What is Nature? A Prototype Analysis of the Natural Environment

Defining Nature With Environmental Features

Nature has been identified as “perhaps the most complex [word] in the [English] language” (Williams, 1983, p. 219) with a singular term encompassing a multiplicity of things and processes (Williams, 1980). Nature can encompass the essence of things (“the nature of something”), as material things that are outside of humanity (“a nature thing”), or even the laws or realities of the universe (“law of nature”; Ginn & Demeritt, 2009). The word nature was derived from the Latin term natura meaning to give birth (see Ducarme & Couvet, 2020). During orthodox medieval times, nature was defined as encompassing all of God’s creation with humans as a part of nature, yet in a dominant position (Williams, 1980). Nature was later disconnected from God in Western thought, then defined as separate from and untouched by humans. The industrial revolution increased the exploitation of physical nature, which may have bolstered humans’ feelings of separation from and dominance over the natural environment as something to use and control (Williams, 1980).

This separation between humans and nature remains common in Western thought. For example, even though the definition of nature is not consistent across mainstream English dictionaries, the majority include a clause like “not made” (Oxford; Encyclopedia Britannica, “not caused” (Collins), “not controlled” (Longman), “independent” (Cambridge), and “without” humans (Dictionary.com; see Supplemental Material for full definitions). Similarly, Chipeniuk (2025) laments that nature as experienced and studied is often “fuzzy” in that it includes substantial human influence, which may dilute its psychological benefits. The dichotomy of humans versus nature also emerges when researchers specifically ask their participants how they think about nature (Haluza-Delay, 2001; Pointon, 2014; Vining et al., 2008). For example, when almost 1,000 New Zealand residents were asked what they thought nature was, approximately 58% noted that it was something where humans and their influence are not a part (Fehnker et al., 2021). Similarly, after participating in a 12-day adventure trip, 8 teenagers were asked about their conceptualization of nature, and they described it as not civilization (Haluza-Delay, 2001). When Vining et al. (2008) asked U.S. student and community participants to identify what came to mind when they thought of a natural environment, the largest theme that emerged was untouched by humans (47% of respondents) with others like preserved land, uninhabited, and not human made as other similar and frequent categories. Participant-listed features of unnatural environments (a separate question) were largely about human presence and damage to nature.

Beyond the (sometimes) defining feature of natural environments as separate from human influence, research has assembled other features listed by participants. For instance, Vining et al. (2008) coded responses into a few categories with broad pleasant adjectives such as clean, peaceful, beautiful, and balanced. In another study, adolescents from the UK were asked to describe what the word “environment” means, and the features listed were more biotic (i.e. living natural elements, such as plants and animals) than abiotic (i.e. non-living natural elements, such as mountains and bodies of water, Loughland et al., 2002), similar to how UK children used the words nature and environment when prompted with photos in another study (Bonnett & Williams, 1998). Similarly, adult New Zealanders provided open-ended descriptions of nature and produced dozens of unique features (Fehnker et al., 2021). Commonly listed items were plants (17%), animals (15%), trees (13%), and oceans/seas (10%), with more specific terms (e.g., flower, bee, stream) typically listed by fewer participants (<2%). These features were also grouped into broader themes of flora (plants, 58%), fauna (animals, 37%), hydrological (water types, 28%), geological (land types, 24%), atmospheric (air, weather, 11%), space (large bodies, 8%) and consumable (food, 1%).

There is clearly variability in what people list when describing natural environments, including variation in whether they see nature as separate from human activity. These variations likely stem from personal experience, culture, and identity (e.g., nature relatedness). For example, when children in rural Northwestern Ontario (Canada) described nature and natural elements, none described parks, but rather described trees, forests, and the “bush” that they had experience with (Tillmann et al., 2019). Similarly, a study of Spanish children found that their associations with nature depended on their usual daily experiences with nature (Collado et al., 2016). Additionally, a study on the concept of nature in 60 different languages found three useful clusters: nature as inclusive of humans, exclusive of humans, or as a deity (Coscieme et al., 2020). In many Indigenous languages, nature is described as inclusive with humans, where humans, non-human animals, and the natural environment are all psychologically connected to one another (Milfont & Schultz, 2016). These cultural variations are similar to individual differences in subjective connections with nature, as suggested by a strong correlation between a community-developed measure of Anishnaabe Indigenous Culture and nature relatedness (Mamaweswen Niigaaniin & MacNeill, 2022). Additionally, participants with high subjective nature relatedness were significantly more familiar with, and positive toward, all but 1 biome (i.e., beaches) of 10 biomes studied (e.g., rivers, swamps, mountains, etc.; Mangone et al., 2021). This may reflect highly nature-related people’s tendency to spend more time in natural environments, as well as to notice more nature in urban scenes (Desrochers, 2023).

