Previous - Next
Abstract
1. The concept of temperament
2. Activity level as a dimension of temperament
3. The behavioral assessment of activity level
4. The role of activity in studies of energy expenditure/energy requirements
5. Activity level and information processing
6. Activity level and the influence of the child's psychosocial environment
7. Conclusions
Acknowledgements
References
T.D. WACHS *
* Department of Psychological
Sciences, Purdue University, West Lafayette, IN 47907, U.S.A.
Temperament is traditionally defined as the individual's behavioral 'style' - the way in which behavior is expressed. This behavioral style appears early in life, has a biological base, and is at least moderately stable across time and situations. Activity level is viewed as one of the dimensions in most current theories of temperament.
At a behavioral level, activity is typically assessed by temperament questionnaires, direct observations or objective measurement procedures (e.g., actometers). While improvements have been made, psychometric problems with existing temperament questionnaires still remain. Both direct observations and objective measurement procedures tend to be reliable only when multiple data points are utilized. Under these circumstances, the most appropriate approach is to utilize multiple measures of activity.
Behaviorally-oriented researchers would like to know more about the role of activity in influencing behavioral development. This requires more than just measuring activity level. For example, underlying many activity studies is the assumption that highly active children are exposing themselves to more developmentally facilitative, varied stimulation than inactive children. However, available evidence indicates that high activity level may reflect either stimulus seeking or stimulus avoidance behavior. What this means in practice is that, in addition to measuring activity level, it is equally important to measure what the child is doing while being active or inactive. Similarly, recent research has indicated that less active children may have greater need of stimulation by caregivers than more active children. If children who are low in activity due to low energy intake are reared by caregivers who also have a low energy intake, these caregivers may be less able to provide the extra stimulation their children need. However, it is impossible to know whether this synergistic model applies unless we measure both the child's activity level, and the characteristic caregiver behaviors experienced by the child.
The interrelation of energy intake, energy requirements and activity must be looked at within a multidimensional framework, emphasizing the process wherein the child's energy intake, energy expenditure, activity level, behavior and environment relate to specific developmental outcomes.
The focus of the present paper will
be on the relevance of temperamental differences in activity level for the behavioral
development of infants and children. The main thesis of this paper is that questions
concerning the influence of energy expenditure and individual differences in activity
level upon behavioral development cannot be satisfactorily answered, unless we take
account of the functional meaning of differences in activity level. However, prior to
getting at this primary issue, some ground work needs to be done in terms of defining what
is meant by temperament, where activity level fits into the construct of temperament and
how behavioral researchers measure temperamentally-based activity level.
Although the concept of temperament, in a form we would recognize today, can be traced back to the writings of Galen in the 2nd century A.D. (IRWIN, 1947), the scientific study of the relevance of temperament for children's development really began only about 35 years ago with the initial reports of Thomas and Chess (e.g., THOMAS et al., 1963). Given the short history of the scientific study of temperament, it is not surprising that there is no formal, agreed upon definition of this construct. Rather, there are a variety of definitions, most of which reflect the theoretical orientation of the researchers studying temperament (e.g., GOLDSMITH et al., 1987) While there are multiple definitions of temperament, this does not mean the situation is hopeless. As McCall (GOLDSMITH et al., 1987) has pointed out, we still do not have a formal, agreed upon definition of intelligence, and yet the construct of intelligence has proven to be very useful, both as a scientific and as an applied tool. What we have for intelligence, and what we also have for temperament, instead of a formal definition, is a working definition an agreement among active researchers about what the construct of temperament includes, as well as what it excludes. Within the framework of a working definition there is general agreement among the majority of temperament researchers on the following definitional aspects of temperament (e.g., BATES, 1989a; GOLDSMITH et al., 1987; STRELAU, 1987):
1. Temperament is not a thing or a series of things. Rather it is a hypothetical organizing construct we use to label and define a complex set of processes.
2. The processes we call temperament refer to the way in which behavior is expressed, independent of the content of behavior or the motivation for behavior. Using Thomas and Chess's term, temperament refers to the how and not the what of behavior. For example, infant crying in and of itself is not an aspect of temperament. Rather, how intensely and how frequently the infant cries, or how easily it is consoled when crying, are the behavioral measures that define temperament.
