Knowledge Co-Creation in the Commons: Facilitating Collaboration in the Circular Economy of Plastic with “Closing the Loop” Game

Abstract

The implementation of circular solutions instead of single-use plastic requires a stronger alliance between science and the private sector, which could be achieved by means of knowledge and strategy co-creation. To investigate their benefits, this study applies common-pool resource theory in serious game design. While other serious games focused on the circular economy exist, none of them consider different modes of knowledge acquisition. To test the game’s potential, we used a convenience sample of 188 biology and geography students (researchers-to-be) from Jagiellonian University. Each of the 20 groups was divided into playing the roles of science and business actors. Two versions of the game were assessed—one with separation and one with personnel exchange. The results of the games were analyzed in relation to the structure of the shared social values as a differentiating co-factor. Science–business collaboration and environmental attitudes were evaluated using pre- and post-questionnaires. Results suggest that the successful resolution of the presented common-pool resource dilemma correlates with the introduced collaboration (p = 0.004, t value = 3.400, df = 1) and the Self-Transcendence values (p = 0.001, t value = 4.329, df = 1). We observed desirable changes in participants’ attitudes after the game (p = 0.000, χ² = 18.810, df = 1). Further investigation into gaming as a science communication tool and its long-term effects is recommended to facilitate the implementation of knowledge co-creation principles in practice.

1. Introduction

1.1. Plastic Dilemma as a Common-Pool Resource Problem

Single-use plastic products constitute 36.7% of global plastic production in 2021—more than any other end-use category [1,2]. Such products are designed with the intent for single use or for a short period before being disposed of [3]. Moreover, single-use plastics are technically difficult and costly to reprocess, and, in effect, only 9% is recycled, while the majority is incinerated [4,5]. Plastic lifecycle CO2 emissions will account for more than 12% of the remaining 1.5 °C carbon budget by 2050 [6]. The negative influence of plastic waste affects agriculture, health, and the environment in ways we are only beginning to understand [7,8,9].

Responsibility for the pollution is dispersed and, despite some institutional control at various administration levels, plastic disposal remains largely unregulated [10]. With Earth’s limited carrying capacity and open global plastic sinks in the form of faulty regulations and export to developing countries, this problem can be framed in terms of a common-pool resource [11]. This theory describes notoriously difficult-to-manage, subtractable, partially or non-renewable resources, understood either as material goods or more abstract environmental capacities that are a source of collective benefit. Limited property rights, competitive consumption, and required collaboration heighten the risk of private extraction of profits. Stakeholders often choose such a strategy even if it can substantially diminish common value generation [12].

Game theory, which is the theoretical underpinning of this concept, assumes that stakeholders have equal rational motivation to maximize expected utility. Therefore, refraining from participation in the commons is seen as giving ground to potential competitors [13,14]. A social dilemma arises where cooperation would yield more beneficial outcomes for all stakeholders but fails due to conflicting interests [15]. In the plastic crisis, the common goods, or in this context, so-called common bads, represent externalized waste. Externalization often means a legal gray area or exporting the waste to countries that are willing to provide such services [16,17].

1.2. Knowledge Co-Creation as a Postulated Solution

The circular economy is one of the frameworks developed to manage the plastic waste problem [18,19]. It is a method of reaching a sustainable society by gradually extending product lifespans to strongly reduce necessary energy consumption and limit the amount of waste released into the environment [20,21]. Although this idea has been strongly promoted, circular economy advocacy has focused on waste generation, resource use, and environmental impact. Meanwhile, the economic perspective and business incentives have been neglected, missing the concerns of key business stakeholders [22,23]. The academic world fostering innovative strategies in the circular economy tends to engage in a linear mode of communication, where knowledge transfer is supposed to occur only in one way: from academia to the public [24,25]. In particular, academia–business relations have been theorized as an Ivory Tower setting, where there is little initiative or incentive for bridging the gap between those groups [26,27]. Similarly, university-level circular economy education lacks authentic workplace interactions and systems thinking [28], while experiments with experiential learning provide promising results [29] and higher education institutions can play a major role in the transition to the circular economy, provided there is effective collaboration with industrial partners and other stakeholders [29,30]. To close the gap between the scientific and business communities, knowledge co-creation has been devised to facilitate knowledge sharing and cooperation between knowledge producers and knowledge users. It also helps create important professional associations that support joint problem-solving [24,31].