Even independent of individual difference measures, people in Western nations seem to have paradoxical views of nature. In the same samples where many participants defined natural environments as untouched by humans, over 75% of people also indicated that they saw themselves as a part of nature when asked directly in another question (Vining et al., 2008). Similar mixed views were observed in the study of UK urban and rural primary school students that prompted small-group discussions with photos (Bonnett & Williams, 1998). The children described humans (including themselves) as part of nature, but they also thought that nature was separate from day-to-day life and something they had to work to protect. Taken together, these apparently contradictory thoughts reveal complexity in the meaning of nature and thereby suggest the usefulness of a prototype approach.

Prototype Approach

One method that can help shed light on the conceptualization of nature in the context of environmental features is the prototype approach. Rosch (1978) suggested that concepts that are difficult to define (like many natural language concepts, including nature) can be characterized by prototypes, rather than in terms of a set of necessary and sufficient criteria (i.e., prototypes are alternatives to classical definitions). Prototypes are organized around their clearest cases, or best examples, which are features that are most typical and familiar to the category of the concept that is being defined. The study of prototypes began with a focus on category membership, the idea that some category examples are better than others. For example, participants view sparrows as a better example of the category “bird,” compared to penguins (Rosch et al., 1976).

Aside from conceptualizing prototypes in terms of category membership, the prototype approach can also be used to assess the defining features of concepts. For example, one early study to adopt this approach found that when defining the concept of a lonely person, some features were viewed as more central (or prototypical) to the definition (e.g., separate from others, unloved) whereas others were less central (e.g., quiet, reserved, introspective; Horowitz et al., 1982). Since that time, the range of natural language concepts that have examined the prototypicality of features of concepts has grown substantially (e.g., emotion, Fehr & Russell, 1984; Shaver et al., 1987; love, Aron & Westbay, 1996; Fehr, 1988; Fehr & Russell, 1991; Fitness & Fletcher, 1993; forgiveness, Friesen & Fletcher, 2007; Kearns & Fincham, 2004; respect, Frei & Shaver, 2002; virtue, Gulliford et al., 2021; heroes, Kinsella et al., 2015; gratitude, Lambert et al., 2009; creativity, Amelang et al., 1991; leadership, Rush & Russell, 1988).

Rosch identified two main criteria for researchers to make claims that concepts are organized as prototypes: (1) participants can make meaningful judgments that some examples are more representative of the category or that some features are more central as defining features than others, and (2) the internal structure of the prototype affects cognition (e.g., judgments, reaction time). A typical prototype analysis begins with laypeople generating features in response to open-ended definition questions (e.g., What is love?), and then the coded features are rated by a second group of participants in terms of how characteristic the features are of the concept (i.e., obtain prototypicality ratings). Initial evidence for a prototype structure of a construct is found if some features are listed more frequently than others and if participants consistently rate some features as more characteristic (i.e., prototypical) than others (i.e., peripheral or non-prototypical). Evidence of a prototype structure would be lacking if only a few common features were listed and/or they were all rated as similarly prototypical. After the prototype structure has been identified and initial evidence has been provided, the next step is to examine is whether the prototype structure affects cognition (Rosch, 1973). Researchers have assessed the effects on cognition in a variety of ways including reaction time, applicability to hypothetical situations, and logical inferences. A lack of cognitive consequences (e.g., no differences in judgments between high and low prototypical scenarios) would also argue against a prototype structure.

There are many benefits of using a prototype analysis to understand concepts aside from providing evidence that the concept might best be conceptualized as a prototype (vs. as a classical definition). A prototype analysis involves a “bottom up” examination in which the content and the structure of a concept are described from the perspective of laypeople. Tapping the collective knowledge that people hold for concepts such as the natural environment is useful because people rely on this knowledge to understand their environment, and this has potential implications for well-being and environment-related behaviors. Analyses of lay conceptions also can highlight aspects of a construct that may have been overlooked by experts and can provide considerable breadth of a construct (Gregg et al., 2008). In other words, the bottom-up knowledge can be used to complement top-down, expert-generated definitions. Finally, the approach typically involves staying as true as possible to the wording generated by laypeople making the list of features valuable when constructing measures and designing studies that are meaningful to the people that participate in those studies. Past research has provided lists of features that laypeople associate with nature, and we build on this by adding centrality ratings and testing whether nature fits a prototype structure.