3. The behaviors defining temperament appear very early in life, certainly within the first 2 months after birth.
4. Temperament is biologically based. This does not necessarily mean that temperament is inherited. Indeed, the question of whether heritability should be part of the definition of temperament is one of the major points of disagreement among theoreticians today (GOLDSMITH et al., 1987). However, there is clear agreement that temperament has a strong biological-constitutional basis. The expression of temperament, however, can be modified by the experiences encountered by the child (MATHENY, WILSON and THOBEN, 1987; WACHS, 1988).
5. Temperament is moderately stable across time and over situations. While stable, temperament is not necessarily invariant across time or situations. The expression of temperament can be influenced by biological, maturational (ROTHBART and DERRYBERRY, 1981) and contextual factors (THOMAS and CHESS, 1977).
In addition to the above definitional criteria it is also important to recognize that temperament is a multilevel construct. AS BATES (1989b) has suggested, temperament can be expressed and studied at three levels. The first is the behavioral level - the how of behavior, centering around such dimensions as intensity, activity, persistence or emotionality. This is the level that has been most often studied over the past 25 years. The second level, the neurological level, refers to the individual neurological underpinnings of the behavioral expression of temperament. This level refers to individual differences in the patterning of central nervous system reactivity to environmental stimulation. Examples in this area are perhaps best seen in the work of researchers with a Pavlovian orientation such as STRELAU (1983). The third level refers to the constitutional basis of temperament, especially genetic influences, though it may also include factors like prenatal nutrition of the fetus, as it affects the development of the central nervous system (level 2). Except for genetic research (e.g., BUSS and PLOMIN, 1984; WILSON and MATHENY, 1986) there has been relatively little study of temperament as a construct at this level. 1
One potential biologically-based approach to the study of temperament defined at level 3 would involve the use of doubly-labelled water to assess individual energy expenditure levels (HORTON, 1984). Specifically, it would be of interest to determine if there were relations between biological assessments of energy expenditures assessed with doubly-labelled water, and behavioral measures of temperamentally-based activity level.1
BATES' (1989a) fundamental point is
that, both conceptually and methodologically, it is critical to specify exactly what level
of temperament the researcher is utilizing. Although interrelated, each level is unique.
Thus, although the behavioral processes defined at level 1 must be consistent with the
processes encompassing temperament at the neurological and constitutional levels, it would
be a form of naive reductionism (ANDERSON, 1972) to reduce the behavioral processes of
level 1 down to the neurological or the constitutional processes of levels 2 and 3. Each
level has unique processes and requires unique methodologies for study. In terms of what
follows, in my discussion of activity, I will be dealing with activity primarily at a
behavioral level (level 1).
Within a level 1 temperamental
framework, activity level has traditionally been defined either as the degree of energy
expenditure through movement (EATON and ENNS, 1986) or the frequency and intensity of
motor activity (BATES, 1987). This definition is not very different from that in energy
expenditure studies, which often refer to the intensity or duration of physical
activities. If we look at the five major models of temperament which currently exist,
activity level forms an essential part of two of them (the THOMAS and CHESS (1987) model;
the Buss and PLOMIN (1984) model). In the GOLDSMITH and CAMPOS (1982) model, which defines
temperament on the basis of emotional reactivity, activity level is included as an index
of emotional arousal. In the ROTHBART and DERRYBERRY (1981) model, which defines
temperament on the basis of self-regulation activities, activity level, while measured,
does not play a major part in the theory itself. In Eastern European models based on
Pavlovian theory, activity level is often viewed as a means of discharging or augmenting
central nervous system arousal level (STRELAU, 1983); however, in general, the concept of
activity is confounded with and often subsumed under reactivity (STRELAU, 1986). Thus, all
of the available behavioral models of temperament do include activity level as a
dimension, although the activity level construct is central to some theories and only
incidental to others.
There are a variety of methods to assess temperamental activity at the behavioral level. 2 Although measurement of activity level is not a prime focus of this paper, a brief discussion of the strength and weaknesses of each method does seem appropriate.