One of the methods to address these issues is to develop alternative ways of knowledge co-creation, which can be achieved using serious games designed to facilitate mutual and experiential learning [32]. Games are increasingly prevalent as both educational and research tools [33] and can help to explore social dilemmas by building on the game theory legacy [34]. This approach is especially valuable when dealing with groups with different incentives, as it has the “power to improve communication between competing stakeholders” [35]. Serious games on circular economy contribute to increasing circular economy awareness, technological understanding, and stakeholder motivations [36,37,38] as well as facilitating dialogue among various stakeholders [39] and stimulating co-development [37]. Serious games can contribute by providing shared boundary objects for diverse stakeholder groups by representing their operational models and constraints. However, serious games so far have failed to address many other issues that are especially relevant to the circular economy. These include low levels of trust, unresolved conflicts between stakeholders, difficulty reaching mutually acceptable decisions, and uneven engagement levels depending on participants’ roles and motivations [40,41,42].

Several serious games describe the circular economy; however, only a few address knowledge co-creation. Until now, no single one has combined those two topics. There have been games designed as tools to co-create circular solutions, but they did not consider the mechanisms of co-creation, such as personnel exchange, in an empirical manner [43,44]. We contribute to filling this knowledge gap by designing the Closing the Loop game as a novel methodology, both to allow detailed observation of the decision-making process and facilitate mutual and experiential learning about knowledge co-creation. We compare the effect of knowledge co-creation and an “Ivory Tower” setting on solving the common-pool resource dilemma of the plastic economy [26,27].

The main aims of this study are to assess the conditions that increase the chances of success in the proposed game and whether it changes attitudes towards science-business collaboration. These goals were achieved by conducting research that addressed the following research questions:

(1)

Does player exchange help govern common-pool resources in the Closing the Loop simulation game?

(2)

Do the values held by the participants affect the results of such a game?

(3)

Does participation in the game affect the attitudes of players towards collaboration between science and business in support of the circular economy?

(4)

Does the game affect environmentally oriented participants differently?

2. Materials and Methods

2.1. Game Model Design

The main inspiration for the game design was an investment game by Ostrom and her team [45]. This Nobel Prize-winning study demonstrated that people can abandon selfish behaviors in favor of a common goal when communication is framed as a strategy for co-creation and behavioral control. The following research design examined whether the Closing the Loop game reproduces similar dynamics when the communication flow is controlled among groups, rather than individuals. We consider two types of knowledge architectures. The first is the “Ivory Tower” [46] approach, where the processes of knowledge acquisition and management are confined by the rules provided by an organization or entity. The second is the knowledge co-creation process, where the process of knowledge acquisition and management takes place through ongoing interaction between two or more entities. In this study, these entities are represented symbolically by academia and the private sector, which exchange internal rules, experiences, and strategies.

The core of the game was based on the double circular industrial process proposed by Grodzińska-Jurczak [24]. Circular material processing is mediated there by a parallel cycle of knowledge generation, requiring synchronized implementation of non-linear schemes. Thus, two different disciplines must cooperate by supporting each other with multidirectional flows of knowledge and capital, set within the broader market context and the corresponding flow of resources.

To allow for experiential learning, players were meant to learn the rules through the gameplay itself. To make this possible during a single game session, the complexity of the game had to be limited. Drawing on second-order cybernetics, we view players as participant observers [47] trying to understand a dynamic system while changing it from the inside [48]. The same situation occurs when young scientists and entrepreneurs become members of an industrial ecosystem that requires transformation [49,50]. Because the focal point of our research was social interaction, a physical board game design—printed cards and instructions—was adopted to encourage experiential learning [51].

2.2. Simplified Description of the Game

Players are divided into two equal teams of inventors and entrepreneurs, which are located in two separate rooms. Each team is guided by different rules, but the end goals are the same: either securing personal gains or investing in a common victory (or both). Inventors can buy random (but sequenced) invention cards—six distinct types that differ in environmental and economic effects. Each round, inventors have the option to (1) voluntarily share the cards with entrepreneurs, who can activate them for economic and/or environmental benefits, or (2) sell them to gain personal benefit. The latter option earns resources for the inventors but consumes valuable time that could be used for further research. In each round, inventors can perform only one action: buy a new invention card or sell one they already own. Sharing the cards does not use up the action.

