Cultural tightness and resilience against environmental shocks in nonindustrial societies Authors: Denis Tverskoi, Carol R. Ember, Michele J. Gelfand, Eric C. Jones, Ian Skoggard, Louise Toutée, Sergey Gavrilets Published: 2024 ABSTRACT SignificanceHumans have historically thrived in diverse ecological systems, facing various natural hazards. What cultural and social adaptations help human groups cope with environmental shocks such as droughts, floods, or tornadoes? Employing mathematical modeling, we demonstrate that different levels of cultural tightness (capturing the degree of norm strength) can help small-scale nonindustrial societies withstand specific types of shocks. We found that mild slow-onset shocks require cultural looseness, leading to moderate cooperation and minimal cultural change. However, to cope with severe shocks, societies need intermediate levels of tightness, fostering significant cultural changes and high cooperation. Our findings can aid in developing effective coping strategies tailored to local cultures, which might be especially important for indigenous communities often located in high-risk disaster areas. The Globe faces accelerating climate change (1, 2) accompanied by a surge in costly weather disasters, such as heat waves, heavy precipitation, floods, droughts, extreme storms, tornadoes, and wildfires (3), each with the potential to cause catastrophic loss (4, 5). Therefore, responding and adapting to climate shocks and developing resiliency has assumed global significance and urgency (6). Effectively addressing these multifaceted challenges necessitates not only adequate funding, appropriate technologies, and the formulation of sound strategies but also a profound understanding of human social behavior. Recognizing the pivotal roles of cooperation, social norms, and culture in shaping human decision-making processes is also vital to devising comprehensive solutions (7–10). In addressing the role of these factors in responding to environmental challenges, valuable insights can be gleaned from the experiences of past societies documented in the ethnographic record (11–14). Many of these societies have encountered a range of environmental shocks, natural disasters, and climate instability. Remarkably, some have not only weathered these challenges but have also thrived by forging specific economic, institutional, and cultural practices that exemplify social resilience (15). Developing a theoretical understanding of how diverse cultural groups have navigated various environmental shocks and the cultural adaptations they have devised can potentially greatly contribute to the formulation of contemporary sustainability practices and policies. Environmental shocks are complex events with three key dimensions. First, they differ in frequency. Second, they can be classified based on their speed of onset (16), where fast-onset hazards are immediate and distinct (17), while slow-onset hazards emerge gradually, leading to ambiguity and uncertainty. Last, the severity of these hazards varies, with more severe shocks capable of causing subsistence challenges, food destruction, infrastructure loss, depopulation, social instability, and collapse. To effectively withstand these diverse shocks, societies must develop tailored social and cultural adaptations. A fundamental cultural response to environmental shocks is the enhancement of cooperation. Cooperation is ubiquitous across cultures (18–20) and a key aspect of human morality and sociality (21). It has been argued that cooperation should be higher in societies facing unpredictable varying environments (22, 23). However, the influence of additional attributes of these shocks remains inadequately understood (15, 24). Although there is some anthropological and ethnographic evidence available (15, 19, 25) that provides partial insights into this matter, a comprehensive understanding of the topic remains a subject of ongoing exploration. Addressing the relationship of cooperation to various attributes of shocks theoretically is the first goal of this paper. In addition to cooperation, cultural adaptations also include changes in cultural tightness-looseness, i.e., in the degree of norm strength and tolerance for deviant behaviors within a group (26–28). Empirical research has yielded conflicting findings regarding the relationship between tightness and the history of environmental shocks. On one hand, a cross-country study involving 33 countries (27) and an examination of the 50 United States (29) indicated that tightness is positively linked to a higher frequency of natural disasters. Conversely, an analysis of nonindustrial societies found no correlation between tightness and the frequency of natural disasters (30). These inconsistent results may arise from variations in sample types (modern vs. nonindustrial societies) and measurement methods but may also reflect the intricate nature of the relationship between cultural tightness and various features of environmental threats (24, 31). Our second objective is to establish theoretically grounded predictions regarding the patterns of these associations. Cultural adaptations may lead to different patterns of resilience (32), by which we will understand the ability of a group to adapt and effectively withstand hazardous events, such as environmental shocks, social instability, and conflicts (33, 34). We will focus on two types of resilience discussed in the existing literature (33, 35). Absorptive resilience implies current cultural adaptations and practices allow the group to remain stable, effectively withstanding shocks. In contrast, adaptive resilience involves significant cultural adjustments to cope effectively with shocks. Our third goal is to explore theoretically what patterns of resilience are expected to be observed in response to various types of shocks and what are the associated adaptations, particularly in terms of cooperation and cultural tightness-looseness. Previous theoretical work has provided insights into these issues but also has limitations. Regarding cooperation, mathematical models, often using evolutionary game theory (36), explore the impact of exogenous environmental changes. These changes are represented as random noise (37), fluctuations between two environmental states (38, 39), or seasonal variations in payoffs (40, 41). The models suggest that exogenous environmental fluctuations can promote cooperation but they tend to ignore certain vital aspects of human behavior such as social norms, diverse subsistence strategies and types of cooperation, and complex decision-making. Human-centric models investigate the role of social norms in responding to environmental shocks (42–44), with climate shocks typically represented as changes in resource endowment. Findings show that a more severe threat leads to higher cooperation and stricter punishment (42). Tight societies adopt cooperative behaviors more swiftly than do loose societies (44). However, some tight societies may initially adhere to defection during the onset of a threat but shift quickly toward cooperative behavior as the threat intensifies (44). Nevertheless, previous models make no distinction between various types of shocks (42–44), do not consider different types of resilience, or employ simplified decision-making mechanisms (e.g., genetically encoded, payoff-biased or conformity-based imitation) (42, 44). Here, we aim to improve our understanding of how various nonindustrial groups respond to different types of exogenous shocks, what cultural and social adaptations (such as cooperation and cultural tightness) are developed in the course of this process, and what types of social resilience (if any) are exhibited. To do this, we develop a human-specific model that takes into account: 1) a diverse range of cooperative activities typical for nonindustrial human societies (e.g., coordinated infrastructure, cooperative subsistence, and food sharing); 2) a decision-making mechanism based on both material and nonmaterial factors; and 3) environmental shocks differing in severity, frequency, and the speed of onset. The Model Agents and Their Actions. We consider a group of N agents over a finite number of time steps. Each agent represents a decision-making unit (e.g., a unique person or a household). We assume the group is characterized by a primarily subsistence economy.