Theoretical orientation

Existing lexicons have applied many different approaches to classifying the states of humans, usually without thoroughly acknowledging the implicit theoretical orientations that underlie the definitions. We here identify three orientations that are used to differentiate between categories. Existing lexicons sometimes use one orientation to define some categories, while using another orientation to define others. Although the inclusion of inconsistent orientations may be acceptable within the context for which an individual lexicon was created, it creates problems for comparing across contexts, and makes it very difficult to apply parallel lexicons in an exclusive and exhaustive manner.

1. Social motivation. We use this term to refer to the diverse ways by which social constructs and cultural rules motivate the undertaking of an activity. For example, ‘paid work’ generally includes activities that are undertaken for the benefit of someone else, who is usually not a friend or family member, through a monetized exchange. ‘Unpaid work’, in contrast, is generally undertaken for the benefit of oneself or one’s family. Meanwhile, the activities of ‘volunteer work’ could be identical to those in paid work but without the monetary exchange. This type of social motivation description underpins the first-order divisions often used in economics and international development, such as the System of National Accounts [32], which starts by separating market from non-market activities, regardless of what is actually being done. For example, childcare will be classified differently whether one is taking care of their child at home (unpaid work), in a daycare (paid work) or if the child is a neighbour (volunteering). As a result, it can be awkward to compare societies with different social and economic systems using the SNA [33]. Differentiating based on social roles also frequently identifies a provider and a recipient, such as a teacher and student, even if the participation of both is essential for the activity to occur. This can lead to subdividing what are essentially multi-participant activities based on whom is being paid. Social motivations are not limited to monetized relationships such as these, but also extend to religious and other cultural mechanisms.
2. Context. This type of description focuses on features of the environment that shape the human experience, often including who else was present, the natural surroundings, or the human-made tools or equipment used during the activity. Common examples in time use surveys include ‘spending time with family’, ‘watching tv’, ‘using a computer’ and ‘reading a book’. Technology-based descriptions can be complicated by the potential use of multiple technologies at the same time, such as listening to recorded music while traveling by train, and are often only relevant to narrow periods of history given technological changes over time.
3. Outcome. These descriptions focus on the outcome of an activity, rather than the contextual state, usually in terms of changes in the objective world (which can include the physical outcomes for humans). Examples found in time use surveys include ‘preparing food’, ‘cutting wood’, ‘washing dishes’, ‘providing medical care’. These descriptions can become confused given multiple simultaneous outcomes, for example riding a bicycle can change someone’s location, provide exercise, and generate an enjoyable experience at the same time.

None of these three theoretical orientations can be thought of as the ‘best’ one. Rather, each can be useful for capturing particular dimensions of a human state, and is well-suited to its own set of research and policy questions. But the use of different orientations leads to incompatibility between dataset and, further complicating the issue, many existing lexicons combine these three types of descriptions, particularly in government time use surveys (e.g. Table 1). We argue that the consistent use of a single type of theoretical orientation within a lexicon limits arbitrariness in activity definitions, decreases ambiguity in interpretation, and helps to ensure exhaustiveness and exclusivity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Inconsistent theoretical orientations of categories in an existing lexicon.

Five examples are drawn from the International Classification of Activities for Time Use Statistics (ICATUS). For each, we indicate the implied theoretical orientation(s).

https://doi.org/10.1371/journal.pone.0270583.t001

Multitasking and aggregation

People frequently describe themselves as doing more than one thing at a time, such as ‘minding children while cooking’ [34]. This might suggest that people can violate the 24-hour limitation in terms of how states can be categorized.

Yet, physical outcomes that can be achieved with the body at one time are typically described in singular terms, given the limitations that a person can only be in one place at a time, and that physical tasks tend to be limited by the ability to coordinate a set of manipulations. The latter is closely related to the limitation of the conscious mind to focus on one thing at a time, due to neurological constraints on working memory allocation [35] and attention limitation [36]. In fact, most instances of ‘multitasking’ can be considered as rapid switching between micro-tasks, as part of a continuum [37]. Consequently, it could be feasible to define categories in physical terms for which multitasking could be considered negligible, at the individual level.