Entrepreneurs, during each round, may choose between (1) producing single-use plastics (which earns them resources), (2) engaging in slow recycling (which costs them resources), or (3) implementing invention cards, if they have received any from the inventors. Additionally, at the end of each round, entrepreneurs can pass resources to inventors, trusting that they will receive invention cards in return. The invention cards allow entrepreneurs to earn resources and/or to remove waste. Special circular economy cards can also be found; these are used to implement circular economy solutions—they constantly reduce waste levels and, when all are activated by entrepreneurs, no new waste is generated, and the game ends with additional payoffs for all the players. They play a critical role in determining a positive outcome.

If players focus too much on individual goals, neglect cooperation, and the sharing of knowledge or resources with the other team, Earth’s carrying capacity for plastic waste might be exceeded, resulting in a collective loss. However, if both groups focus on implementing invention cards, they can collectively win by Closing the Loop (by implementing all the circular economy cards) and replacing single-use products with circular solutions. Therefore, the game has three possible endings:

(1)

Negative—the waste limit is exceeded: no one wins prizes;

(2)

Neutral—the waste limit is not exceeded: prizes are distributed in relation to the resources acquired individually;

(3)

Positive—all the circular economy cards are implemented: prizes are distributed with an additional bonus for each participant of +1 per person.

An external, undisclosed time limit of 12 rounds for all games was established. This limit was used to evaluate how quickly players can comprehend the rules and how strongly invested they are in cooperation to achieve a circular economy as a form of common victory. There are two versions of the game with different scripted scenarios of communication between the players, named knowledge co-creation and Ivory Tower, representing two factor levels. The Ivory Tower scenario allows only for the transfer of resources and inventions from one team to the other, shared evenly between transferring players. In the knowledge co-creation scenario, one player from each group changes rooms and switches roles. That player is also transferring the resources and inventions, deciding whether to keep them or share them with the others.

Participants were rewarded with 20 PLN (approx. 4,7 EUR) coupons for every 10 resources they collected in the case of a neutral ending, and an additional coupon for each participant in the case of a positive ending. The role of the coupon is to create the initial engagement incentive [52]. Before the game starts, it is announced that there is no limit to collecting the coupons during the game. And while playing, this initial motivation is transformed into playful engagement. The payoff matrix (

Table 1. Payoff matrix as presented to players (left), and as simulated with the algorithm (right), based on Supplementary Information—Algorithm. The numbers represent the number of prizes that can be won in each scenario. “A” means “as many prizes as resources you collect during the game”.

Entrepreneurs	Accumulate		Invest		Entrepreneurs	Accumulate		Invest
Inventors		0		0	Inventors		0		0
Accumulate	0		A		Accumulate	0		2
		A		1 + A			2		5
Invest	0		1 + A		Invest	0		1

) was created with an algorithmic game engine. Detailed rules are described in Supplementary Information—Materials; the game engine is provided in the form of a flowchart (Figure 1) and in full version in Supplementary Information—Algorithm. The study began with a calibration of the scales in the form of eight pre-trial sessions with 60 participants experienced in board games. Following the implementation of all game design improvements, formal data collection began.

2.3. Model Evaluation

This study employed a mixed-methods design integrating qualitative and quantitative approaches. Balliet [53] performed a meta-analysis for communication and cooperation in social dilemmas. Based on his results, he estimated Cohen’s d = 1.21 for the positive effect of face-to-face communication. Based on this value, we calculated that for power = 0.8, we would need ~12 results per factor level (“Ivory Tower”, knowledge co-creation). Because recruitment was challenging, a total of 20 games were conducted (N = 20). An equal number of games were assigned to each of the two factor levels. Participants were undergraduate and graduate students aged 18–27 from the Biology and Geography faculties at Jagiellonian University in Kraków, Poland. The research protocol was approved by the Jagiellonian University Ethics Committee for Social Research (opinion no. 1027.0041.3.2025). The studied game sessions took place during selected classes in the summer semester of the 2024/25 academic year., after all participants had provided informed consent (see Supplementary Information—Materials). Participation in the study was voluntary, and choosing not to participate had no consequences for course attendance. Altogether, 188 students took part in the 20 game sessions. Game parameters were adjusted to the group size to maintain scalability. Since interactions in pairs differ from those in larger groups, each game included an even number of players, ranging from 6 to 14 [54,55].