* Specifically, at each time step, an agent can simultaneously engage in three types of labor. First, an agent can make effort y ≥ 0 to independent subsistence exploiting some resource (e.g., by cultivating a plot of land). The maximum possible amount of resource that an agent can secure is Rmax. Second, an agent can contribute (i.e., x = 1) or not (i.e., x = 0) to a collective action that potentially enhances the productivity of each agent in independent subsistence. For example, agents can cooperate to build, clean, and maintain an irrigation system or drainage channels, construct and maintain dykes, hillside terraces, communal storehouses, and clear new fields (15) (for ethnographic examples, see Materials and Methods). We will refer to these activities as “coordinated infrastructure.” We assume that, typically, the importance of coordinated infrastructure increases with environmental shocks. Contributions to coordinated infrastructure do not lead to direct material benefits, but affect them indirectly by enhancing the productivity of each agent in individual subsistence. Third, an agent can contribute (i.e., z = 1) or not (i.e., z = 0) to “cooperative subsistence,” which is a type of collective action not directly related to independent subsistence and coordinated infrastructure. For example, agents can participate in cooperative fishing or hunting (for ethnographic examples of cooperative subsistence, see Materials and Methods). The maximum group benefit from cooperative subsistence is Mmax. A contribution to cooperative subsistence brings direct material benefits to agents. We assume that the two types of collective action are related indirectly through a synergistic effect, such that the effectiveness of the group effort in one type of collective action increases with increasing group effort in another and vice versa as the group becomes more effective in terms of coordination, intelligence, management, and task division (48, 49). The strength of the synergistic effect is controlled by parameter r ≥ 0. With r = 0, the effectiveness of the group effort in one type of collective action does not depend on the group effort in another, while increasing r enhances the positive effect of the group effort in one type of collective action on group effectiveness in another. In our model, agents can engage in all three types of labor simultaneously. This mirrors many nonindustrial societies (see Materials and Methods for ethnographic examples), with groups focusing on only one or two types of labor seen as specific cases within this model. The two types of collective action considered in the model differ in how they affect individual payoffs (directly or indirectly through enhancing individual productivity) and how the associated benefits change in response to environmental challenges. Overall, the model captures the variety of collective actions widespread in nonindustrial societies, from cooperative game or fish drives to the construction of infrastructural objects such as drainage canals, fences, levees, or storage pits. Note that our model does not include any aspect of formal leadership. This omission is intentional, as many nonindustrial societies in anthropological records demonstrate minimal formal political leadership. Instead, cooperative activities in these societies are largely organized informally, guided by social norms and based on residential and kinship connections (47). Because food sharing is widespread in nonindustrial societies (19, 50–56), we include it in the model. For the sake of simplicity, we postulate that the benefit from cooperative subsistence is shared equally among all group members, while benefits from independent subsistence are shared according to parameter s∈[0,1]. With s = 0, benefits from independent subsistence are not shared, while with s = 1, all the benefits produced by group members are shared equally. For simplicity, we postulate s to be the same for all agents in the group, so it can be treated as a group trait. Environmental Shocks. We consider two types of environment: A stable environment and a shock-prone environment in which environmental shocks occur randomly and independently with frequency ω. Shocks decrease material benefits both from independent and cooperative subsistence. The strengths of these negative effects are controlled by parameters β∈(0,1) and α∈(0,1), respectively. We posit that coordinated infrastructure can lessen the environmental challenges associated with independent subsistence, but as these challenges approach their peak, it becomes harder to mitigate their impact. The latter effect is represented by an increase in parameter X0. The changes in α, β and X0 happen synchronously. Each shock lasts T time steps and is characterized by its severity σ determining maximum values of α, β and X0 observed during the shock and onset duration τ capturing the number of rounds required for reaching the maximum values of α, β and X0 (Fig. 1A). Further, we will consider four main types of shock: 1) mild, fast-onset; 2) mild, slow-onset; 3) severe, fast-onset; and 4) severe, slow-onset (for more details, see Materials and Methods). (A) Schematic representation of the three phases of a shock affecting environmental parameters α, β, and X0. During the onset phase (which lasts τ rounds), parameters synchronously increase linearly from the baseline value toward the maximum (determined by severity σ). During the plateau phase, they remain constant. At the termination phase, they decline linearly to the baseline values. (B) Schematic representation of the decision-making process of each agent in the model. Decision-Making. At each time step, agents update their decisions (x,y,z) independently with probability μ∈(0,1), aiming to maximize their utility function using myopic best response. The utility function u(x,y,z) encompasses material considerations [i.e., material payoff π(x,y,z)], social norms (conformity), and psychological rewards associated with participation in collective actions. Using this norm-utility approach (57–60) allows us to better understand the mechanisms of individual decision-making by controlling for the weights of various material, social, and normative factors. It also gives us greater flexibility in studying different patterns of human behavior. For example, with very strong conformity, agents tend to align their decisions with the average actions of groupmates, while if the weights of all nonmaterial factors are equal to zero, agents maximize their material payoffs. Cultural Tightness. By definition, tight groups exhibit high rates of compliance with social norms and a low tolerance of deviant behavior, while loose groups are characterized by low rates of compliance with norms and a high tolerance of deviant behavior (27). In tight societies, individuals place greater emphasis on the behavior of others, while in loose societies, individuals tend to act on their own values (61, 62). In social psychology, social norms are categorized as descriptive (what people actually do) and injunctive (what people ought to do) (63, 64). Since the original tightness-looseness scale measures what people actually do (27), we concentrate on descriptive norms and manipulate cultural tightness using parameter A, which quantifies the strength of conformity to observed peers’ actions [c.f. Pan et al. (44)]. We model the effects of conformity as a quadratic term in the utility function, such that higher deviations from the average group behavior result in greater disutility (65–68). Social Resilience. In our model, societal state is characterized by well-being, i.e. the average payoff to individuals π¯, the average effort to independent subsistence y¯, and total efforts to coordinated infrastructure, X=∑i=1nxi, and cooperative subsistence, Z=∑i=1nzi. We