An individual’s overall AR to a given situation, such as a bout of exercise, may be shaped by more complex affective phenomena such as specific domains, emotions and mood states (Ekkekakis and Petruzzello 2000). For example, an individual’s overall feelings during a given situation may range from very good or pleasurable to very bad or displeasurable. However, when an individual decides the specific point along a continuum that accurately describes overall AR, a process of cognitive appraisal is engaged. This process, whether subconscious or conscious, integrates information from the internal and external environment to evaluate the current situation regarding its meaning for the individual’s survival and well-being (Ekkekakis 2003; Lazarus 1991).

Since the development of the Borg RPE Scale over 50 years ago (Borg 1962), perceived exertion has dominated the literature attempting to explain the subjective response to an exercise stimulus. As demonstrated by the Global Explanatory Model of Perceived Exertion in Chap. 2 (see Fig. 2.​1), RPE is a gestalt-like perceptual response that involves a complex pattern of physiological, psychosocial, performance-related and symptomatic process mediators. These mediators, functioning individually or collectively, ultimately shape the RPE response to exercise. Hardy and Rejeski (1989) cited Borg (1962) when they suggested “Because RPE represents a ‘gestalt’ of various sensations related to the stress and strain of physical work, it may not accurately reflect the affect a person feels during exercise.” In their example, two individuals performing exercise may give the same RPE for a given workload (e.g., 15 on the 6–20 Borg Scale). However, one individual may feel “good” while the other feels “bad” at the same level of exertion (Hardy and Rejeski 1989). Research has suggested that one’s overall AR during exercise, as well as specific affective states such as exercise enjoyment and social physique anxiety, can play an important role in regular PA participation, the adherence to an exercise program, and the potential of an individual to withdraw from participating in exercise (Ekkekakis and Lind 2006; Parfitt et al. 2006; Wankel 1993). Therefore, it is not only important to determine “what” one is feeling during exercise (i.e., RPE), but also “how” one feels during exercise (i.e., AR) (Hardy and Rejeski 1989).

In the early factor analyses of terms used to describe various affective states reported by subjects, researchers identified up to a dozen independent affective states (Nowlis and Nowlis 1956), including anger, anxiety, elation, sadness, and tension, to name a few. It was generally agreed that each of these affective states could be treated as separate dimensions, which served as the basis for early monopolar scales used to measure the degrees of these various states (Izard 1972; McNair et al. 1971; Nowlis 1965; Thayer 1967). Another school of thought was that these various self-reported affective states were related to one another and that this relation could be described using two bipolar scales that bisect one another: one ranging from pleasantness/pleasure to unpleasantness/displeasure; the other from attention/arousal to rejection (i.e., degrees of arousal) (Schlosberg 1952). This was termed as the circumplex model. Studies involving subject identification of facial expressions (Cliff and Young 1968; Royal and Hays 1959; Schlosberg 1952; Shepard 1962a, 1962b) and vocal expressions of emotions (Green and Cliff 1975) agreed with this circumplex model of affect. Whereas each observed affective state could be placed on the continuum of each of these bipolar scales, certain affective states were found to be better explained as a combination of these two bipolar scales (Russell and Pratt 1980). For example, one could feel excitement, an affective state being pleasant and having a high degree of arousal. Other examples include contentment (pleasant, low arousal), distress (unpleasant, high arousal), and depression (unpleasant, low arousal) (Russel 1980).

Affect includes numerous dimensions that can each be analyzed separately. In addition, the study of PA can include physiologically and psychologically diverse subjects performing exercise of varying intensity and modalities in diverse environments (Ekkekakis 2008). Therefore, numerous metrics have been used to study the relation between PA and AR. Two such metrics involving affective domains include the State-Trait Anxiety Inventory (Spielberger et al. 1970) and the Profile of Mood States (McNair et al. 1971). However, the use of these metrics has assumed that very specific affective domains can represent the whole affective spectrum as may be represented in a diverse group of individuals (Ekkekakis 2008). Other metrics developed specifically for the study of the PA–AR relation have defined, and therefore have limited, this relation to include specific affective domains. These include the Exercise-Induced Feeling Inventory (Gauvin and Rejeski 1993), the Subjective Exercise Experiences Scale (McAuley and Courneya 1994), and their derivations (Annesi 2006; Lox et al. 2000; Rejeski et al. 1999). However, these metrics were developed in healthy, physically active individuals and may not represent the affective experience of other subject populations (Ekkekakis 2008), especially those who have significant barriers to the adoption and maintenance of regular PA. Ekkekakis (2008) described this effect as domain under-representation, which has been demonstrated in sedentary adults (Gauvin et al. 1997) and older adults with osteoarthritis (Focht et al. 2004).

Other limitations of scales used to measure the PA–AR relation is that they are often difficult for the subject to understand and difficult to use during the actual exercise performance. It has been suggested that the average person’s understanding of affective states may be overly simplified if not mistaken (Russel 1980). However, from a public health perspective, it is important to understand an individual’s decision making process regarding PA participation. Therefore, PA research should employ an easy-to-use AR metric that can accurately describe the overall feelings of a wide range of individuals. Many AR scales involve questionnaires with numerous test items each targeting specific affective domains. These questionnaires cannot be used effectively to assess a subject’s AR during acute PA. This is especially the case during high intensity exercise when one’s affective state can change rapidly due to increases in the intensity of interoceptive physiological cues originating from exercising muscle and cardiorespiratory responses (Ekkekakis 2003). Therefore, to measure AR during acute exercise, it is necessary to have a metric that can be used quickly by subjects. Therefore, it would be beneficial to have affect rating scales that assess the overall AR rather than specific affective domains during an acute bout of exercise. This type of rating scale could be similar to an RPE scale. The overall AR is formed by incorporating all of the various affective domains important to that individual. This type of scale allows individuals to interpret their own subjective feelings about the exercise experience using their own unique thought processes. In addition, rating the overall AR may provide a construct that is easier to understand for the average individual, especially a child, rather than measuring an entire set of affective domains.