2.3.1. Model Variables

The model variables were adapted from the study by Grodzińska-Jurczak [24] and common-pool resource theory [45]. The game’s outcomes were more complex than its three possible resolutions. They are expressed through the World State Coefficient (S), a composite variable derived by integrating three distinct scales: Waste Amount (W), Circular Economy Innovation Level (C), and Time Efficiency (T). Each variable contributes differently to the game’s final outcome. Waste Amount (W) indicates how close the group is to losing the game. Circular Economy Innovation Level (C) represents the potential for winning. Time Efficiency (T) captures how quickly a group reached an outcome, whether positive or negative, relative to the fixed number of rounds. The Time Efficiency (T) scale influences the score only in the event of a positive or negative ending, by adjusting for the number of rounds remaining up to a maximum of 12. Each of the three scales was standardized to a range from 0 to 1. Because each circular economy card permanently reduces the amount of waste, these variables can be combined into a single coefficient as follows:

World State (S) = C − W + (−1^{sgn(ln C)} × T)

(1)

Positive World State (S) values mean that Circular Economy (C) cards counterbalanced the Waste (W) in the common pool, which can be achieved only in a neutral or positive ending. Here, C = 0 would mean a fundamental misunderstanding of the rules by the team; therefore, it is an invalid experiment, represented by an undefined logarithm. A single World State (S) value was calculated for each game session and treated as the dependent variable in the analysis to answer the first and second research questions. Although resources (R) are an essential part of the gaming model, they were not included directly in the calculations. They were treated as instrumental because they are used to manipulate the game state and have no direct link to environmental outcomes.

2.3.2. Predictors

The World State (S) was the dependent variable, and Treatment was the independent variable: 0 for groups with limited communication and 1 for groups with knowledge co-creation. Before the game, participants completed the Polish version of the Revised Portrait Value Questionnaire (PVQ-RR) to assess the influence of personal values [56]; see Supplementary Information—Materials. The scale was reduced to two dimensions—Self-Transcendence and Self-Enhancement—based on 27 items measuring values that may affect cooperative behavior.

To address the third and fourth research questions on the game’s educational value and its impact on attitudes, pre–post questionnaires were used. The questionnaire included 21 items on general awareness of the plastic problem and science–business cooperation, plus 5 post hoc items on opinions about the game design and experience. Attitude questions were developed by the authors based on themes identified as relevant in their former research on single-use plastic and the circular economy [24,31,57]. The questionnaire was adjusted during the pre-trial period. Attitude change was tested by comparing responses to the 21 items before and after the game. Responses were (1–10, from strongly disagree to strongly agree). The model included PrePost questions as a factor, Treatment as a co-factor, and Person as a random effect.

2.3.3. Validation of the PVQ-RR Scale

Although the modified scale showed good internal consistency (α = 0.78), the confirmatory factor analysis revealed significant imbalances, requiring questionnaire decomposition. A better-fitting measurement model was obtained by reducing the number of items per latent factor (Supplementary Information—Materials). New fit indices indicated good fit: CFI = 0.951, TLI = 0.94, RMSEA = 0.049, SRMR = 0.069. All standardized factor loadings were acceptable (0.4–1.00). The final model included two latent variables: Self-Enhancement (Enhancement), comprising Achievement, Power Dominance, and Power Resources; and Self-Transcendence (Transcendence), comprising Universalism Nature, Universalism Concern, and Universalism Tolerance. A weak negative correlation emerged between Transcendence and Enhancement (estimate = −0.257, SE = 0.099, p = 0.009). Scores for Enhancement and Transcendence were extracted and entered as covariates in the model. A two-level model with a nested group variable could not be estimated due to small group sizes, an important limitation given that the pre-existing groups may differ in intragroup dynamics [58]. Despite generally favorable indices obtained during the validation process, it has to be stated that the modification of the scale might have influenced the alignment with the original theoretical constructs.