say that a group demonstrates resilience to environmental disturbances if its long-term well-being in nonshock periods of the shock-prone environment is not lower than in the stable environment. Note that well-being and resilience are not interchangeable concepts: Well-being characterizes the group’s performance, whereas resilience refers to changes in long-term well-being in shock-prone compared to stable environments. Resilience is absorptive if shocks lead to some fluctuations around the values of X,y¯,Z observed in the stable environment. Conversely, resilience is adaptive if shocks lead to a transition to new significantly different values of X, y¯, or Z. Parameters of the model are summarized in Table 1. The structure of the model is illustrated in Fig. 1B. For more information on our modeling approach see Materials and Methods. Table 1.Main model variables, functions, and parametersSymbolsTheir meaningVariables x i Individual contribution to coordinated infrastructure (xi=1 or xi=0) y i Labor effort to indep. subsistence (yi≥0) z i Individual contribution to cooperative subsistence (zi=1 or zi=0)Functions X Group contribution to coord. infrastructure Z Group effort to cooperative subsistence π i Material payoff, Eq. 8 u i Utility function, Eq. 9Parameters N Number of agents Rmax Maximum resource from indep. subsistence Mmax Maximum group benefit from coop. subsistence s Sharing rate (independent subsistence) r Synergistic parameter A Degree of cultural tightnessEnvironm.β, αStrengths of forces preventing agents from obtaining maximum benefitsparameters X 0 Half-effort parameter related to coordinated infrastructureShock ω Frequencydimensions σ Severity τ Onset time Main model variables, functions, and parameters Results Groups’ Response to New Shocks. To examine how groups differing in the degree of cultural tightness A respond to environmental shocks in terms of cooperation (captured by variables X and Z), effort to independent subsistence y¯, and well-being (captured by variable π¯), we used agent-based simulations. In each run, we first let the system reach an equilibrium in the stable environment, and then randomly introduce shocks. We will focus on the case of fairly low initial levels of cooperation in the stable environment to examine whether high levels of cooperation can emerge in response to shocks. (When starting with high levels of cooperation, meaning the necessary cooperative practices are already in place, tight groups prove to be among the most effective in any environmental context.) The effects of cultural tightness. Here, we examine how the degree of cultural tightness A shapes group responses to environmental shocks. As mentioned earlier, we will focus on four main types of shocks: 1) mild, fast-onset; 2) mild, slow-onset; 3) severe, fast-onset; and 4) severe, slow-onset (see Materials and methods for details). By analyzing long-term behavior of groups differing in cultural tightness, we identify five model dynamics illustrated in Fig. 2. First, groups with loose cultures (with small A values) often show moderate cooperation and minimal cultural adjustment to a shock-prone environment, illustrating absorptive resilience (Fig. 2A). Second, tight groups (with very large A values) may experience cooperation collapse and well-being decline, indicating a lack of resilience (Fig. 2E). In groups of intermediate tightness, shocks can lead to three additional outcomes characterized by shifts to highly cooperative states: complete engagement in coordinated infrastructure (X = N) with relatively low contributions to cooperative subsistence Z (Fig. 2B); full participation in cooperative subsistence (Z = N) with relatively low coordinated infrastructure contributions X (Fig. 2D), and total involvement in both activities (X=Z=N) (Fig. 2C). We will call these three outcomes adaptive resilience types 1, 2, and 1+2 (summarized in Table 2). Groups with intermediate tightness can demonstrate adaptive resilience in response to shocks because they are loose enough to initiate the necessary changes (unlike very tight groups) and tight enough to reinforce and stabilize the shift to a new cooperative state, making it irreversible (a trait missing in very loose groups). After these shifts, persistently high cooperation (i.e. large values of X or Z or both) benefits the groups even during periods without shocks. Examples of five types of model dynamics. (A) Environmental shocks lead to fluctuations around the existing equilibrium (absorptive resilience, A = 0). (B) Shocks trigger a transition to full participation in coordinated infrastructure X = N (adaptive resilience type 1, A = 0.6). (C) Shocks trigger a transition to full participation in coordinated infrastructure X = N and cooperative subsistence Z = N (adaptive resilience type 1+2, A = 0.7). (D) Shocks trigger a transition to full participation in cooperative subsistence Z = N (adaptive resilience type 2, A = 1). (E) Shocks undermine cooperation and well-being (no resilience, A = 2). Parameters used describe the case of severe, fast-onset shocks (Table 5). Table 2.Types of adaptive resilienceGroup characteristics in the shock-prone environmentType 1Full engagement in coord. infrastructure (X = N), low ZType 2Low X, full participation in cooperative subsistence (Z = N)Type 1 + 2Total involvement in both cooperative activities (X = Z = N)Each type is characterized by a shift from a state with low cooperation in the stable environment (with low X and Z) to a substantially different state in the shock-prone environment. For illustrations see Fig. 2B–D. Types of adaptive resilience Each type is characterized by a shift from a state with low cooperation in the stable environment (with low X and Z) to a substantially different state in the shock-prone environment. For illustrations see Fig. 2B–D. The mechanisms behind the emergence of adaptive resilience types 1 and 2 differ significantly. For type 1, the growth in group contribution X toward coordinated infrastructure is a response to the increased value of such activities during shocks. For type 2, the rise in contribution Z to cooperative subsistence occurs after the shock ends. During a shock, Z drops due to the inevitable loss of benefits from cooperative subsistence. However, when the shock ends and benefits increase, some agents start participating more in cooperative subsistence. This increased engagement in Z is then sustained and amplified by conformity, facilitating the transition. Fig. 3 illustrates the conditions for existence and frequency of the five dynamics as a function of parameter A. Severe shocks often lead to highly cooperative states, evident from the peaks in the average values of key variables at intermediate values of A. These outcomes are much less likely with mild shocks. This occurs for two reasons. First, more severe shocks cause greater material losses, prompting more agents to contribute to coordinated infrastructure to minimize these losses (thereby inducing adaptive resilience type 1). Second, more severe shocks imply greater potential benefits following shock termination which encourages agents to enhance their participation in cooperative subsistence (hereby promoting adaptive resilience type 2). Consequently, the group’s contribution X to coordinated infrastructure, the average effort y¯ for independent subsistence, and well-being π¯ generally decrease with A during mild shocks but show a hump-shaped pattern during severe shocks. This pattern, observed at intermediate A values, results from adaptive resilience types 1 and 1+2. Similarly, the group contribution Z to cooperative subsistence displays multiple peaks at intermediate A levels due to adaptive resilience types 2 and 1+2. Notably, adaptive resilience type 2 often yields lower material benefits than does type 1, as maintaining maximum group contribution to cooperative subsistence during shocks may not be as advantageous. Effects of the degree of tightness A on (A) the group contribution X to coordinated infrastructure; (B) the average effort y¯ to independent subsistence; (C) the total contribution Z to cooperative subsistence; and (D) the average payoff π¯ for four different types of shocks (Table 5). Shown are averages in different runs (circles) and across all runs (solid black line). 