Accordingly, confirming and describing the natural environment prototype has value for a large body of social sciences research on nature, as well as applications that seek to increase people’s exposure to nature and its salubrious effects. The prototype description can help researchers assess how well their objective nature stimuli match lay views, and how their participants interpret questions or instructions about engaging with the natural world. That is, scientists sometimes impose meaning via the specific nature stimuli they use, such as showing human participants images of a forest and comparing them to others who view images of buildings. Here, nature is defined as forests, assuming this broader interpretation is taken (vs. narrower conclusions about particular forests and buildings). With this approach, nature is defined clearly, yet it may or may not be interpreted as nature by participants. Other approaches do not explicitly define nature and instead invoke the term while leaving the interpretation up to participants. For example, the Inclusion of Nature in Self (INS) measure asks participants to choose among pairs of circles, labeled “nature” and “self,” and where the degree of overlap is varied (Schultz, 2001). Answers are meant to indicate a stronger or weaker sense of connection with nature (self-nature overlap), akin to the nature relatedness construct mentioned earlier. Participants may think about nature in idiosyncratic or stereotypical ways, yet it is typically not possible to know what they have in mind as they answer. This approach has the advantage of capturing people’s personal views (it is “their nature”), yet with the possible disadvantage that different people see somewhat “different questions.” Understanding the nature prototype provides insight on typical interpretations, as well as less frequent variations. Similarly, intervention programs and studies may ask people to spend time in or reflect on experiences natural environments, yet often leave the details to participants (e.g., Nisbet, 2015; Passmore et al., 2022). There are advantages and disadvantages to letting participants define nature as they like, yet it is clearly useful to better understand what they have in mind. Finally, architects and designers might draw on the natural environment prototype to select features that will convey a sense of nature in physical spaces.

Current Study

The goal of the current research was to understand how laypeople think of nature in terms of features of the natural environment across three studies using a prototype analysis. In Study 1, student and community participants were asked to list features of natural environments within a 3-min window. We determined the 100 most frequently listed features to then use in Study 2, where new participants rated them for centrality, providing another quantitative estimate that goes beyond most previous qualitative work. In other words, Studies 1 and 2 define the prototype by ranking participant-generated features for commonality and centrality. A third study helped confirm the prototype structure by comparing participants’ reactions to scenarios written with high- or low-prototypical features. We predicted that scenarios with more central features would be seen as better examples of nature (natural environments) than scenarios with less central features. The dependent variable items were phrased as both nature and natural environments to assess possible divergence across the terms, yet that we predicted would be very similar. Finally, all studies included an individual difference measure of subjective connection with nature (nature relatedness) to explore possible personality differences. We did not have strong a priori hypotheses, yet we reasoned that highly nature-related people may have richer and more elaborate views of nature (e.g., list more features of the natural environment) that could emerge in the prototype analysis.

Method

Analysis, Results, and Discussion

All responses were compiled into one data set. We did not expect substantive differences between students and community members, and casual comparisons aligned with this expectation (see details in the Supplemental Material). Responses that were identical or synonymous were combined following prototype protocol (e.g., Fehr, 1988). For example, “trees” and “tree” were combined as were “animal,” “animals,” and “sounds of animals.” There were themes that varied slightly while expressing similar content, such as “fresh air” and “good air quality,” that were combined. Yet, other related features like “bodies of water,” “rivers,” and “lakes” were not combined and remained separate. Similarly, “sky” and “clouds” were left as separate features. Additionally, some responses were counted as representing more than one feature. For example, “blue sky” was counted as “blue” and “sky.” This was done with the intent of staying true to data and minimizing researcher imposition on category development. We performed the integration and counting twice, with the more recent and higher-confidence version reported here conducted after data for Studies 2 and 3 were collected. It was undertaken to verify results due to unavailability of some initial count documentation and study personnel. The samples resulted in a combined 2,349 responses (M = 12.30 per participant, SD = 8.56), of which 1,811 were instances of features contributed by 5 or more participants.

We selected the 100 most mentioned features (from our original count) for further investigation in Study 2. These 100 top-listed features appear in Table 1, along with some additional features that were frequently listed (>5 spontaneous mentions) yet omitted or under-counted in the first attempt. (The uncommon features omitted in Table 1 can be found at: https://osf.io/f3vqt). The most commonly listed features (>50 spontaneous mentions) were: trees, animals, grass, water, plants, and rocks. These largely mirror similar previous research, and we elaborate on these findings in the General Discussion, after adding information from Studies 2 and 3.