2.4. Other Methods

All data, including covariates, were analyzed with R (ver. 2025.05.0+496). Confirmatory factor analysis was used to assess the robustness of the PVQ-RR questionnaire. Using the lmer and lm functions, we built linear models and analyzed them with Anova tools (‘car’ package ver. 3.1-3). Pre- and post-questionnaire results were modeled with a mixed cumulative link model (‘ordinal’ package ver. 2025.12-29) and compared using the emmeans function (‘emmeans’ package ver. 2.0.1). Additional qualitative observations of the game flow, based on observers’ notes, are briefly reported in the results section to further illuminate game dynamics.

2.5. Data Collection

The rules of the game were presented by a game facilitator in a standardized manner, followed by an open-question session. Instruction shortcuts were continuously made available throughout the session. Although facilitators were allowed to clarify rule-related ambiguities, they were prohibited from offering any suggestions regarding strategies. The total number of game rounds remained undisclosed; participants were only aware that the session would fit within a standard 90 min class period. In cases of an odd number of participants, one individual was excluded from active participation and prohibited from interacting with other players. They were then assigned the role of observer, documenting behavioral patterns and subjective opinions. Two initial “test” rounds introduced a buffer regarding the number of actions required for a successful outcome.

3. Results

3.1. Effects of Communication and Values on the Game Results (The World State)

Out of all the results, the only two positive endings were recorded as an effect of the knowledge co-creation scenario and three negative endings as an effect of the Ivory Tower. Statistical analysis strengthens the correlation thesis. Diagnostic statistics show that model residuals have a normal distribution (W = 0.93129, p-value = 0.1635) and fulfill the assumptions of homoscedasticity. The outcome variable (World State) correlates significantly with both knowledge co-creation (Treatment) and the declared level of Self-Transcendence values (Transcendence), which allows us to reject the null hypothesis (

Table 2. ANOVA-like table showing the influence of different factors on the result of the game represented as the World State dependent variable (interval). Intercept (nominal) for Ivory Tower, Treatment (nominal) for introduced knowledge co-creation, Transcendence (interval) for mean declarations of Self-Transcendence values per group (scaled towards the group total). Enhancement (interval) for mean declarations of Self-Transcendence values per group (scaled towards the group total). Null hypothesis: predictors do not influence the World State Coefficient significantly. Levels of significance: *** = p < 0.001; ** = p < 0.01. For further statistical output, see Supplementary Information—Statistics.

Term	Estimate	Std.Error	F Statistic	2.5%	97.5%	p.Value	Sig
(Intercept)	−0.595	0.118	−5.024	−0.847	−0.343	0.000	***
Treatment	0.595	0.175	3.400	0.222	0.968	0.004	**
Transcendence	0.497	0.115	4.329	0.252	0.742	0.001	***
Enhancement	0.094	0.097	0.960	−0.114	0.301	0.352
Treatment:Transcendence	−0.255	0.173	−1.478	−0.623	0.113	0.160

).

The best fit (adjusted R² = 0.633) results from removing all insignificant terms except the Treatment × Transcendence interaction (p = 0.160), indicating that a larger sample may reveal a significant effect. If confirmed, this would suggest that the knowledge co-creation scenario reduces the impact of values. The Cohen f² coefficient was extracted for Treatment (f² = 0.860, power = 0.946), Transcendence (f² = 1.395, power = 0.995) and Interaction term (f² = 0.146, power = 0.313); both factor effects substantially exceeded Cohen’s threshold for a large effect (f² ≥ 0.35), while interaction term’s influence might be medium (f² ~ 0.15) if significant, meaning that those two factors have exceptionally good explaining power.

Ivory Tower groups low in Self-Transcendence scored just above −1, indicating that the game is difficult for groups that are not altruistic and cannot communicate between teams. Transcendence had a more significant effect than Treatment but a slightly smaller impact on World State (0.497 vs. 0.595), and the interaction term in the reduced model had an estimated slope of −0.255 (Figure 2).

3.2. Experiential Learning Outcomes: Effects of Participation on Players’ Attitudes

A linear model with visual diagnostics showed homoscedasticity, but the residual Q–Q plot indicated substantial imbalance for the sample (n = 188). Consequently, a non-parametric cumulative link mixed model was used with a stricter significance threshold (p < 0.01). A contrasted pairs table was built to identify questions whose responses changed significantly between before and after, accounting for the categorical Treatment and random Person effects. For all the pre- and post-responses below, a scale of 1–10 has been applied.