100 runs each of 500 time steps for each parameter combination. Results in each run are averages based on the last 125 of time steps. Dashed black curves show the average values of these characteristics in the absence of shocks. Fig. 3 also demonstrates that mild shocks lead to lower average values of y¯, and π¯ compared to a stable environment, especially in groups with a tight culture. In looser groups, mild shocks increase the average engagement X in coordinated infrastructure relative to that in the stable environment, while, in tighter groups, X decreases. Additionally, mild shocks can boost the average group contribution Z to cooperative subsistence at intermediate A values, which is a result of adaptive resilience type 2. Notably, in groups with intermediate tightness, severe shocks significantly raise the average values of all model variables, suggesting that these groups may perform better under severe shocks than in the stable environment. Overall, our findings indicate that loose groups tend to be more effective against mild shocks (exhibiting absorptive resilience; this outcome is more marked for slow-onset shocks), while groups with intermediate tightness are typically the most effective against severe shocks (demonstrating adaptive resilience types 1 or 1+2). The effects of dimensions of shocks. So far we focused on how the degree of tightness A shapes a group’s response to four specific types of shocks (specified in Table 5). Here, we summarize the effects of severity σ, onset time τ, and frequency ω) of shocks on model outcomes. Our results show that reducing the onset duration τ, increasing the severity σ, or raising the frequency of shocks ω does not influence cooperation yet lowers well-being in overly tight groups. In these groups, conformity and tightness favor low-cooperation states, even with shocks (as we consider initially low cooperation levels). In contrast, in loose societies, reducing τ, increasing σ, or increasing ω slightly raises X as coordinated infrastructure becomes crucial with more noticeable shocks. Simultaneously, such changes decrease Z due to their negative effects on benefits from cooperative subsistence. This also leads to decreases in y¯ and π¯. For more information, refer to SI Appendix, Figs. S1–S16. Groups with intermediate tightness show interesting patterns. In such groups, the independent subsistence efforts and the levels of well-being can show hump-shaped relationships with severity σ and frequency ω (i.e., they rise initially but begin to decline after reaching a certain threshold). Cooperation levels can either increase with σ and ω or exhibit a hump-shaped dependence on these parameters. Additionally faster onset speeds τ can boost cooperation and well-being. These patterns stem from adaptive resilience and the tendency of more pronounced shocks to more likely initiate a shift to new, highly cooperative states, particularly those with a high contribution X to coordinated infrastructure. The marked increase in cooperation counters the adverse effects of shocks by improving group efficiency and creating extra material benefits. This offsets the material losses from shocks, allowing groups to achieve higher material payoffs than with less pronounced shocks or even without shocks. However, when shocks become too severe and frequent, they may hinder this shift if it is not made quickly, reducing cooperation and well-being. For more information, see SI Appendix, Figs. S1–S16. The effects of other model parameters. Fig. 4 shows how parameter A and the synergistic parameter r work together. Increasing r boosts cooperation (X and Z) and enhances effort y¯ in independent subsistence and well-being π¯. High synergistic effect r is important for adaptive resilience but not always required. For example, with severe shocks, adaptive resilience type 1 can occur even without synergistic effects, although within a fairly narrow range of tightness A. Raising r widens the range of A values where adaptive resilience type 1 appears and also makes this outcome more frequent. This suggests that high group contributions to coordinated infrastructure in response to shocks are more likely in the presence of cooperative subsistence, even though it is not essential for such a response. However, without shocks, the synergistic effect alone is not enough to lead to highly cooperative states. Effects of the strength of conformity A and the synergistic parameter r on (A) the group contribution X to coordinated infrastructure; (B) the average effort y¯ to independent subsistence; (C) the total contribution Z to cooperative subsistence; and (D) the average payoff π¯. Shown are the averages based on 100 runs each of 500 time steps for each parameter combination. Results in each run are averages based on the last 125 of time steps. In SI Appendix, we explore the impact of other parameters. Raising the benefit parameters Rmax or Mmax enhances cooperation (X and Z, respectively) and increases payoffs (SI Appendix, Figs. S17 and S18). Notably, increasing Rmax or Mmax widens the range of tightness A values where transitions to highly cooperative states (adaptive resilience) become more prevalent. Specifically, tighter societies can better handle environmental shocks if the maximum resource available for independent subsistence Rmax is large, though societies with medium tightness A still perform best. Generally, with high Rmax, moderately tight societies excel even under mild slow-onset shocks, while loose societies fare better even against severe fast-onset shocks with low Rmax. This implies that as independent subsistence benefits, like those from farming or horticulture, increase, a greater degree of tightness is needed to effectively withstand shocks. In SI Appendix, section S1.3 we also inspect the effects of the sharing rate s and the group size N confirming the robustness of our results to small changes in these parameters. In fact, increasing the group’s size undermines cooperation and lowers well-being. However, if an increase in N is coupled with a corresponding increase in the benefit parameters, our main findings on group responses to shocks are robust to changes in N. Groups’ Response to Regular Shocks. So far, we analyzed how groups with a constant degree of tightness A react to new environmental shocks, starting from a baseline of low cooperation levels. We now seek to determine whether changes in tightness can yield higher payoffs after a group reaches a stochastic equilibrium in a shock-prone environment (i.e., when shocks become regular). To implement this, we examine the same four types of shocks mentioned earlier (refer to Table 5) and use the previously studied stochastic equilibria reached by loose groups (with A = 0), tight groups (with A = 2), and groups with intermediate tightness (with A∈{0.6,0.7,0.8,0.9,1}) in response to these shocks as initial values in our simulations. Given the initial values and a type of shocks, we then identify associations between cooperation/well-being and the degree of tightness in the regular-shocks environment. In SI Appendix, Figs. S22–S28 we show that our main findings related to new shocks are stable to these changes in tightness. Specifically on average, 1) cultural looseness is necessary to respond effectively to regular mild slow-onset shocks; and 2) intermediate levels of cultural tightness are necessary to effectively withstand regular severe shocks. Interestingly, our results indicate that in one scenario excessive tightness can be very beneficial in coping with regular shocks. Specifically, in response to new shocks (typically severe), a group with moderate tightness can reach the equilibrium with total involvement in both collective activities (adaptive resilience 1+2). Once the equilibrium is reached, increasing tightness has no effect on cooperation and payoffs, whereas decreasing tightness