There are also a variety of physiologically-based methods, but since these are not directly relevant to this paper they will not be reviewed here. Readers interested in this topic can refer to chapters by HORTON, BRUN and TORUN in Volume 11 of Current Topics in Nutrition and Disease, edited by POLLITT and AMANTE (1984).2
At a behavioral level the most common approach for assessing temperamental activity level is the use of temperament questionnaires, in which the parent is asked to rate the child's behavior across a variety of dimensions, one of which typically involves activity. A recent review by my research group (SLABACH, MORROW and WACHS, in press) indicated that there are 16 temperament questionnaires which are currently utilized. The use of questionnaire assessment of temperament has been very controversial, with earlier reviews critically questioning many of the psychometric qualities of temperament questionnaires (e.g., HUBERT et al., 1982). There has also been concern that temperament questionnaires may tell us more about parent characteristics or parent test-taking style than about the nature of the child's temperament (SAMEROFF, SEIFER and ELIAS, 1982; VAUGHN et al., 1987). However, the most recent review of the psychometric qualities of available temperament instruments (SLABACH et al., in press) has concluded that many of the psychometric problems previously associated with temperament questionnaires have been satisfactorily resolved. In terms of the second problem, namely what temperament questionnaires tell us about the child, the consensus of current opinion is that, while temperament instruments contain both objective and subjective factors, the influence of subjective factors (maternal characteristics) does not outweigh that of objective factors (child characteristic) (BATES and BAYLES, 1984; BATES, 1987).
Across existing instruments the activity dimension is psychometrically among the strongest (SLABACH et al., in press). A representative sample of activity items from the Toddler Temperament Scale, one of the most commonly used measures of temperament, is shown in Table 1.
Table 1. Sample activity items from the toddler temperament scale
All items are rated by the caregiver on a 6-point scale from almost never to almost always The child fidgets during quiet activities (story
telling, looking at pictures). |
While temperament questionnaires appear to be a very promising method for quickly assessing the child's activity level, some care must be taken when using them in energy studies in nonwestern settings. Although, modified versions of standard temperament questionnaires have been successfully utilized in a variety of cultures, including both Asian (HSU et al., 1981) and African countries (DEVRIES, 1984), this does not mean that a simple translation of temperament questionnaire items into a different language will guarantee validity. All too often, cultural constraints may influence the nature of ratings (BATES, 1987). For example, because of different cultural patterns, different criteria are used by parents in Kenya and America when characterizing their infant as temperamentally difficult (SUPER and HARKNESS, 1982). Thus, while temperament questionnaires do have the potential to provide appropriate measurements of activity level crossculturally, they should not be used without considering whether cultural factors will affect scores (for more detail on this question see the chapter by Super in this volume).
A second approach is direct observation of the child's behavior in naturalistic settings, such as the child's home environment. In this procedure, observers typically watch the child for periods of several hours and either record the occurence of specific child behaviors, using a preset behavioral code (EATON, ENNS and PRESSE, 1987), or rate the child's temperament across a variety of dimensions after the observation is completed, using one of the standard temperament questionnaires (ROTHBART and GOLDSMITH, 1985; GOLDSMITH and RIESER-DANNER, in press). Observational studies of temperament have the advantage of being highly flexible and applicable across a variety of settings. However, available research suggests that it is essential to utilize multiple observations whenever possible, both as a means of increasing the stability of findings, and as a way of minimizing non-representative responding due to the presence of an observer (GOLDSMITH and RIESER-DANNER, in press; WACHS, 1987a). In addition, in naturalistic observation procedures it is difficult to differentiate between spontaneously occurring child behavior, and the child's reaction to covert or overt signals from the caregiver (GOLDSMITH and RISER-DANNER, in press). In addition, the observers' knowing such factors as the sex of the child, or believing certain stereotypes about how particular children should behave (e.g., children who look malnourished should be less active) may result in reliable ratings that are, nonetheless, unrepresentative reflections of the child's individual characteristics (EATON and ENNS, 1986).
A variant of direct observation are laboratory-based assessments. Typically, in this type of procedure, the child is presented with a series of structured situations (peak-a-boo games, being talked to, interruption of feeding) in a laboratory setting. The child's behavior is videotaped, and the videotapes are then rated on a number of dimensions, one of which typically involves activity level. At present there are two fully developed laboratory assessment procedures which involve the assessment of activity (GOLDSMITH and ROTHBART, 1988; MATHENY WILSON and THOBEN, 1987). The major advantage of laboratory-based assessment procedures is that they permit precise assessment of behavior, and allow for tight control of extraneous variables. However, while laboratory ratings of activity can be reliably scored, and do show a moderate degree of agreement with parental ratings (GOLDSMITH and RISER-DANNER, in press), for the most part the degree of psychometric adequacy of these procedures is still open to question. In addition, although the tasks children perform in the laboratory situation are relatively simple, the requirement for videotaping or precise time control may reduce the usability of these procedures in less developed countries, where necessary equipment may not be easily available. It is possible to utilize structured task situations as a means of obtaining data on temperamental activity level in field situations, without some of the constraints of a laboratory setting. For example, the open field activity test utilized by Chavez and his colleagues (CHAVEZ and MARTINEZ, 1985) is one example of an approach to assessing activity level using a structured task in a naturalistic setting. However, while more usable in naturalistic settings, the open field measure suffers from some of the same previously discussed problems associated with other naturalistic observational methods.