OPEN IN VIEWER

Table 1.

Counts (Study 1) and Centrality Ratings (Study 2) of Commonly Listed Natural Environment Features.

Feature	Count (#)	Frequency (% of participants)	Mean centrality	SD centrality
wildlife	16	8.38	6.31	1.05
trees	128	67.02	6.23	1.10
ecosystem	8	4.19	6.23	1.14
plants	54	28.27	6.22	1.04
earth	14	7.33	6.13	1.31
forest	29	15.18	6.10	1.24
fresh air	31	16.23	6.09	1.24
animals	100	52.36	6.09	1.15
water	83	43.46	6.07	1.24
sun	39	20.42	6.07	1.40
organisms	5	2.62	6.06	1.22
sky	29	15.18	6.04	1.37
sunlight	30	15.71	6.02	1.30
air	43	22.51	6.01	1.44
outdoors	10	5.24	6.01	1.33
vegetation	6	3.14	6.01	1.22
oxygen	9	4.71	6.00	1.48
not man made	12	6.28	5.99	1.56
rivers	38	19.9	5.97	1.22
ocean	28	14.66	5.95	1.35
soil	27	14.14	5.93	1.32
life	9	4.71	5.92	1.37
lakes	32	16.75	5.89	1.27
mountains	37	19.37	5.88	1.36
land	6	3.14	5.88	1.34
leaves	25	13.09	5.84	1.32
rain	29	15.18	5.83	1.37
weather	20	10.47	5.81	1.42
habitats	6	3.14	5.79	1.37
streams	10	5.24	5.76	1.31
flowers	48	25.13	5.74	1.30
climate	7	3.66	5.72	1.44
seasons	4	2.09	5.72	1.44
clouds	20	10.47	5.70	1.52
grass	92	48.17	5.69	1.48
rocks	53	27.75	5.68	1.40
flora	3	1.57	5.68	1.46
dirt	47	24.61	5.67	1.46
waterfalls	8	4.19	5.66	1.45
wild	5	2.62	5.64	1.48
wind	39	20.42	5.62	1.41
atmosphere	4	2.09	5.57	1.61
bees	5	2.62	5.57	1.53
hills	14	7.33	5.57	1.45
minerals	9	4.71	5.54	1.47
wood	15	7.85	5.53	1.41
valleys	7	3.66	5.49	1.51
organic	4	2.09	5.49	1.56
moss	6	3.14	5.48	1.56
fish	10	5.24	5.48	1.54
ponds	9	4.71	5.47	1.43
stars	6	3.14	5.45	1.72
pollen	4	2.09	5.45	1.53
bushes	14	7.33	5.43	1.45
fauna	3	1.57	5.43	1.63
living	7	3.66	5.41	1.53
green	29	15.18	5.40	1.55
desert	8	4.19	5.37	1.59
fields	5	2.62	5.37	1.50
sticks/twigs	5	2.62	5.37	1.50
breeze	13	6.81	5.35	1.54
moon	6	3.14	5.31	1.87
weeds	5	2.62	5.30	1.56
snow	16	8.38	5.25	1.65
temperature	7	3.66	5.24	1.70
shrubs	4	2.09	5.22	1.53
squirrels	6	3.14	5.22	1.56
vines	4	2.09	5.21	1.56
sand	18	9.42	5.19	1.53
chirping	4	2.09	5.14	1.71
fruits	6	3.14	5.13	1.53
light	8	4.19	5.11	1.66
lightning	5	2.62	5.11	1.76
moisture	5	2.62	5.08	1.60
beach	8	4.19	5.05	1.58
beautiful	13	6.81	4.97	1.76
untouched	9	4.71	4.93	1.93
resources	6	3.14	4.92	1.65
shade	4	2.09	4.84	1.60
cycles	4	2.09	4.68	1.84
peaceful	16	8.38	4.64	1.68
serene	7	3.66	4.63	1.64
cold	6	3.14	4.61	1.59
free	5	2.62	4.58	1.90
smell	7	3.66	4.58	1.74
warm	8	4.19	4.58	1.64
spaces	6	3.14	4.53	1.69
open	9	4.71	4.49	1.79
calm	9	4.71	4.44	1.76
relaxing	12	6.28	4.34	1.72
food	7	3.66	4.31	1.69
cows	3	1.57	4.29	1.86
clean	11	5.76	4.17	1.76
quiet	12	6.28	4.07	1.64
fire	7	3.66	3.98	1.79
park	5	2.62	3.78	1.83
comfort	6	3.14	3.76	1.73
humans	6	3.14	3.74	1.89
gravel	4	2.09	3.57	1.85
people	6	3.14	3.52	1.86
birds	38	19.9	N/A	N/A
blue	5	2.62	N/A	N/A
bugs	24	12.57	N/A	N/A
dog	5	2.62	N/A	N/A
Earth	5	2.62	N/A	N/A
fresh	10	5.24	N/A	N/A
house	5	2.62	N/A	N/A
insects	27	14.14	N/A	N/A
no human	5	2.62	N/A	N/A
noises	8	4.19	N/A	N/A
vegetables	5	2.62	N/A	N/A