Because the sample (biology and geography students) already showed strong pro-environmental attitudes, questions on science funding, the pace of environmental research, and circular economy knowledge (Q4, Q8, Q20) were expected to yield positive responses. All other questions were expected to be negative. Results were standardized so that all items were negatively oriented, so a negative estimated change matched the predicted direction.

To reduce inflated variance from limited room for improvement, the two lowest PRE scores and their corresponding POST scores were removed. The reduced dataset was then refit. Diagnostics showed a strong random Person effect (p < 2.2 × 10⁻¹⁶), justifying its inclusion. The model revealed strong effects of the game session (PrePost) and its interaction with Question (

Table 3. ANOVA-like table for a cumulative link mixed model testing the effect of game session (PrePost: Pre/Post) on attitude towards knowledge co-creation and plastic crises (ordinal 1–10), with Treatment (two-level) and Question (21-level) as fixed factors and Person as a random effect. Null hypothesis: none of these factors significantly affect the response. Significance levels: *** p < 0.0001. Additional statistical results are provided in Supplementary Information—Statistics.

Term	df	Chi-Square	p.Value	Sig
Question	20	228.332	0.000	***
Treatment	1	1.922	0.166
PrePost	1	18.810	0.000	***
Question:Treatment	20	31.823	0.045
Question:PrePost	20	102.299	0.000	***

), though even significant results were hard to interpret. Many filtered responses indicated participants were already fully convinced, while fewer responses reflected more diverse attitudes. Since it is less important to further reassure convinced participants than to shift negative views on knowledge co-creation and the circular economy, two additional models were fitted. These models were fitted separately for responses on the positive and negative sides of the scale (agree/disagree).

All significant pre–post attitude changes aligned with the predicted item orientation. Of the 21 items, 12 changed significantly; no substantial differences between the groups were detected. The greatest shift occurred for the separation of the disciplines of science and business (Q2; see

Table 4. Estimated marginal means (ordinal responses converted to a 1–10 interval, 1 = strongly disagree, 10 = strongly agree) comparing post- and pre-game session questions (treated as a 21-level nominal factor), based on a cumulative link mixed model. Null hypothesis: the true mean response does not differ between post- and pre-game questionnaires. For items 4, 8, and 20, PRE and POST values were swapped to allow uniform comparison of estimates. “Filtered responses” gives the proportion of responses removed due to extreme pre-questionnaire values (1 or 2). Significance levels: *** p < 0.0001; ** p < 0.001; * p < 0.01. Further statistical details are provided in Supplementary Information—Statistics.