Despite the minor importance of multitasking for an individual on short timescales, an observer could choose to describe activities at multiple scales, which reflect different levels of aggregation over time and/or aggregation over social units (multiple people). For example, the activities of a single person could be described in terms of their physical manifestation on a second-to-second basis. This type of approach has been applied in studies of everyday life for disability and robotics research, where individual physical movements are recorded by cameras and hierarchically coded [38]. This description is relatively unambiguous, physically-oriented and highly-detailed, and would capture sequential activities that might otherwise masquerade as true multitasking. Thus, a sequence of activities might include opening the refrigerator, pulling open a drawer, extracting a bag of carrots, taking out two carrots, closing the refrigerator, carrying the carrots to the counter, opening the drawer, extracting the peeler, etc. Alternatively, the same activity could be described at a temporally aggregated level (e.g. a 15-minute interval timescale) as part of a composite activity, ‘preparing dinner’. The importance of temporal scale becomes evident when one considers that the first scheme might never explicitly capture the fact that the overall intended outcome was to make dinner, while the second would lack all detail on what kind of dinner was made and what steps were required to do so.

Social aggregation also frequently appears within activity descriptions when they are used to capture composite activities that are distributed among people. For example, a food packing plant might have employees driving forklifts, moving bags of food, operating packaging equipment, organizing sales and purchasing, and managing personnel. These activities could be described in an aggregated sense as all being part of food packing, and indeed would typically be recorded at this level of aggregation in economic data. There is no objectively-correct level of aggregation, so the level of aggregation must be strategically chosen for pragmatic and/or theoretical reasons, and must be borne in mind when interpreting results. In future, one might conceive of including additional measures to quantify the social aggregation of activities as a parallel lexicon.

High-level MOOGAL categories

To provide an overview of the MOOGAL (version 1), the eight high-level categories are described according to their outcomes on the six state variable classes illustrated in Fig 1. The six state variable classes are designed to provide a simple conceptual superstructure for the state of the entire human-Earth system, and within which the time allocation to activities plays a central role as a determinant of rates of change, the physical outputs of activities (e.g. Eq 1). The framework explicitly considers neural structures as the physical basis of mental attributes—such as knowledge, beliefs and values—an uncommon perspective in social sciences, but one which has growing empirical support [42–44]. Similarly, experience is considered, in physical terms, as the stimulation of existing neural structures (including associated emotions and feelings) [45, 46] which occurs during all waking hours, but can also be deliberately modified by experience-oriented activities. The close alignment between physical state variables and activities is intended to facilitate the identification of mechanistic linkages within the human-Earth system.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. High-level MOOGAL categories.

Eight categories of activities are shown schematically in relation to the human-Earth system state variables suggested by [40]. Regions with solid outlines indicate the six state variable classes of the soma (orange), neural structure (dark blue), neural activation (light blue), things (brown), time allocation (grey circle) and remainder of Earth System (green). Dashed outlines indicate relationships of MOOGAL activity categories to state variables. MOOGAL categories can each be associated with one state variable class except maintenance of the inhabited environment, which spans both human-created and natural spaces, and organization, which spans both time allocation and neural structures (via laws and other cultural norms). Note that organization also includes changing the locations of humans, food, materials and artifacts.

https://doi.org/10.1371/journal.pone.0270583.g001

In brief, the high-level categories are defined as follows:

1. Food provision. All activities that contribute towards the provision of food to humans. This includes anything to modify the growth of edible organisms, the collection, storage and processing of edible organisms, and the final preparation of food prior to eating (usually at home or in a restaurant) including cleanup.
2. Non-food provision. The collection of non-edible materials from the natural environment, including essential initial processing to non-specialized forms.
3. Inhabited environment maintenance. Cleaning and maintenance of surfaces and textiles, and rearrangements of objects and plants within the spaces inhabited by humans. These can include both the interiors of buildings (homes, offices, public buildings) and exterior spaces (lawns, gardens, urban parks).
4. Transformation of materials. The activities that physically transform extracted materials into the entire range of human creations (other than food), including movable objects and machines (referred to as artifacts) as well as immovable buildings and infrastructure.
5. Somatic maintenance. Care of the surface and interior of human bodies, including decorative aspects. This encompasses physical care and health care as well as personal hygiene, grooming and sleep.
6. Deliberate neural restructuring. Activities that deliberately aim to alter the neural structure of humans, including formal education, teaching through oral tradition, research and religious practices.
7. Organization. Activities that do not directly aim to achieve any of the prior categories, but which have the outcomes of a) altering time allocation through communication with others, and/or b) mediating the spatial relationships between other humans and the physical world. The latter includes the locations of people, artifacts and materials, as well as the access rights of people to locations and artifacts via their codification in neural structures and documents.
8. Experience-oriented. Activities that are motivated primarily by the subjective experience, such as meals, recreation, games, hobbies, passive observation and idle relaxation. Although oriented towards experience, activities included here may be socially-aggregated, such as socializing and participation in cultural events.