Note. Features in the table are sorted by centrality ratings; features without centrality ratings (N/A) were inadvertently excluded from Study 2.

We also explored whether a person’s level of nature relatedness would predict the features listed. The number of features listed by an individual participant ranged from 1 to 61, with M = 12.30 and SD = 8.56. We found a significant positive association between nature relatedness and number of features listed, ρ (189) = .20, p = .005, computed as a Spearman’s rank correlation given clear outliers in the number of features. Additionally, data were split at the median score of nature relatedness (Mdn = 3.52) so we could visualize possible differences. See Figure 1 for Word Clouds of the responses provided by the two groups. Similar to the correlation, those above the median on nature relatedness provided more features, on average (M = 13.62, SD = 7.86), than those lower in nature relatedness (M = 10.81, SD = 9.10), t(189) = 2.29, p = .023, d = 0.33. This suggests that highly nature-related people see or can recall a larger set of features that describe natural environments. Despite this statistically significant difference, the word clouds appear qualitatively similar across nature relatedness levels.

OPEN IN VIEWER

Figure 1.

Word clouds of terms used to describe the natural environment separated by levels of nature relatedness using a median split. Panel A: Low nature relatedness participants. Panel B: High nature relatedness participants.

Method

Results and Discussion

The ratings of prototypicality (centrality) for 100 features appear in Table 1. The top five (highest) prototypical features were wildlife, trees, ecosystem, plants, and earth, whereas the bottom five (lowest) prototypical features rated were people, gravel, humans, comfort, and park. As expected (e.g., see Fehr, 2005), count frequency and centrality ratings did not mirror each other perfectly, yet there was a moderately sized positive association, ρ (98) = .51, p < .001. Some notable deviations include organisms, vegetation, and ecosystem, which were rated as highly prototypical but spontaneously listed relatively infrequently. We also observed that standard deviations were smaller for the more prototypical items. This could be interpreted as the consequence of very high average ratings (i.e., ceiling effect), yet it also implies higher consensus about the most prototypical features (vs. the least prototypical ones). Note that centrality ratings are missing for some features because they were not in our initial top-100 count. Many of these features had comparably low counts, where small differences fall on either side of the Study 2 selection threshold (e.g., blue, vegetables, dog); however, our second count also discovered some commonly listed features where initial exclusion was a clear oversight (i.e., birds, bugs, insects). It seems reasonable to predict that features with high counts would have higher centrality ratings too, given that these two elements correlated moderately strongly. Additionally, it seems likely that most low-count features (i.e., excluded from Table 1) would receive even lower centrality ratings; 3.52 was the minimum observed (for “people”), and centrality ratings were only gathered for frequently listed features (i.e., the initial top 100).

We also explored nature relatedness as a predictor of centrality ratings. Across all rated features, those higher in nature relatedness also rated natural environment features as more central to the prototype, r(426) = .22 p < .001. This suggests that highly nature-related people have a more expansive view of nature (i.e., features are generally seen as more central), and this may stem from similar processes that had highly nature related people spontaneously list more features in Study 1.

Methods

Results

To test the hypothesis that scenarios using more prototypical features, as identified in Study 2, would be rated as better describing a natural environment than the scenarios using less prototypical features, independent-sample t-tests were conducted. There was a significant difference in the composite fit ratings between those who read high versus low prototypicality scenarios, t(194.76) = 5.06 p < .001, d = 0.71. More specifically, those who read the scenarios with the more prototypical features rated the scenarios as more like a natural environment (M = 5.90 SD = 1.08) than those who read the scenarios with less prototypical features (M = 5.04 SD = 1.34). The same pattern emerged for each individual item too (see Table 2), indicating that results were quite similar whether participants were rating fit with “nature” or a “natural environment.”

OPEN IN VIEWER

Table 2.