Question	Contrast	Filtered Responses	Estimate	SE	Asymp.LCL	Asymp.UCL	p.Value	Sig
1. Current free-market principles promote the flow of knowledge from scientists to entrepreneurs.	POST–PRE	10.64%	−0.848	0.196	−1.2312	−0.4638	0.0000	***
2. Scientists should focus on acquiring knowledge; the free market will make the best ideas come to life.	POST–PRE	43.09%	−2.241	0.262	−2.7554	−1.7267	0.0000	***
3. Entrepreneurs will find out on their own which scientific discoveries are worth implementing.	POST–PRE	31.38%	−1.089	0.241	−1.5605	−0.6165	0.0000	***
4. Financing science should be the domain of the state.	PRE–POST	39.89%	−0.002	0.245	−0.4831	0.4788	0.9930
5. Innovation should primarily serve economic prosperity.	POST–PRE	10.64%	−1.145	0.199	−1.5343	−0.7556	0.0000	***
6. It is enough to develop economic potential and the free market will solve the waste problem.	POST–PRE	45.74%	−0.513	0.252	−1.0061	−0.0189	0.0419
7. We have enough time to solve the waste problem.	POST–PRE	43.62%	−1.090	0.265	−1.6092	−0.5703	0.0000	***
8. There are already many scientific discoveries that could solve many environmental problems if someone implemented them.	PRE–POST	26.06%	0.032	0.217	−0.3936	0.4568	0.8843
9. Scientists should focus on acquiring knowledge; the free market will ensure that the best ideas are implemented.	POST–PRE	32.98%	−0.418	0.218	−0.8456	0.01	0.0556
10. Scientific discoveries are communicated effectively enough.	POST–PRE	34.04%	−0.804	0.227	−1.2488	−0.3592	0.0004	**
11. Science is primarily the domain of universities and colleges.	POST–PRE	13.30%	−0.413	0.2009	−0.8063	−0.0187	0.0400
12. Entrepreneurs, when introducing new technologies, should above all consider whether they can earn profits from them.	POST–PRE	42.02%	−1.073	0.254	−1.5703	−0.5758	0.0000	***
13. Scientists should not engage in communication with the public or business.	POST–PRE	61.17%	−1.730	0.329	−2.375	−1.0844	0.0000	***
14. Science should be independent and set its own goals	POST–PRE	7.45%	−1.185	0.204	−1.5848	−0.7851	0.0000	***
15. Science should focus on the most profitable areas	POST–PRE	55.32%	−0.579	0.286	−1.1386	−0.0194	0.0426
16. The only commitment of the industry towards the environment is ecological neutrality	POST–PRE	32.45%	−1.132	0.228	−1.5782	−0.6859	0.0000	***
17. Since you cannot make money from ecology, entrepreneurs shouldn’t deal with it too much.	POST–PRE	76.60%	−0.784	0.428	−1.6234	0.0547	0.0669
18. Science will sooner or later solve the problem of plastic waste on its own.	POST–PRE	42.55%	−0.624	0.253	−1.1183	−0.1286	0.0135	.
19. Recycling is a sufficient form of dealing with plastic waste.	POST–PRE	60.11%	−0.388	0.3	−0.9761	0.201	0.1968
20. I know well what the circular economy is.	PRE–POST	13.83%	−1.730	0.222	−2.1658	−1.2942	0.0000	***
21. Plastic waste is not one of the most urgent problems.	POST–PRE	41.49%	−0.804	0.257	−1.3068	−0.3015	0.0017	*

for full questions), where the groups showed a significant decrease in the attitude (estimate = −2.241 scale points). The second-largest change can be attributed to two items. These are an increase in self-reported knowledge of the circular economy (Q20) (estimate = −1.730 scale points) and attitudes towards the need for science communication (Q13), despite the highest number of significant filtered responses (115) for this item. All remaining items showed estimated changes within the 0.805–1.186 scale points. Participants’ support for active business involvement in addressing the plastic crisis increased, even when such involvement was not immediately profitable. The absolute right for science independence in setting its goals and strategies was less valued than engaging in collaboration. Items without statistically significant change included topics such as the source of science funding, the current state of sustainability research, the free market of ideas (Q4, Q8, Q9, Q11, Q18), and items with the highest number of filtered responses (Q6, Q15, Q17, Q19).

3.3. Differences in Response with Respect to Primary Attitude

Because the sample consisted of Biology and Geography students, all items were skewed in accordance with their orientation, increasing the likelihood of a Type II error. The magnitude of the difference between pre- and post-questionnaires varied according to the primary attitude of the respondent (negative “pro-environmental”: pre-response 1–5; positive “anti-environmental”: pre-response 6–10) and the Question, as well as their interaction. It also depended slightly on the Question:Treatment (

Table 5. ANOVA-like table for a cumulative link mixed model showing how game session ratings (ordinal 1–10) are influenced by preceding attitude (two levels: positive/negative, responses standardized toward negative item orientation), Treatment (two levels), and Question (21-level factor), with Person as a random effect. Significance levels: *** p < 0.0001. Additional statistical details are in Supplementary Information—Statistics.

Term	df	Chi-Square	p.Value	Sig
Question	20	92.282	0.000	***
Attitude	1	147.259	0.000	***
Treatment	1	1.109	0.292
Question:Attitude	20	61.941	0.000	***
Question:Treatment	20	36.386	0.0139

). The negative attitude was estimated to increase by 0.35 points, and the positive attitude fell by 1.53 points, according to the estimated marginal means test (z.ratio = 23.312, p < 0.0001). Only eight questions did not change significantly in respondents with “anti-environmental” attitudes. These included questions about science funding, current effectiveness of science communication (Q4, Q8, Q9, Q10, Q18), and items with the highest number of filtered responses (Q15, Q17, Q19). In the “pro-environmental” group, only six questions changed significantly. These included statements about the time limit and industry commitment to addressing the plastic crisis (Q7, Q10, Q16) and the questions with the highest estimates (Q2, Q13, Q20). The only response that did significantly change in the convinced group and did not in the unconvinced group was with respect to the question about the current effectiveness of science communication. The results suggest that the game affected people without positive attitudes towards the circular economy more strongly than those already convinced.