reduces them. This outcome indicates that excessive tightness may be as effective as moderate tightness in withstanding regular severe shocks if highly cooperative practices are already established and the group merely needs to maintain them. Moreover, one can hypothesize excessive tightness to be the most effective against regular severe shocks if there are some additional forces favoring it (e.g., some types of intergroup conflicts). Overall, the frequency of outcomes demonstrating adaptive resilience 1+2 in response to new shocks determines the average effectiveness of tight groups in the face of these shocks becoming regular. With respect to withstanding severe fast-onset shocks, very tight societies are less effective than are societies with moderate tightness, but fare better than do loose groups (SI Appendix, Fig. S27). Otherwise, tight groups handle regular shocks worse than do groups with other types of cultures (SI Appendix, Figs. S22–S28). Discussion Using a mathematical model we have explored how cultural groups may respond to different environmental shocks, the resulting cultural and social adaptations, and the evolved types of social resilience. Our model has introduced elements often overlooked in prior research, including, 1) environmental shocks that vary by severity, frequency, and onset speed; 2) a spectrum of cooperative behaviors observed in human societies, including coordinated infrastructure, cooperative subsistence, and food sharing; and 3) a decision-making process informed by material factors, personal values, and social norms. Our theoretical framework enhances comprehension of how culture can foster adaptability and resilience of human societies to environmental challenges. Main Findings. Our main findings are summarized in Table 3. First, our results reveal that societies need a certain degree of cultural looseness to effectively manage mild slow-onset environmental shocks. This cultural looseness enables societies to maintain a moderate level of cooperation with minimal need for significant changes to their social and economic structures. We refer to this ability as “absorptive resilience.” The principle is quite simple: When faced with minor challenges, a society does not have to overhaul its existing systems. Instead, it just needs the capacity to make small, timely adjustments to cope with the challenges. Our findings also indicate that cooperation and well-being decline with increasing tightness in the case of mild, slow-onset shocks. Table 3.Summary of main findingsParameterMain resultsTightness A• Societies need cultural looseness to manage mild slow-onset shocks effectively. Such shocks lead to fluctuations around the existing equilibrium, resulting in absorptive resilience and moderate levels of cooperation. • Societies need an intermediate level of cultural tightness to manage severe shocks effectively. Such shocks prompt major cultural shifts, fostering adaptive resilience and high cooperation. • Societies with very high levels of tightness struggle to adapt to new shocks but can exhibit high levels of cooperation and well-being in response to regular severe shocks.Severity σ, onset time τ, and frequency ω (new shocks)• In loose societies, increasing shocks (i.e., reducing the onset time, increasing severity, or raising frequency) results in a slight increase in coordinated infrastructure but reduces all other types of economic activities and well-being. This is a consequence of absorptive resilience. • In moderately tight societies, well-being can exhibit a hump-shaped dependence on severity and frequency (i.e., they rise initially but begin to decline after reaching a certain threshold). Cooperation levels can either increase monotonically with severity and frequency or show a hump-shaped dependence on these parameters. Faster onset shocks can boost cooperation and well-being. These patterns stem from adaptive resilience and the tendency of more pronounced shocks to be more likely to initiate a shift to new, highly cooperative states. • In tight societies, increasing shocks does not affect cooperation and lowers well-being.Synergistic effect r• With shocks, a stronger synergistic effect widens the range of tightness values where the transition to states with high levels of coordinated infrastructure appears. • Under severe shocks, such a transition can occur even without the synergistic effect within a narrow range of moderately tight societies. • Without shocks, the synergistic effect alone is not enough to lead to highly cooperative states.Benefit param. Rmax, Mmax• Increasing the benefit parameters enhances cooperation and well-being, widening the range of tightness values where adaptive resilience becomes more prevalent. • As the maximum resource available for independent subsistence Rmax (e.g., for farming or horticulture) increases, a greater degree of tightness is needed to effectively withstand shocks.Group size N• Increasing group size undermines cooperation and lowers well-being. • If an increase in the group size is coupled with a corresponding increase in the benefit parameters, our main findings on group responses to shocks are robust to changes in the group size. Summary of main findings Second, our research shows that societies need an intermediate level of cultural tightness to manage severe shocks effectively. These shocks prompt major cultural shifts, fostering adaptive resilience and high cooperation. Having some degree of looseness (relative to very tight groups) allows groups to initiate a necessary change. Moreover, cultural tightness in these groups reinforces the transition, once the majority of agents have already changed their behavior, and ensures stability after the shift to a new social equilibrium making it irreversible. Once these changes take place, cooperation is persistently high, which benefits societies even in stable conditions, leveraging good environmental situations (69). Generally, the model shows that under severe shocks, cooperation and well-being peak at moderate levels of cultural tightness. The above conclusion is consistent with some empirical findings. Specifically, a study of 32 modern nations shows that many indicators of societal well-being exhibit hump-shaped dependencies on the degree of cultural tightness, indicating that excessive tightness and excessive looseness contribute to poorer psycho-social, health, economic, and political outcomes, as well as lower overall well-being at the national level (31). However, it is important to note that Harrington et al. (31) concentrated on modern societies, whereas our model is tailored for small-scale, nonindustrial groups (see the discussion below). Additionally, they did not account for the types of natural hazards prevalent in the societies they sampled. Our findings suggest that future research should specifically examine nonindustrial groups, focusing separately on societies predominantly exposed to severe shocks—where we predict well-being will show a hump-shaped relationship with cultural tightness—and those mainly experiencing mild, slow-onset shocks—where we anticipate well-being will decrease with increased tightness. Third, our findings indicate that very tight societies can struggle to adapt to environmental shocks, suggesting that extreme tightness can be disadvantageous. Our model reveals that in the face of new challenges, tight societies fare poorly, particularly when the potential benefits from cooperation are not very high. However, these societies tend to perform better when dealing with regular severe fast-onset shocks, typically achieving higher levels of cooperation and well-being than do loose societies, though not as high as those with moderate tightness. It is noteworthy that, as our model allows for multiple coexisting stochastic equilibria, in some of them, tight groups can exhibit cooperation and well-being levels similar to moderately tight groups. This observation leads us to propose a hypothesis: In specific scenarios not yet fully explored in our model, such as those involving frequent intergroup conflicts (27, 