"1. Summary of binary logistic regression analyses with posture features as predictors of each affective state from neutral for data sets based on affective judgments of the four judges. Next, we consulted the numerical directions (i.e. signs, + and -) of the statistically significant coefficients of the logistic regression models in order to explore relationships between body position, movement, and affect. Although all 16 features were used as predictors for the logistic regression analyses, we focus on the average pressure and the change in pressure; these are the most interpretable features and can be theoretically aligned within the attentive-arousal framework. The logistic regression analyses were used to discriminate between each affective state versus neutral, hence, a statistically significant predictor implies that the feature is heightened (significant positive predictor) or suppressed (significant negative predictor) during the emotional experience when compared to neutral. For example the back average pressure change feature was a significant positive predictor for the boredom- neutral logistic regression, so the episodes of boredom were accompanied by an increase in movement on the back when compared to neutral. Table 2. Significant predictors for the multiple regression models for emotions in each data set. A number of relationships surface when one considers the significant predictors of the affective states in which at least two judges agreed. The requirement that two judges agree on the significance and direction of each predictor is motivated by a desire to establish a degree of convergent validity in exploring the posture-affect relationships. By requiring that the features need to be significant predictors of affect for at least half of the judges models ensures that, to some extent, they generalize across judges. Boredom. Our results suggested that during episodes of boredom, the learners leaned back and presumably disengaged from the learning environment (low attentiveness indicated by an increase in pressure on back and a significant decrease in pressure on the seat). Experiences of boredom were also accompanied by an increase in the rate of change of pressure exerted on the seat. Therefore, heightened arousal was associated with the boredom experience, presumably as learners mentally disengage from the tutor and divert their cognitive resources to fidget around and alleviate their ennui. Some may view the heightened arousal accompanying boredom to conflict with the preconceived notion of boredom in which a learner stretches out, lays back, and simply disengages. Our results suggest that the learner lays back, disengages, but is aroused. Furthermore, this pattern of increased arousal accompanying disengagement (or boredom) replicates a previous study by Mota and Picard [14]. They monitored activity related posture features and discovered that children fidget when they were bored while performing a learning task on a computer. Delight and Flow. In contrast to boredom, learners experiencing the positive emotions of delight and flow demonstrate increased attentiveness towards the learning environment by leaning forward. Learners experiencing these emotions also demonstrate heightened arousal on the back of the chair – at least when compared to the neutral state. Confusion and Frustration. Similar to delight and flow, learners experiencing confusion and frustration tend to lean forward. However, it appears that during episodes of delight and flow learners lean forward at a steeper inclination than with confusion and frustration. We arrived at this conclusion because the increase in the pressure exerted on the seat of the chair was accompanied by a commensurate decrease in pressure exerted on the back of the chair for delight and flow. On the other hand, confusion was accompanied by a decrease in pressure exerted on the back of the chair without any accompanying increase on the seat. Similarly for frustration the increase in pressure on the seat was devoid of a notable (statistically significant) decrease on the back. We suspect that the pattern of body position with confusion and frustration indicates that learners are in an upright position when they experience these states, as opposed to the forward lean that seems to accompany experiences of delight and flow. Arousal. In addition to boredom, it also appears that experiences of delight, flow, confusion, and frustration are accompanied by significant arousal, either on the back or the seat of the pressure sensitive chair. The arousal that is affiliated with confusion and frustration occurs on the back while there is an increase in movement on the seat when the emotions of delight and flow are experienced. While the significance of the location of the arousal (back or seat) during experiences of these is unclear, what is important is that all affective experiences (including boredom) were accompanied by heightened arousal when compared to neutral. In summary, the experience of each affective state is accompanied by a significant increase in arousal, at least when compared to the neutral baseline. General Discussion. We discovered relationships between body position, degree of movement, and learners’ affective states. Our findings are in line with an attentive-arousal or engagement-arousal framework (see section 2). With respect to the attentiveness dimension, it appears that there are three bodily configurations that are associated with the affective states. These include heightened attentiveness, which is manifested by a forward lean when the positive emotions of delight and flow are experienced. On the other hand, bored learners tend to lean back, presumably in a state of disengagement (boredom). States such as confusion and frustration occur when learners confront contradictions, anomalous events, obstacles to goals, salient contrasts, perturbations, surprises, equivalent alternatives, and other stimuli or experiences that fail to match expectations [28]. Learners are in a state of cognitive disequilibrium, with more heightened physiological arousal, and more intense thought. Our results suggest that the bodily corollary to the mental state of cognitive disequilibrium is an alert position, where the learner sits upright and pays attention. The single major finding with respect to the arousal dimension is that each of the affective experiences is accompanied by a significantly higher arousal when compared to a neutral baseline (i.e. no emotion). This finding has important theoretical implications because some of our colleagues view some of these emotions (i.e., flow, confusion, etc.) as cognitive states, whereas other researchers would classify them as either emotions or affect states. We have traditionally agreed with the latter group because we hypothesize that the single major discriminator of an affective state over a cognitive state is that the affective state is accompanied by enhanced physiological arousal (compared with neutral). Our results indicate that in most cases there is a significant increase in bodily movements (bodily arousal) during the experience of emotional episodes indicating that both cognitive and affective processes are at play. Therefore, it might be the case that the term cognitive-affective state is the most defensible position for mental states such as confusion, flow, and frustration. We acknowledge that the aforementioned relationships between body posture and affect ignore individual differences in affect expression. In ideal circumstances, from a statistical point of view, the landscape of postural configurations would be evenly distributed among the 28 different students. However, this claim is implausible and it is therefore important to contrast contributions of individual learners versus generalizable posture features in predicting affect. Our results focused on broad patterns observed across all learners and should be interpreted with a modicum of caution. We are currently addressing these concerns by building models that attempt to separate variance explained by individual student characteristics versus variance explained by posture features above and beyond individual differences."