Rejeski and colleagues (1987) developed the Feeling Scale (FS, Fig. 4.1) to measure AR during exercise. The scale was intended to be easy to use and easily understandable. The FS is not intended to measure various categories or specific domains of affect because it was believed that the strongest aspect of AR is the initial determination of good or bad, a more global response (Hardy and Rejeski 1989; Weiner et al. 1979). The FS was based on one of the core bipolar constructs of Russell’s (1980) circumplex model that could be used to differentiate between feelings along the continuum of core emotions ranging from pleasantness/pleasure to unpleasantness/displeasure (Frijda 1988). The FS is an 11-point bipolar (i.e., valenced) metric with numerical categories ranging from −5 to 5. Verbal descriptors were situated in juxtaposition to each odd integer, with numerical categories ranging from “very bad” (at −5), representing maximal displeasure/unpleasantness, to “neutral” (at 0), to “very good” (at +5) representing maximal pleasure/pleasantness (Rejeski et al. 1987).

The FS has demonstrated face, content and construct validity (Hardy and Rejeski 1989; Rejeski et al. 1987) and has been widely used to measure AR in various subject populations performing a variety of exercise situations (Ekkekakis et al. 2000; Ekkekakis et al. 2004; Ekkekakis and Petruzzello 2002; Lind et al. 2008; Parfitt et al. 2006). This line of research has indicated that, although affective states are considered as a psychosocial correlate of exertional perceptions, AR measured using the FS and RPE are not isomorphic constructs. When subjects were asked to rate both AR and RPE during cycle ergometer exercise bouts of various workloads (30, 60, and 90 % of peak VO₂), the constructs were only moderately inversely correlated, with resultant coefficients ranging from r = −0.33 to −0.55 (Hardy and Rejeski 1989).

The FS was based on the bipolar construct of the circumplex model. This model differentiates between feelings distributed along the affective continuum from pleasantness/pleasure to unpleasantness/displeasure (Russel 1980). Therefore, consideration has been given to the bipolar construct of the circumplex model that bisects pleasure–displeasure during exercise, and analogously differentiates between feelings along the affective continuum from low to high arousal (or activation). The single-item Felt Arousal Scale (FAS), developed by Svebak and Murgatroyd (1985), is a 6-point scale that has been used to measure the degree of perceived activation associated with exercise (Kerr and Vlaswinkel 1993; Kerr and van den Wollenberg 1997; Hall et al. 2002). Previous research has measured the AR to exercise using both the FS and FAS to investigate the application of the circumplex model in explaining exercise behavior (Hall et al. 2002). Evidence suggests that FS ratings of AR that distribute along the pleasure–displeasure continuum have a strong practical application to exercise intensity prescription and PA program adherence. The application of ratings of perceived activation using the FAS requires further investigation. As such, this manual will only focus its discussion of AR as measured using the FS. The assumption here is that the FS provides a global AR measurement indicating good or bad feelings during exercise (Hardy and Rejeski 1989).

Ekkekakis (2003) proposed the “dual-mode model” to explain inter-individual differences in AR during exercise of varying intensities. As described by Parfitt and colleagues (2006), the model “is based upon the interplay of relevant cognitive processes and interoceptive cues prior to and following the transition from aerobic to anaerobic metabolism”, namely, the anaerobic threshold (AT) as identified using the ventilatory threshold (VT) or lactate threshold (LT). At exercise intensities below the AT, when the energy metabolism is supported by aerobic pathways, the acute AR is primarily influenced by cognitive processes such as appraisal, self-efficacy, and social context. These cognitive processes are shaped by personal experience, individual personality variables, personal goal achievement, etc. All of these factors are unique to the individual (Ekkekakis 2003). Thus, there may be heterogeneity in AR at intensities below the AT due to the inter-individual differences in interpretation of the exercise (Rose and Parfitt 2007). At exercise intensities above the AT, when energy metabolism is supported by ever-increasing anaerobic sources, lactacidemia in active muscle and decreasing tissue and blood pH become significant mediators of the acute AR. In addition, AR is influenced by interoceptive cues from baroreceptors, thermoreceptors, and visceroreceptors in the heart and lungs (Rose and Parfitt 2007). Thus, there would be less inter-individual variability in AR during exercise at intensities above the AT because the affect experience is shaped less by cognitive processes and more by physiological cues that disrupt metabolic homeostasis (Ekkekakis 2003; Parfitt et al. 2006).

Research has supported the “dual-mode model” in that the acute AR to exercise declined once the intensity exceeded the AT (Ekkekakis et al. 2005; Hall et al. 2002). It was shown that, while 47 % of subjects exhibited a decline in AR during 15 min of treadmill exercise intensities below their VT, 80 % of subjects exhibited a decline in affect during similar exercise at intensities above their VT (Ekkekakis et al. 2005). Similar results were reported by Parfitt and colleagues (2006) when subjects performed 20 min of treadmill exercise. AR was more positive and stable when subjects performed intensities below their LT, with only 25 % of subjects exhibiting a decline in AR during exercise. When subjects performed intensities above their LT, AR became increasingly more negative with 83 % of subjects exhibiting a decline in AR during exercise (Parfitt et al. 2006).