30), extremely tight cultural norms might actually be most advantageous. This could be particularly true in environments that frequently experience severe and fast-onset challenges. Overall, the results indicate that moderately tight groups are more prone to rapid transformation in response to environmental shocks, especially with severe, fast-onset hazards. This is because these groups are loose enough to initiate the necessary changes (unlike very tight groups) and tight enough to reinforce the shift (unlike very loose groups). These findings deepen our understanding of how culture and environment shape a group’s acceptance/resistance to change (70). Fourth, our model demonstrates that larger shock severity, frequency, and rapid onset can paradoxically boost both cooperation and well-being in moderately tight societies. What happens is that more salient shocks are more likely to trigger significant cultural adjustments leading to settling into a new, more beneficial state characterized by a fairly high cooperation level. Ethnographic evidence from 90 societies supports this, revealing more technological adaptations, such as irrigation systems, drainage channels, dikes, cattle ponds, hillside terraces, and wells, in societies facing fast-onset shocks (15), which aligns with our model’s predictions. Fifth, our research indicates that societies with cooperative subsistence practices, like cooperative hunting or fishing, are more likely to demonstrate a transition to a very high participation in coordinated infrastructure activities in response to shocks. Likewise, existing coordinated infrastructure activities can boost the evolution of cooperative subsistence. This dynamic is typically seen in societies with moderate tightness facing severe shocks, particularly when there is a synergy between different collective actions by group members. Therefore, we expect societies employing diverse subsistence strategies and exhibiting adaptive resilience to be more common in regions experiencing severe climate-related disturbances. Sixth, our model can be easily generalized to account for the effects of religious rituals and beliefs. Specifically, religious rituals foster group cohesion (71–73), which can be captured in the model through increased synergistic effect. As a result, emerging rituals during the onset phase of a shock may act as an additional force to trigger significant cultural adjustments, including the rapid development of nonreligious cooperative coping mechanisms (e.g., coordinated infrastructure practices). Since our model predicts that the above transition is less likely to be observed during slow-onset shocks, such rituals may offer more evolutionary benefits during slow-onset shocks compared to fast-onset ones. This is well in line with the results of Pierro et al. (15) showing that religious coping mechanisms are associated with slow-onset shocks, and an increased number of religious coping mechanisms is associated with an increased number of nonreligious coping mechanisms. Finally, while tailored for small-scale nonindustrial groups, the model is expected to have relevance for larger societies. First, in the model, the group size N represents the number of decision-making agents, which, depending on the empirical context, can be interpreted as the number of individuals in a small group, the number of households in a village, or the number of villages in a system of villages. This implies the actual number of individuals in a group can be 1 to 2 orders of magnitude greater than the number of agents N. Second, small-scale societies share similarities with larger ones, especially with small groups within larger societies. Indeed, even larger societies consist of many small egalitarian communities and groups that share common interests and behave much like small-scale societies. Examples include subcastes in India, rural villages in China and Mexico, and even volunteer organizations in the industrial US (74–77). Overall, we believe the model can be applied to large groups to at least examine the dynamics of some smaller geographic, cultural, or political subgroups. For example, the model captures some basic features shared by many human societies regardless of their level of complexity (such as the role of collective action in mitigating natural hazards) and can be directly applied to some local subgroups (e.g., to some indigenous or rural communities). Yet we also caution that the model does not take into account key aspects of modern societies including wage labor, the market economy, various political structures, government interventions, online behavior, and many others. These limitations can potentially be addressed in future work. Limitations and Future Work. Our model has several limitations. Participation in collective actions and food sharing are crucial forms of cooperation and common strategies to manage various environmental hazards. Our model primarily focuses on collective actions while accounting for the food sharing rate. Following Bowles (78), we model food sharing as a proportional deduction at rate s from each agent’s individual payoffs, with the proceeds distributed equally among all group members. However, treating the food sharing rate s as a group-level parameter is a simplification which should be addressed in future work. Potential model generalizations could include introducing individual heterogeneity in s or allowing individual food sharing rates to change endogenously. Another simplification is the model’s exclusion of how individual choices impact the environment, such as overharvesting leading to degradation (79). Future extensions could integrate resource depletion and resultant ecoevolutionary dynamics (79–85). Additionally, the model overlooks inventions (86–88) and cultural knowledge accumulation as responses to previous hazards. Our model also overlooks group internal diversity and heterogeneity. Future research could include variation in the individual perception of the importance of material and normative factors in decision-making (67), social network dynamics (89–93), economic inequality (94), and political institutions and power dynamics (95–97) to more accurately simulate societal responses to environmental shocks. Here, we examine how societies cope with and adapt to various types of environmental shocks employing the norm-utility approach to delve into individual decision-making. However, alternative strategy update methods like payoff-biased or conformity-biased imitation can also be investigated. Additionally, the model does not address intergroup dynamics such as migration (98, 99), trade, intergroup cooperation (100, 101), and conflict (102–104), which are vital for understanding the evolution of cooperation and cultural tightness. These areas are identified for future investigation. Conclusion. In this paper, we have shed light on the complex ways in which nonindustrial societies adapt to environmental shocks, introducing testable hypotheses about the interplay between such shocks and key social dynamics. Our findings, which illustrate the impact of various shocks on crucial social attributes like cooperation and cultural tightness, set the stage for designing more effective, culturally specific sustainability strategies. Such strategies would be tailored to the unique cultural, social, and institutional contexts of human groups, enhancing their resilience in the face of global environmental changes. This work not only contributes to the theoretical understanding of societal responses but also has practical implications for developing adaptive policies that could lead to more sustainable practices worldwide. Materials and Methods Relevant Ethnographic Examples. Ethnographically recorded examples of coordinated infrastructure include: construction and maintenance of a sophisticated system of terraces by the Burusho (15, 105); construction and maintenance of a complex mechanism that combined drainage of surplus water through ditches with composting mounds in response to frequent floods by the Kapauku (15, 106); building levees