Means, Standard Deviations, and Inferential Tests Comparing the High- and Low-Prototypical Scenarios on Fit With Concept of Nature (Study 3).

Dependent variables	M (SD) low prototypicality	M (SD) high prototypicality	t	d
Natural environment	5.29 (1.29)	6.12 (.93)	5.25***	0.74
Walk in nature	4.82 (1.55)	5.66 (1.40)	4.12***	0.58
Consider nature	5.01 (1.49)	5.91 (1.22)	4.73***	0.66
Composite	5.04 (1.34)	5.90 (1.08)	5.06***	0.71

***

p < .001.

As exploratory analyses, we also tested whether the scenarios differed in their ratings of happiness, stress, restoration, relaxation, interpersonal conflict, and wanting to be alone with independent-sample t-tests (see Table 3). Most tests were non-significant, but desire to be alone was significantly higher in the highly prototypical scenarios.

OPEN IN VIEWER

Table 3.

Means, Standard Deviations, and Inferential Tests Comparing the High- and Low-Prototypical Scenarios on Exploratory Item Ratings (Study 3).

Dependent variables	M (SD) low prototypicality	M (SD) high prototypicality	t	p	d
Happiness	5.41 (1.20)	5.62 (1.31)	1.20	.23	0.17
Stress	2.64 (1.66)	2.64 (1.69)	0.003	.99	0.00
Restoration	5.07 (1.21)	5.12 (1.45)	0.27	.79	0.04
Relaxation	5.44 (1.18)	5.73 (1.32)	1.69	.09	0.24
Interpersonal conflict	2.56 (1.79)	2.84 (2.00)	1.05	.30	0.15
Alone	4.54 (1.61)	5.04 (1.38)	2.36	.02	0.33

Exploring personality, the correlation between participants’ levels of nature relatedness and their ratings of how well the scenario fit the description of natural environment (composite) was positive yet not statistically significant, r(201) = .13 p = .07. The trend is similar to positive correlations between nature relatedness and the number of features listed (Study 1) and higher centrality ratings (Study 2), yet its failure to meet conventional levels of significance resists a confident conclusion.

Describing the Nature Prototype

Turning to the specific contents of the nature prototype, our studies produced features similar to past, mostly qualitative, work, while also adding the novel dimension of lay prototypicality ratings. For example, other studies (and dictionaries) have highlighted separation from human activity as a defining characteristic of nature (e.g., Fehnker et al., 2021). Our results echo this as a theme, along with some paradoxical complexity. Many of the top (most frequently listed and highest rated in terms of centrality) features are things more abundant where there are fewer humans (e.g., wildlife, forest, fresh air), yet they can be present in urban areas too. Some features explicitly noted nature as separate from human activities (e.g., not man made, untouched), yet these were not rated among the most prototypical. Other features explicitly included people as examples of nature (e.g. humans, people), though they were rated among the least prototypical examples assessed in Study 2, albeit still near the scale midpoint. Therefore, it seems that our participants see humans as part of nature, but also not very good examples of it, and where human presence can detract from better examples of nature. This is consistent with Vining et al.’s (2008) paradoxical findings that many participants defined natural environments as places without humans, while also indicating that they were part of nature on another question. The prototype provides a more affirmative picture of what nature includes based on common features (e.g., trees, animals, sun), compared to other definitions focused on excluding human activity.

Vining et al. (2008) also found that common associations of a natural environment include purity, cleanliness, beauty, and peacefulness. These themes appeared in our data too, but perhaps a bit less strongly. For example, our participants listed fresh air commonly, and rated it as highly prototypical, yet clean was much less frequently listed and rated as modestly prototypical. Peaceful, serene, relaxing, and beautiful were spontaneously listed, but were rated as less prototypical than many other features connoting flora and fauna. It is noteworthy that negative emotions were absent from our list. Even as some natural environments can be scary, those places do not seem like common or good examples when thinking about nature. This is consistent with other work suggesting that nature is typically, if obviously not universally, quite pleasant (e.g., Meidenbauer et al., 2020, but cf. Bixler & Floyd, 1997).

Our results are also consistent with work showing that terms representing flora and fauna are among the best examples of nature. Like Loughland et al. (2002), we saw that the most prototypical items were living (wildlife, trees), yet many non-living examples were also rated as highly prototypical (earth, fresh air, water), so we hesitate to conclude a clear preference for biotic examples of nature. The term living itself was listed relatively infrequently and was rated in the middle of features assessed in Study 2. Still, nearly all the living and non-living features are conducive to healthy humans, so it seems that the nature prototype is largely pleasant (as previously noted) and consistent with the target of biophilia (Wilson, 1993), that is, an evolved need to affiliate with life and lifelike natural elements that support human health. Our instructions asked participants to list examples of the natural environment, which might have encouraged listing physical features, yet it is clear that our participants still saw plants and animals as key parts. Participants also included some dynamic processes that occur in these spaces with examples like seasons, cycles, climate, and weather seen as moderately prototypical.