Choices in assigning hierarchical associations

Although the definition of subcategories and their associations with high-level categories was straightforward in most cases, we were still left with the need to make a few choices in order to be able to make effective use of existing time use data. We briefly discuss the most important of these.

First, meals could be considered as somatic maintenance, given that eating is an essential part of maintaining a living body. However, breathing is also essential, and is never recorded as an activity, given its ubiquity and therefore lack of usefulness as an observation. Similarly, all living humans must eat regularly in order to stay alive, and for most humans the time spent physically ingesting food is considerably less than what is expended at meals. Meals, instead, hold a large cultural value in many societies, and are often an enjoyable activity given the sensory experience and social context. Because the meal experience typically goes far beyond the simple act of ingestion, and it is this extended experience that is likely to account for most variation in the meal time expenditure as recorded in time use surveys, the decision was made to include meals within the experience-oriented category.

Similarly, active recreation, including exercise, could be considered a somatic maintenance activity when undertaken with the deliberate aim of improving physical fitness. However, exercise is often recorded in combination with other activities that are undertaken purely for enjoyment. It is therefore included within the experience-oriented high level category.

Religious activities are diverse, ranging from individual meditation and prayer to large gatherings, rituals and social events. Because they are largely experiential, they could be categorized as experience-oriented. However, they are generally differentiated from non-religious cultural and social activities by their deliberate influences on beliefs, norms and ways of thinking, all of which involve persistent modifications of neural structure. Religious activities are therefore included within deliberate neural modification.

Maintaining built environments and their surrounding plant life could be theoretically separated, which would make it more in line with the state variables (cleanly separating buildings vs. rest of Earth system, Fig 1). However the distinction between the two domains is problematic, given the gradation from yards and gardens to public parks, and the fact that the activities are often aggregated together in existing datasets. The decision was therefore made to combine all maintenance of the inhabited environment together, including modifications of the immediate vegetation (not intended to produce food).

Finally, the organization category also produces outcomes that span two state variable classes—time use and neural structures. This was unavoidable given the fact that norms and relationships, which are encoded in neural structures, are often modified through the same activities that allocate time use in existing datasets. We also note that the transportation aspects of organization do not correspond in an obvious way with any of the state variables in ref. [40], but can be considered as modifying the location properties of soma, things and Earth system materials.

Priority scheme

Although true multi-tasking is a rare occurrence for individuals on short timespans, as discussed above, the aggregation of activities over time and between individuals unavoidably results in the co-occurrence of multiple outcomes within the same time interval. To return to the food processing factory example above, the people involved are undertaking many organizational activities, as well as food processing, while the participants are themselves physically modified by their activities, through the neural activation stimulated by the experience as well as the outcomes on the growth of their bodies through the types of physical activity in which they are engaged. The co-occurence of simultaneous activities within a social aggregate can introduce significant ambiguity in how it should be classified.

To address this challenge, we introduce a priority scheme to the MOOGAL activities (Fig 2), similar to the coding priority used in the Cross-cultural Studies in Time Allocation monographs [2]. The priority scheme determines which physical outcomes of an aggregate activity are used to identify it, and gives higher priority to more easily identified outcomes. Deliberate modifications of the physical world are given the highest priority: if a compound activity has for its motivating outcome a modification of the non-human Earth system, human-made creations and environments, or humans themselves, this takes precedence over any organizational or unintended human outcomes. In the absence of a deliberate modification, organizational outcomes take precedence over neural activation (which is omnipresent) or unintended consequences for human bodies. Finally, if the activity is not undertaken to deliberately modify the world nor to produce organizational outcomes, it is identified as being undertaken for neural activation. Unintended consequences on human bodies, although always occurring, are (by definition) not motivating outcomes and are therefore never identified.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Prioritization of simultaneous outcomes.