to control land erosion caused by floods, and digging wells to supply the necessary water for irrigation by the O’odham people (107); organization of water reserves by the Koreans in response to droughts (108); coordinated actions to clear new fields by burning them and communal works to clear roads in Garo communities (109); digging canals and clearing new fields by the Mapuche of Chile (110); as well as construction and cleaning of irrigation canals by the Kazakhs (111, 112). Ethnographically recorded examples of cooperative subsistence include: hunting battues (i.e., a form of hunting in which game is driven out of the woods by beating the bushes) and seasonal communal game drives by the Lozi people (113, 114); cooperative fish poisoning and communal net-hunting among the Bemba people (115); cooperative hunting for pigs during the rainy season among the Andamans (116); as well as group hunting for peccaries and cooperative net fishing among the Kuna (117, 118). Ethnographic examples of both coordinated infrastructure and cooperative subsistence are discussed below and are summarized in Table 4. Table 4.Some ethnographically recorded examples of both, coordinated infrastructure and cooperative subsistenceCultureIndependent subsistenceCoordinated infrastructureCooperative subsistenceLoziLand cultivationDigging drainage canalsCooperative game drivesBembaGardeningCooperative tree-cuttingCooperative fish poisoningAymaraAgricultureIrrigation worksCooperative hunt for vicuñasWolofCultivation of rice/millet/cottonConstruction of fencesCommunal fish drivesSomaliHerding/land cultivationDigging ponds/construction of fences/tree fellingCooperative antelope drivesHidatsaGrowing corn/beans/squash/sunflowersConstruction of storage pitsCooperative buffalo huntsTukanoManioc cultivationCooperative clearing of new fieldsCooperative tapir hunt Some ethnographically recorded examples of both, coordinated infrastructure and cooperative subsistence The Lozi people participated in digging drainage channels. In addition, they practiced cooperative hunting in a hunting-party called lisulo. Men on foot and with dogs drove bucks into the water, where they were killed by other men from dugouts or by men who entered the water. The meat and skins were shared among all the hunters (113).The Bemba people organized work groups known as ukutumya to do tree-cutting. In addition, the whole village participated in another cooperative activity—fish poisoning (ukusungilo buba) (115).Among the Aymara of Peru, members of an ayllu group or clan participated in collective house construction and irrigation works. Individual families performed agricultural activities such as plowing, sowing, and harvesting. In addition, up to three hundred individuals participated in a vicuñas hunt, forming a large circle and slowly moving toward the center to round up the animals (119).The Wolof formed cooperative work groups to build fences around rice, millet, and cotton fields to prevent wild animals from eating or trampling the crops. Also, the Wolof participated in seasonal communal fish drives (120).Some of the Somali tribes were engaged in both herding and land cultivation. All able-bodied men from a settlement cooperated to dig out ponds, construct fencing, and felling trees. In addition, the Somalis participated in cooperative hunts in which they beat the bushes to drive animals such as antelopes into a semicircle of preset nets (121, 122).Among the Hidatsa of the northern Great Plains, cooperative summer buffalo hunts typically involved the entire village, and the meat was shared among all group members. During the summer, the Hidasta worked in the gardens growing corn, beans, squash, and sunflowers. They also participated in the cooperative construction of storage pits for the crops (123).The Tukano people participated in communal works to clear new plots of land. In terms of manioc cultivation on these plots, each household was autonomous. In addition, the Tukano people organized hunting parties with a strong commitment to teamwork. A hunting party tried to track down and catch tapirs at a salt lick. The meat was typically brought to the settlement and shared among group members (124). The Lozi people participated in digging drainage channels. In addition, they practiced cooperative hunting in a hunting-party called lisulo. Men on foot and with dogs drove bucks into the water, where they were killed by other men from dugouts or by men who entered the water. The meat and skins were shared among all the hunters (113). The Lozi people participated in digging drainage channels. In addition, they practiced cooperative hunting in a hunting-party called lisulo. Men on foot and with dogs drove bucks into the water, where they were killed by other men from dugouts or by men who entered the water. The meat and skins were shared among all the hunters (113). The Bemba people organized work groups known as ukutumya to do tree-cutting. In addition, the whole village participated in another cooperative activity—fish poisoning (ukusungilo buba) (115). The Bemba people organized work groups known as ukutumya to do tree-cutting. In addition, the whole village participated in another cooperative activity—fish poisoning (ukusungilo buba) (115). Among the Aymara of Peru, members of an ayllu group or clan participated in collective house construction and irrigation works. Individual families performed agricultural activities such as plowing, sowing, and harvesting. In addition, up to three hundred individuals participated in a vicuñas hunt, forming a large circle and slowly moving toward the center to round up the animals (119). Among the Aymara of Peru, members of an ayllu group or clan participated in collective house construction and irrigation works. Individual families performed agricultural activities such as plowing, sowing, and harvesting. In addition, up to three hundred individuals participated in a vicuñas hunt, forming a large circle and slowly moving toward the center to round up the animals (119). The Wolof formed cooperative work groups to build fences around rice, millet, and cotton fields to prevent wild animals from eating or trampling the crops. Also, the Wolof participated in seasonal communal fish drives (120). The Wolof formed cooperative work groups to build fences around rice, millet, and cotton fields to prevent wild animals from eating or trampling the crops. Also, the Wolof participated in seasonal communal fish drives (120). Some of the Somali tribes were engaged in both herding and land cultivation. All able-bodied men from a settlement cooperated to dig out ponds, construct fencing, and felling trees. In addition, the Somalis participated in cooperative hunts in which they beat the bushes to drive animals such as antelopes into a semicircle of preset nets (121, 122). Some of the Somali tribes were engaged in both herding and land cultivation. All able-bodied men from a settlement cooperated to dig out ponds, construct fencing, and felling trees. In addition, the Somalis participated in cooperative hunts in which they beat the bushes to drive animals such as antelopes into a semicircle of preset nets (121, 122). Among the Hidatsa of the northern Great Plains, cooperative summer buffalo hunts typically involved the entire village, and the meat was shared among all group members. During the summer, the Hidasta worked in the gardens growing corn, beans, squash, and sunflowers. They also participated in the cooperative construction of storage pits for the crops (123). Among the Hidatsa of the northern Great Plains, cooperative summer buffalo hunts typically involved the entire village, and the meat was shared among all group members. During the summer, the Hidasta worked in the gardens growing corn, beans, squash, and sunflowers. They also participated in the cooperative construction of storage pits for the crops (123). The Tukano people participated in communal works to clear new plots of land. In terms of manioc cultivation on these plots, each household was autonomous. In addition, the Tukano people organized hunting