2008, for some evidence of an additive effect) or that autonomy support is sufficient by itself to boost both perceived competence/self-efficacy and learning (Tafarodi et al., 1999). The fact that the Self-4 group, despite having the same self-controlled feedback opportunities as the other two groups, did not experience the same psychological and learning benefits appears to indicate that when one need (competence feedback suggesting relatively poor performance) is thwarted, or counteracts the other’s (autonomy) effect, the benefits of one (e.g., autonomy) are reduced. Alternatively, the Self group findings suggest that enhancements to perceived competence and self-efficacy may be effects of autonomy support. In the present study, in which self-controlled conditions meant the receipt of relatively positive feedback, we cannot decouple autonomy from competence information – one nearly automatically confirms the other when participants are choosing better performance trials on which to receive feedback. However, self-control effects have been found in conditions with less obvious conveyance of perceived competence and self-efficacy, such as when fewer rather than more practice trials are chosen (Post et al., 2011), under conditions of chance in which illusions of control are generated (e.g., Langer, 1975), or when choice involves incidental or trivial options that would have no ostensible impact on perceived competence or self-efficacy (Cordova and Lepper, 1996; Tafarodi et al., 1999). Tafarodi and colleagues found confidence effects for experimental options that involved choosing among names to be used in reading comprehension assessments.

The results highlight the role of motivational influences on motor learning. From an information-processing perspective, no learning differences among groups would have been expected, as all groups experienced the same active engagement in the learning process and had the same opportunity to choose feedback. In pursuit of the motivational circumstances that optimize motor learning (Lewthwaite and Wulf, 2012), several related but distinct psychological constructs have been directly or indirectly studied, including perceived competence, self-efficacy, perceived choice, intrinsic motivation or task interest and enjoyment, and positive and negative affect. These cognitive and affective motivational variables have been influenced by motivational conditions, including autonomy support (i.e., learner control of some practice conditions) and social-comparative and other forms of augmented instructions or feedback. Future research can provide more definitive evidence of the conditions and constructs, as well as the neurophysiological (e.g., dopaminergic; Jay, 2003) or neural activation effects (Leotti et al., 2010; Leotti and Delgado, 2011) associated with particular practice conditions or related cognitions or affective experiences. To do so will involve the capacity to measure distinct, preferably theoretically derived, constructs, as well as the experimental power to detect differences among them. For example, perceived competence and self-efficacy have theoretical differences but share some common conceptual ground in perceived ability and associated affective consequences generated by these perceptions (Bandura, 1997). Yet, they diverge in a number of potentially important ways for skill acquisition. It may be the cognitive distinctions or affective similarities that contribute to their motivational impacts on motor learning.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.