Schematic illustration of how simultaneous outcomes can occur, in relation to the prioritization scheme. Simultaneous outcomes are those which co-occur horizontally. For example, many organization outcomes (thin grey bars) could co-occur with deliberate modification activities throughout the day, but are not identified as the motivating outcomes because they are lower priority. Similarly, somatic and neural outcomes are always occurring, but somatic outcomes are only identified when they are the primary deliberate outcome, and neural outcomes are only identified when not co-occurring with one of the tier I or II outcomes. Context, social motivation and wellbeing (left-hand side) are not captured by the MOOGAL, but are examples of dimensions of time use that could be addressed with additional parallel lexicons.

https://doi.org/10.1371/journal.pone.0270583.g002

Priority is not intended to reflect a judgement regarding the normative value of activities, but simply to avoid ambiguity in the results by codifying an explicit procedure. This strategy introduces a systematic error, in that lower-priority categories will tend to be underestimated, so that the time recorded in lower priority activities should be interpreted as a lower bound. The intent of the priority scheme is to reduce arbitrariness and make results more reproducible.

Fractional time allocation

The primary intended use of the MOOGAL is to provide a cross-mapping between the lexicons of existing datasets. These other lexicons frequently include aggregated activities that clearly span multiple motivating physical outcomes. For example, the aggregated activity ‘Housework’ (defined by social motivation and context) will typically include, among its motivating physical outcomes, some maintenance of surroundings, some food preparation, and may also include a fraction of organization in some lexicons. In order to disambiguate such aggregations, the coder uses whatever information is at their disposal (additional data, detailed descriptions, etc.) to estimate the fractional distribution of the activity between MOOGAL subcategories. The total allocation of time across the activity must sum to one, ensuring exclusivity.

The recategorization of any activity between two or more MOOGAL categories unavoidably adds a layer of uncertainty to the final results. To provide a quantitative estimate of this categorization-related uncertainty, we use a structured subjective assessment of the confidence range during the association of MOOGAL categories with the original activities, and propagate this through to the final result by weighting each category by the time devoted to each activity. The subjective confidence ranges were calculated by following three steps: 1) considering all MOOGAL categories that could conceivably be associated with an activity, according to the prioritization scheme, 2) for the dominant category, assessing the minimum and maximum feasible time fraction that it could represent (aiming for 95% confidence interval), 3) similarly assessing the feasible minimum and maximum time fractions for the remaining categories, in diminishing importance. The ranges were recorded as 1 standard error, symmetrically distributed about the mean. For example, an activity that is estimated as 60% interactive stimulation (best guess), with feasible minimum and maximum of 30% and 90%, respectively (95% confidence range) would be recorded as mean time fraction 0.6 with standard error 0.15.

Comparison with existing lexicons

We next show how the MOOGAL compares to three of the four lexicons introduced above: one lexicon designed for hunter-gatherer study (CCSTA), one lexicon designed for time use surveys (ICATUS), and one economic lexicon (ISIC). Four coders independently allocated each activity in these three lexicons with one or more of the 24 MOOGAL subcategories listed and briefly described in Table 2. Where an activity was identified as including multiple MOOGAL categories, this was expressed by coding time fractions between 0 and 1 (and summing to 1) using the provided descriptions of the activity components and/or supplemental information on the detailed breakdown (e.g. ISIC tier 3 subdivision). Initial disagreements between coders were discussed and resolved by consensus. The final results are shown for all activities in the Supplementary Figures and summarized here.

Fig 3 shows the distribution of activities across the MOOGAL categories. The distribution is more evenly-spread for the time use lexicons (both ICATUS and CCSTA), which can be explained by the fact that—like the MOOGAL—these surveys aimed at capturing largely holistic views of life. The economic (ISIC) categories are much less widely distributed across MOOGAL categories, with foci in the Artifacts and Allocation sub-categories, but still have some degree of representation across most other categories, consistent with the diverse character of economic activities. Importantly, there were no activities that could not be associated with one of the MOOGAL categories in any of the three lexicons, indicating that the MOOGAL lexicon is exhaustive. Human travel is absent from the hunter-gatherer lexicon because, by design, the lexicon is only for recording activities observed within the hunter-gatherer camp.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Sum of time fractions by category.