parties with a strong commitment to teamwork. A hunting party tried to track down and catch tapirs at a salt lick. The meat was typically brought to the settlement and shared among group members (124). The Tukano people participated in communal works to clear new plots of land. In terms of manioc cultivation on these plots, each household was autonomous. In addition, the Tukano people organized hunting parties with a strong commitment to teamwork. A hunting party tried to track down and catch tapirs at a salt lick. The meat was typically brought to the settlement and shared among group members (124). Note that most of the above ethnographic examples were taken from eHRAF World Cultures database (https://ehrafworldcultures.yale.edu/). Additional Details on the Modeling Approach. Benefits from independent subsistence. We postulate that productivity of agent i making effort yi>0 to independent subsistence is described by a Cobb–Douglas production function with diminishing returns to labor (125):[1]pi=Ryiγ, where R is the available resource (defined below) and γ∈(0,1) is the elasticity parameter showing the responsiveness of productivity due to a change in labor. Then, benefit from independent subsistence is[2]bi=(1−s)pi+sp¯, where s∈[0,1] is the sharing rate, and p¯ is the average productivity in independent subsistence. With s = 0, there is no sharing so each agent keeps the produced benefit (i.e., bi=pi), while with s = 1, all the benefits produced by group members are shared equally (i.e., bi=p¯). Sharing can be treated as part of a feast, death, or religious ceremony (19, 109), or as something that happens on a regular basis (116, 117, 126, 127). We assume that the actual resource R available to each agent can be smaller than maximum possible resource Rmax due to environmental conditions. However, by contributing to coordinated infrastructure, agents can mitigate the negative effects of environments. The success of the group in the coordinated infrastructure is an increasing function of the total group effort X=∑j=1Nxj (128):[3]P=XX+X~0, where X~0>0 is a half-effort, which specifies the level of group contribution at which the group success equals to 1/2. Note, Eq. 3 describes a public good game with nonlinear production function capturing diminishing marginal return of group effort (129, 130). Finally, the resource R is defined as[4]R=[1−β(1−P)]Rmax, where β∈[0,1] captures the strength of forces preventing agents from obtaining the maximum resource Rmax. Benefits from cooperative subsistence. Agents can participate (i.e., zi=1) or not (i.e., zi=0) in cooperative subsistence potentially leading to the group benefit M. The group success Z/(Z+Z0~) is an increasing function of the number of participants Z=∑j=1Nzj (131), where Z~0>0 is a half effort. The individual benefit from cooperative subsistence is[5]bi′=MNZZ+Z~0. Mathematically, we model cooperative subsistence as a generalized public goods game called “us vs. nature” game (128, 131). We assume the maximum group benefit is Mmax>0, and the actual benefit M can be less than Mmax due to environmental conditions. Specifically,[6]M=(1−α)Mmax, where α∈[0,1] is the strength of the environmental effects. Synergistic effect. We assume there is a synergistic effect between the two types of collective actions in the model (i.e., between coordinated infrastructure and cooperative subsistence) so that the effectiveness of the group effort in one type of collective action increases with increasing group effort to another and vice versa. This effect is captured by the dependence of half-effort parameters X~0 and Z~0 on the group efforts Z and X respectively:[7]X~0=X0(1+Z/N)r,andZ~0=Z0(1+X/N)r, where X0 and Z0>0 are baseline half effort parameters, and r ≥ 0 is a synergistic parameter. Note that if there is a synergistic effect (i.e., r > 0), increasing X decreases Z~0, while increasing Z decreases X~0, so it becomes easier to succeed in collective actions. Material payoffs. The material payoff of agent i is[8]πi(xi,yi,zi)=bi+bi′−C, where C=cxxi+cyyi+czzi is the total cost function, cx, cy, and cz>0 are individual cost parameters related to three types of actions. Individual decision-making. We postulate agents make their decisions (xi,yi,zi) based on the utility function using myopic best response. The utility function of a focal agent encompasses: 1) the material payoff; 2) nonmaterial rewards associated with participation in collective actions, i.e., agents who participate in coordinated infrastructure or cooperative subsistence can get more respect and have a better reputation than those who do not (115); and 3) social norms, i.e., agents tend to align their actions with the average action of group-mates and may experience psychological discomfort [internalized (132–134) or caused by ostracism (135)] if they deviate significantly:[9]ui(xi,yi,zi)=πi(xi,yi,zi)⏟material payoffs+[A1xi+A1′zi]⏟Psychological reward fromparticipation in collective actions−[A2(xi−x¯)2+A2′(zi−z¯)2]⏟Conformity with others, where A1, A1′, A2, and A2′ are nonnegative parameters capturing the strengths of normative components. For simplicity, we will assume A2=A2′=A, where A is treated as the degree of cultural tightness (42, 44). Main model parameters are summarized in Table 1. Environmental shocks. We consider two types of environment: a stable environment and a shock-prone environment. Three parameters, α, β, and X0 reflect environmental changes. During non-shock periods, environmental parameters are set to their baseline values (i.e., α = 0, β = 0.4, and X0=0.25N; note β > 0 to capture regular works related to coordinated infrastructure such as building and maintaining storehouses or clearing new fields). Shocks occur randomly and independently with frequency ω and are associated with a simultaneous increase in the environmental parameters (i.e., shocks negatively affect benefits R and M, and make it more difficult to mitigate negative environmental impacts). Note that the effect of shocks on individual decisions to participate in coordinated infrastructure is twofold: 1) on the one hand, shocks provide an additional incentive to participate in order to mitigate the growing material losses (captured by β); but 2) on the other hand, shocks can inhibit participation as it also becomes more difficult to succeed in shock mitigation (captured by X0). Note, one can also assume that, together with X0, shocks increase Z0. However, this effect works in the same direction as that of α, inhibiting the contribution to cooperative subsistence. As a result, for simplicity, we model the effects of shocks on cooperative subsistence only through parameter α. Each shock lasts T time steps and consists of three phases (Fig. 1). During the onset phase which lasts τ∈{1,..,T/2} rounds, α, β, and X0 increase linearly from their baseline values toward the maximum values, which are determined by severity parameter σ∈{1,..,6} (for more details, see SI Appendix). During the plateau phase which lasts T−2τ rounds, all the environmental parameters remain constant at their maximum values. Finally, during the termination phase which lasts τ rounds, α, β, and X0 decline linearly toward their baseline values. Baseline Parameters for Numerical Simulations. In the simulations, we assume that each shock lasts T = 10 time steps and the baseline shock frequency is ω = 0.05. Baseline characteristics of four main types of shocks are shown in Table 5. We assume that initially, each agent participates in each type of collective action randomly and independently with probability 0.2. In our simulations, we use a set of baseline parameters: N = 8, Rmax=16, Mmax=56, r = 0.35, and s = 0.25. For more details on numerical simulations, see SI Appendix, section S1.1. Table 5.Baseline characteristics of four main types of shocks. For all types of shocks, their frequency is ω = 0.05Types of shocksSlow-onsetFast-onsetMildσ = 2, τ = 5σ = 2, τ = 1Severeσ = 4, τ = 5σ = 4, τ = 1 Baseline characteristics of four main types of shocks. For all types of shocks, their frequency is ω = 0.05 Supplementary Material Appendix 01 (PDF)