Like Fehnker et al. (2021), we found that broader, rather than more specific, terms (e.g., animal vs. fish, plant vs. vines) were more commonly listed, and Study 2 suggests that they are also seen as more prototypical. We speculate that when participants are asked to rate the prototypicality of a broad term like water, they may think of water-related examples that are familiar and, thereby, good examples; thus, a broad term can be rated as a good example by most participants. In contrast, a more specific term (e.g., ocean, beach, river) may be locally familiar to fewer participants and thus seem less prototypical for some. Underlying this speculation is the finding that people seem to prefer biomes that are familiar across a wide range of likability dimensions (Mangone et al., 2021). Taking this idea further, some biomes were rated as more prototypical than others, with forests, rivers, oceans, and mountains seen as considerably better examples of nature than beaches and deserts. Imagery and design invoking the former may evoke a sense of nature more strongly than the latter biomes, at least on average across people in USA and Canada. Researchers who wish to evoke a nature context (e.g., manipulation with images) might also then prefer stimuli that includes these more prototypical biomes. Our suggestions about individual differences in experiences affecting individuals’ prototypes are speculative, yet consistent with other research suggesting these things matter to views of nature (e.g., Bixler & Floyd, 1997; Collado et al., 2016; Milfont & Schultz, 2016) and prototypes more generally (Sun et al., 2024).

We can speak to individual difference in nature relatedness better because it was measured in all studies. Consistent with broad expectations, people who had a strong subjective sense of connection with nature listed more examples of nature in 3 min (Study 1) and rated most features as being more prototypical (Study 2). The results of Study 3 were somewhat inconsistent, with nature relatedness positively associated with fit ratings, yet with an effect size that appeared smaller and that was not statistically significant. Still, the contradiction seems minor and perhaps due to statistical noise or method details. Across studies then, it seems like nature relatedness may include a more elaborated and inclusive understanding of nature. This accords with scale items about frequently noticing nature, even in the city, spending and enjoying time in nature, and high awareness of environmental issues. High levels of physical and mental experience with nature may be both cause and consequence of more elaborate views of nature over time. Interestingly, the core of nature relatedness is about feeling connected to, and even a part of, nature, so the paradox of some prototype features suggesting that nature is separate from human activities seems particularly strong for highly nature-related people. Future research will be needed to better understand how individual differences in nature relatedness and nature prototypes may or may not translate to the actual effects of natural environments. For example, will people high or low in nature relatedness disagree about whether a particular place (image, sound, etc.) counts as nature? Will highly nature related people notice more nature features and/or benefit more from (some types of) nature exposure? Although intuitively appealing, such personality moderation effects often go unreported or are null (e.g., Passmore et al., 2022), though such tests likely require quite large samples for good power.

The nature prototype presented here also has implications for much past and future research on nature. As noted earlier, studies often expose participants to stimuli deemed nature a priori (e.g., images of forests). Researchers’ intuitions about how their participants interpret such stimuli can now be compared to the nature prototype presented here. So, for example, images of forests likely evoke a sense of nature in participants as features like trees, forest, plants, sun, and sky were viewed as excellent examples of natural environments. In contrast, a researcher may be interested in the effects of parks, farms, or beaches for good reasons, yet there seems a greater chance that some participants will not interpret them as nature given that they were seen as less prototypical. Other kinds of research ask participants questions about nature. When no definition is provided, the protype here helps fill in gaps about what participants are more likely to be thinking about (i.e., highly prototypical features like wildlife, ecosystems, fresh air, and water). Alternatively, the nature prototype could be used to construct a definition of nature as a preamble to such questions, for example, “when it comes to nature, such as animals, plants, and water, how concerned are you about degradation?.” There are many useful and valid ways to study nature, yet across most of them, considering the nature prototype helps inform stimuli selection and participants’ interpretation.