The final class of behavioral measures includes objective, mechanical assessments of activity level. Devices such as actometers are examples of instruments used to collect this type of measure. They have the advantage of being easily applicable to a wide range of subjects and to a wide range of contexts (EATON, 1983; EATON and DURESKI, 1986; HALVARSON and POST-GORDON, 1984). But, it is important to remember that objective measures of activity do not automatically result in greater predictive validity than more traditional measures, such as temperament questionnaires (EATON and ENNS, 1986).
The basic point to be made about measurement of temperamental activity is that there are a variety of measures available, all of which have some strengths and some drawbacks. Probably the strongest measurement strategy is to use multiple measures of activity, including questionnaires, objective measures such as actometers and, if possible, laboratory or direct observational methods. Assuming that there is some degree of correlation across these multiple measures, aggregation across measures should lead to more stable and representative activity scores than if just a single measurement technique were utilized (EPSTEIN, 1983).
However, from a behavioral point of
view, the critical problem is not our ability to measure the child's activity. As the
above brief review indicates, there are a number of different types of measures which can
provide satisfactory measurement of the child's activity. Rather, from a behavioral point
of view, there is a somewhat paradoxical problem, namely that the ease of behavioral
measurement of activity may conceal a more fundamental difficulty with the use of activity
measures in studies of energy intake and energy expenditure.
Traditionally, activity level has been a popular outcome measure in energy expenditure and energy requirement studies (POLLITT and AMANTE, 1984). In part, this is not only due to the ease with which activity can be measured, but also to the fact that the child's activity will influence it's energy expenditure, which in turn will influence the balance between energy expenditure and energy requirements. Thus, many studies utilize activity level as a means of aiding in the definition of energy expenditure, and in the definition of energy requirements for a population (e.g., REINA and SPURR, 1984; TORUN, 1984). When activity level is used in this way, with no implication that differences in activity level have functional consequences for differences in behavioral development, then the problem primarily is one of choosing the most appropriate behavioral measures of activity. However, in other studies activity level is used less as a pure dependent variable, and more as a mediating variable. In this latter group of studies it is assumed either indirectly (MALINA, 1984), or directly (BARRETT, RADKE-YARROW and KLEIN, 1982; BARRETT and RADKE-YARROW, 1985), that activity level is an essential part of the process wherein inadequate energy intake relates to subsequent decrements in behavioral development. In these studies it is typically assumed that, if energy intake is insufficient to meet energy requirements, there will be a decrement in activity level. This assumption can easily be viewed as an example of a level 3 (constitutional) approach to temperament (BATES, 1989a). This reduction in activity level is thought to have certain consequences, such as less exploration of the environment, less active processing of information by children, and less adequate patterns of caregiver-child transactions. These consequences, in turn, are assumed to result in subsequent behavioral deficits.
It is this latter usage of activity level that is of concern, particularly when activity level is viewed within the framework of temperament theory and empirical research on temperament. Within a temperamentally-based view of activity it is doubtful if we can understand the relevance of activity for behavioral development, just by measuring activity. Rather, to understand the role of temperamental activity level in the developmental process, we also need measures of the child's environment, the child's characteristics, the parents' characteristics and the social context in which the child is functioning.
Please note that I am not
questioning the fact that an imbalance between energy intake and energy requirements leads
to reductions in measured activity level; ample evidence is available supporting this
proposition (RUSH, 1984); BARRETT and RADKE-YARROW, 1985). Similarly, I am not questioning
the fact that inadequate caregiver-child interactions or inhibition of exploration will
adversely affect behavioral development. Again, ample evidence is available to support
this proposition (WACHS and GRUEN, 1982). What I am questioning is the assumption that
simply measuring activity level per se will tell us something about the nature of the
child's information processing, or the process whereby variability in activity level
relates to variability in developmental outcomes. As a means of underscoring this point I
will present some theoretical statements and some empirical research from the temperament
(and other) literature involving activity, both as a dependent variable and as a potential
mediator of environmental influences.