Each bar shows the sum of all time fractions associated with each category, for each of the three illustrative lexicons. The sum reflects how frequently activities were associated with each MOOGAL category.

https://doi.org/10.1371/journal.pone.0270583.g003

The comparison shown in Fig 4 reveals that, although there are only 24 MOOGAL subcategories, most activities in the three larger lexicons can be directly associated entirely with one MOOGAL subcategory, indicating relatively good alignment. Each bar shows the average time fraction (ranging from 0 to 1) of all activities associated with each MOOGAL category. Where the average time fraction is 1, the corresponding MOOGAL category was unambiguously associated with one or more activities in their entirety. This direct equivalence reduces the potential for coder error and the need for subjective judgments. Where the value is less than 1, one or more activities were only partially associated with the MOOGAL category. Note that the values shown in Fig 4 cannot be directly related to uncertainty in the resulting time use calculation, since this also depends on the relative distribution of time between different activities (i.e. rare activities contribute little to the overall uncertainty).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Average time fraction by MOOGAL category.

Each bar shows the average of all time fractions associated with each category, for each of the three illustrative lexicons. Where the average time fraction is 1, only unambiguous, direct associations occurred within the category. Where the value is <1, the association required some subdivision of activities amongst multiple categories. For example, food processing was only unambiguously associated with activities in all three lexicons, whereas infrastructure creation and maintenance was estimated as a subcomponent of at least one activity in both ICATUS and ISIC (and is entirely absent from hunter gatherers, who did not create infrastructure).

https://doi.org/10.1371/journal.pone.0270583.g004

A close examination of the supplementary figures shows that almost all of the partial time fraction designations split activities between subcategories of a high-level MOOGAL category, rather than spanning multiple high-level categories. For example, activities are frequently split between two or more experience-oriented subcategories. The only activities that are fractionally allocated between multiple high-level MOOGAL categories are some volunteer activities (ICATUS), veterinary activities (ISIC), undifferentiated goods and services production of private households for own use (ISIC), and fetching and managing household water and fuel (CCSTA). Thus, the MOOGAL associations with these three lexicons have very little uncertainty for the eight high-level categories.

Application to two countries

Finally, we provide two example applications of the MOOGAL to time use data from specific domains. We choose Canada and India, given the large contrast in the resolution of their time use survey categories (n = 18 in Canada vs. n = 155 in India) as well as income levels. The Canadian data include the 2015 Time Use Survey [47] for non-paid work time and the 2019 economic data for paid work time [48, 49]. The Indian data include the 2019 Time Use Survey [50] for non-paid work time and the 2019 economic data for paid work time [51]. The results are shown in Figs 5 and 6.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Time use in Canada using the MOOGAL.

Combination of time use and economic activity data. Shaded bars show the total time allocated to each MOOGAL category, while hatched portions indicate the time allocated through the formal economy (work for pay or profit) Thin black lines indicate the 95% confidence range for the total time use in each MOOGAL category.

https://doi.org/10.1371/journal.pone.0270583.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Time use in India using the MOOGAL.

As in Fig 5.

https://doi.org/10.1371/journal.pone.0270583.g006

In terms of the uncertainty introduced by the translation to the MOOGAL subcategories (indicated by error bars), the comparison shows very little additional uncertainty for the well-resolved Indian data, but considerably more for some categories of the highly-aggregated Canadian data. Primarily, the inclusion of all ‘housework’ as a single category in Canada leads to a large uncertainty in food preparation and inhabited environment maintenance.

Despite the coarse aggregation of the Canadian data, some differences in time use between the countries are clear. Although we do not dwell on these differences here, we note the much greater time in food growth and collection in India, greater time devoted to health care in Canada, greater time in hygiene and grooming in India, and much greater time in organizational activities (both transportation and allocation) in Canada. Given that the purpose of this comparison is simply to illustrate the potential of the method, we leave it to future work to explore differences such as these between populations, as revealed by consistent application of the MOOGAL.