Limitations and Future Directions

As noted in previous paragraphs, we believe that personality, culture, geography, and experience may produce individual differences in understanding nature, even as there appears to be a consensus core prototype. An important implication is that the prototype described here will apply best to the “average” American/Canadian, and less well to people with different cultures or geography, and perhaps also to smaller or distinctive subgroups in the USA or Canada. For example, many Indigenous cultures include specific ways of relating with nature (e.g., Niigaaniin & MacNeill, 2022), and racist policies have controlled access and experiences of nature in harmful ways (e.g., Roberts et al., 2023). Even within these nations, our sampling was far from representative. This suggests some broader caution in generalizing results, yet we suspect our convenience sampling gathered enough diversity to reasonably speak to the average or majority experience. Future research might then prioritize exploring more specific places or peoples, even as direct replication always has some value. For example, it would be interesting to test whether people who have grown up or lived in different biomes (deserts vs. mountains) differ in their nature prototypes. Other variations might be tested across age, urban versus rural locations, cultural dimensions (e.g., interdependence, religion), etc. Given the important qualitative elements of prototypes, such comparisons will likely work best when clearly defined, distinct groups can be compared. (We explored and found no significant urban/rural prototype differences in our data, yet we see this as a weak test with vague and modest variation in our samples). It may also be useful to craft prototypes for more specific types of nature more directly, either based on biome (e.g., asking about features of forests) or more psychological qualities of nature (e.g., restorative nature or scary nature). Many possibilities exist, and we thus recommend pursuing new nature prototypes based on theoretical relevance, local conditions, and applied needs. This will produce both useful data for those goals, as well as provide tests of how similar the prototype we offer here applies across such variations (generalizability). Indeed, some of the modest variation between previous research and the nature prototype here may be due to differences in the kind of people studied; alternatively, some variation might be attributed to the particular questions used to solicit answers (e.g., asking for associations vs. examples). Generating more prototypes, using similar methods, will also help address this open question (see Sun et al., 2024).

The nature prototype describes the way people (our participants) think about nature—it is subjective, based in phenomenology. This can be distinguished from research using specific physical or virtual examples of nature, and people’s perceptions may differ from the realities discovered by the physical sciences (e.g., “fish” is a common intuitive category, yet it includes diverse species with less common genetic relatedness compared to some land-based animals). Both subjective and objective views are useful, yet they should not be confused. Future research might also further explore the links between the two. For example, some low-level visual features of images predict subjective ratings of naturalness, including fewer straight edges, less color saturation, and less hue diversity (Berman et al., 2014); such low-level features also predict participants’ preferences for nature images (Kardan et al., 2015). In Chicago parks, subjective naturalness ratings were associated with objective measures of grass, water, and (negatively) buildings, but not significantly with trees, soil, or roads, results that both affirm and contrast with our prototype (Schertz et al., 2025). Sometimes stimuli are ambiguous and then understanding them as nature seems to improve perceptions. For example, when a pink noise recording was described to participants as being from a waterfall versus factory, both subjective pleasantness ratings and physiological (EEG) indicators preferred the nature interpretation (Koivisto et al., 2022). Similarly, although visual disorder is generally associated with low likability, natural images have much disorder and are well liked (Kotabe et al., 2017). Thus, both bottom-up and top-down interpretative processes matter and seeing something as nature seems to produce positivity. Study 3 explored whether scenarios constructed from highly prototypical nature features produced more pleasantness (happiness, relaxation, less stress) than scenarios with less prototypical features, yet results were null. Still, this seems like a weak test as all scenarios used nature features (cf. built or control) and were short and not particularly evocative (cf. actual nature, images, sounds, etc.). Studies that manipulate actual nature exposure provide stronger tests and have suggested differences (Barragan-Jason et al., 2023). It remains worthwhile to explore whether pleasantness co-varies with features’ prototypicality. Given the strong links between nature and positivity, it seems reasonable to think that the most prototypical features would have the best chance of evoking the idea of nature, including the positivity that commonly comes with that.

Conclusion

In sum, we present a nature prototype based on laypeople’s nominations and ratings of its important features. Meaningful variation in prototypicality ratings (Study 2) and judgment differences across scenarios (Study 3) further suggest that a prototype structure applies well to how people think about nature. The nature prototype is rich, including many general and specific terms that describe plants, animals, geology, waterways, feelings, and processes. Although participants rated people as part of nature, they were not seen as particularly good examples of it. Researchers and practitioners can use this prototype to design stimuli or spaces that are more likely to evoke a subjective sense of nature, or to compare these lay views of nature to stimuli and spaces that already exist. This particular prototype likely applies best to majority groups in the USA and Canada, and future research could create additional prototypes that apply to other places, identities, or specific types of nature, thereby also building a better understanding of how much and where we find similarity and differences.