Underlying the use of activity level as a measure in studies of energy expenditure and energy requirements is the assumption that children who are highly active are exploring more, and thus encountering more varied stimulation than children who are inactive (CHAVEZ and MARTINEZ, 1984). This assumption incorporates the idea that not only are more active children exploring more stimuli, but also that these children are more actively processing incoming stimulation. Is there evidence to support these assumptions? Certainly in many early theoretical discussions activity level was seen as a fundamental component of exploration (BERLYNE, 1960). In addition, in many of the early animal studies of exploration, activity level was often viewed as a proxy for exploratory behavior, as in studies of open field 'exploration' (DENENBERG, 1967, 1969; GOODRICK, 1971). However, questions quickly began to be raised about equating activity level with exploratory behavior, in part because of increasing evidence showing that, in animals, activity level was encompassing a variety of other dimensions besides exploratory behavior (WHIMBY and DENENBERG, 1967).
Similar doubts also began to be raised at the human level, based on evidence indicating that temperamentally-based measures of activity level do not load on the same factor as measures of attention span and orientation (MATHENY, BROWN and THOBEN, 1971). It was this state of affairs that led one group of researchers (RHEINGOLD and ECKERMAN, 1970) to conclude: 'Exploratory behavior has proved a troublesome class of behavior conceptually... (is the organism exploring or just active?)". Given this concern it is perhaps not surprising that in more recent studies of exploration in humans, quantitative measures of activity level play no part in defining this construct (e.g., BELSKY and MOST, 1981; POWER, CHAPIESKI and McGRATH, 1985).
The fact that activity and exploration are not necessarily isomorphic does not, however, preclude the possibility that more active organisms may still be exploring more, and thus processing more information than less active organisms. However, when we look at this question more closely, it quickly becomes apparent that we cannot arrive at a satisfactory understanding of the relation between activity level and exploratory behavior or stimulus processing, without taking into account the characteristics of the organism. Particularly within the framework of activity level as a construct of temperament, one of the leading European theoreticians in this area, Jan STRELAU (1986), has pointed out that activity level can reflect either stimulus seeking or stimulus avoidance behavior, depending upon the individual's characteristic level of arousal. For an underaroused child, high activity level may be a form of stimulus seeking, whereas for an overaroused child, high activity level may be a form of stimulus avoidance. Without knowledge about the purpose of the child's activity, it is difficult to establish whether high levels of activity reflect active stimulus processing or active avoidance of stimulus processing. Thus, the same level of activity may have a totally different meaning, depending upon organismic characteristics. Although Strelau was speaking theoretically, examples are readily available to support his point of view.
One line of evidence comes from studies investigating the impact of social stimulation on the behavior of infants at risk. Given that preterm infants may be less responsive to parental stimulation, one of the critical goals in this area is the development of environmental manipulations that maximize preterm infants' responsivity. For example, tactile stimulation has been demonstrated to increase environmental orientation of preterm infants (SCAFIDI et al., 1986). 3 However, when very low birth weight preterms are exposed to tactile stimulation, while there is an increase in activity level, the higher activity level for this population suggests behavioral disorganization rather than stimulus processing activities, as seen in the increased number of behavioral cues suggesting stimulus avoidance rather than stimulus processing activities (OEHLER, ECKERMAN and WILSON, 1988). In contrast, infants at lower risk, while showing less motor activity, showed more visually attentive behavior following stimulation. Thus, for one group of infants increased activity was associated with stimulus avoidance behavior, whereas for a second group of infants there was less activity, but there appeared to be more information processing.
Interestingly, even though the tactually-stimulated and control group did not differ in the amount of caloric intake, there was greater weight gain in the stimulated infants. One hypothesis is that increased metabolic efficiency may be associated with increased activity level (SCAFIDI et al., 1986).3
At an older age, a similar dichotomy can be seen in studies with developmentally disabled children. Specifically, ZENTALL and ZENTALL (1983) have argued that the activity level of autistic children, who appear to be chronically overaroused, is based on a need to avoid stimulation, whereas the activity level of hyperactive children, who are chronically underaroused, appears to be based on a need to obtain increased stimulation.
The problem of activity level as a proxy for information processing becomes even more complex when we consider the possible influence of environmental context. Again, using temperament theory as a guide, within the framework of developmental-contextual approaches to the study of temperament (e.g., LERNER and LERNER, 1983), the significance of temperament for development is not in the specific attributes the individual possesses, but rather in the degree of fit of temperament to environmental context. In the area of personality, KENRICK and FUNDER (1988) have expressed the same difficulty in understanding the meaning of behaviors without also understanding the context in which these behaviors are occurring. Again, examples are available to support this viewpoint.
Although I know of no direct evidence relating context to activity in humans, across a number of areas the importance of contextual factors for understanding activity can be seen indirectly. For example, over the past 10 years Kagan and his colleagues (KAGAN, REZNICK and SNIDMAN, 1984; KAGAN et al., 1988) have been investigating the developmental course of temperamentally-inhibited behavior in infants and children.
While behavioral inhibition is not the same thing as activity level, as Kagan and his colleagues have clearly shown, in unfamiliar situations highly inhibited children remain extremely quiet, preferring to passively watch rather than actively interact with unfamiliar adults, unfamiliar peers or novel objects (KAGAN et al., 1986, 1988). "The words restrained, watchful and gentle capture the essence of the inhibited child, free, energetic and spontaneous capture the style of the uninhibited youngster" (KAGAN et al., 1986, p. 54). Thus, in unfamiliar situations, even though temperamental inhibition is not the same as temperamental activity level, inhibited children should be less active than uninhibited children. These group differences in activity level should be less likely to occur when inhibited children are interacting with familiar peers, familiar adults, or are in a familiar situation. Thus, in some situations, children characterized as low in activity may, in fact, be displaying contextually-driven inhibited behavior rather than low activity level. Inhibition and activity can only be distinguished in relation to the context the child is in. Moving from a familiar to an unfamiliar context affects inhibited behavior, but should have no effect if what we observe is simply the activity level of the child.
Another contextual aspect of activity level can be seen in regard to the question of when high activity level is a good thing. In the majority of studies in the energy expenditure and energy requirement literature, the implication seems to be that higher levels of activity are desirable, since this suggests that the child is in energy balance. However, high activity level may not be particularly appropriate for all types of situations; indeed, in some situations low activity level may be an indicator of both more advanced developmental status and better future developmental potential than high activity level. For example, we know from studies of temperament involving school-age children that, in the classroom, high activity levels are associated with higher levels of distractibility (PAGET, NAGLE and MARTIN, 1984) and lower academic achievement (PALISIN, 1986). Given this background we should not be surprised at the results from the Kenya project, relating chronic low energy intake to school-age children's behavior (SIGMAN et al., 1989). One of the outcome measures in this project was the amount of time the child was off-task in the classroom. Off-task behaviors are typically characterized by higher activity levels (fidgeting, playing with objects, looking around the class) than on-task behaviors. As reported by SIGMAN et al.. (1989), mildly malnourished females were more off-task than adequately nourished females, which means that, in the classroom situation, well-nourished females probably showed lower activity level than malnourished females.
Given the demonstrated importance of context, it is noteworthy that one of the major models of temperament (ROTHBART and DERRYBERRY, 1981) defines temperament in terms of the individual's ability to self-regulate and modulate reactivity to the environment. Applying this notion to the study of activity level suggests that what may be particularly critical is not so much the child's activity level per se, but rather the child's ability to modulate activity level, depending upon the demands of the circumstances it encounters. In some situations (such as free play with peers), high activity levels may be more appropriate, whereas in other situations (such as the classroom), low activity levels may be more appropriate.
We intend to test this hypothesis within the next year, using data from our Egyptian project on the impact of chronically low energy intake upon children's development. Specifically, we have measures of children's activity level both in the classroom and on the playground. If the above hypothesis is correct, we would expect higher energy intake to be associated with higher activity levels on the playground and lower activity levels in the classroom. We would also predict that more advanced development would be shown by children who are more active on the playground but less active in the classroom, as opposed to children who are active or inactive in both contexts. The critical point is that these types of predictions cannot be made without taking into account contextual factors, as well as activity levels.
What the above suggests is that the meaning of activity depends upon the child's goals, the child's characteristics and the social context the child is operating in. What this means in practice is that to understand the functional consequences of activity levels for development, particularly in relation to energy expenditure and energy requirements, it is important to measure what the child is doing while being active, as well as where the child is being active or inactive. For example, is the child actively engaged in interacting with stimuli in the environment, or is the child aimlessly moving with little attempt to actively engage objects or persons in the environment? Similarly, it is important to assess whether the child's activity level changes as a function of contextual demands. These types of distinctions are potentially measurable through direct observation. Indeed, we have made some of these distinctions and used these types of differential measurements in our Egyptian data base. I would maintain that the developmentally most appropriate measures of activity in studies of energy expenditure and energy requirements will be those that are sensitive to context effects.