3.4. Qualitative Observations

These remarks draw on debriefings, facilitator and observer notes, and first-hand observations. Clear differences emerged between groups with knowledge co-creation mechanisms and those with restricted communication. Participants in the Ivory Tower conditions showed more confusion, lower engagement, weaker forward planning, and more irrational decisions at both individual and group levels. Intergroup antagonism frequently appeared and seemed to rise with increasing waste. All three recorded game failures occurred only in Ivory Tower groups, usually due to passive play and the lack of a shared strategy. While no explicit hoarding of resources (R) for high-value rewards was seen, resources remained tightly controlled by individual players. In contrast, knowledge co-creation groups engaged more actively in intragroup information exchange. Although forced player transfers between rooms initially caused brief confusion, the group dynamic usually enabled quick onboarding. Groups that adopted the cycle of research, rotation, and innovation implementation early were more successful, but many ultimately ran out of time before fully executing their winning strategies.

Participants’ decisions were not always strategically optimal. At times, this was linked to explicitly stated anxiety or uncertainty about the consequences of their actions, even though the rules were available. In some cases, players chose not to act despite having a clear advantage in doing so, or they decided to follow others’ directions. This pattern suggests that the presence of a confident, strong leader among players may have been an important factor for achieving positive results. In contrast, an anxious leader could steer the team toward poorer outcomes.

Moreover, players seldom emphasized research intensity or recognized the relationship between greater research efforts and the likelihood of obtaining circular economy cards, which provide cumulative benefits and are crucial for winning. Instead, they frequently pursued immediate rewards and overlooked long-term strategies. The groups that performed best concentrated on keeping waste levels below a critical threshold. In some cases, they even reduced waste levels further, without an obvious economic or strategic rationale. This behavior points to a preference for alleviating environmental anxiety rather than for strictly optimizing problem-solving.

A common emergent pattern was the spontaneous establishment of a shared resource (R) pool, collectively used to finance common goals. Although such actions are technically disallowed by the rules and the study’s assumptions, they open a valuable line of inquiry for future research. Future studies could examine the motivations and conditions that produce collective problem-solving strategies and assess whether they are more effective.

3.5. Participants’ Subjective Evaluation of the Game

Participants showed moderate enthusiasm: 61% would repeat the experience (Q22, score 8–10), which is important if the game were to be used as a frequent exposition tool. 52% felt the game accurately represented scientific communication (Q23, 7–9), and 60% agreed it proposed viable science-business reforms (Q26, 8–10). Meanwhile, 47% were unsure whether the mechanics needed revision (Q24, 3–5) (Figure 3; full questions in Supplementary Information—Materials). Interest in acquiring the game differed significantly (p = 0.007), with higher commitment in the knowledge co-creation group (Q25: 46% scoring 6–8 vs. 41% scoring 5–7).

4. Discussion

Here we discuss the most notable comparisons with previous serious game studies. Whalen et al. [59] created In the Loop, which focuses on recycling and market prices for rare earth elements. While In the Loop deals with recoverable, reusable resources, Closing the Loop addresses single-use plastics with minimal recycling potential. In our game, engaging in “recycling practices” diverts attention from the main goal: replacing plastic with environmentally neutral materials. The games also differ in educational aims and mechanics. In the Loop is competitive with a single winner, whereas our model uses mixed incentives. Our model lets players aim for personal or collective victory.

Risk&RACE, developed by Manshoven & Gillabel [36], is a circular economy game focused on a business perspective. It explains how companies are affected by external pressures and how their sustainability varies with circular versus linear business models. It also provides detailed insights into circular economy strategies using qualitative methods. The game involves teamwork and rivalry between groups. Like Whalen’s [59], it simulates a free market but does not address common-pool resource issues or the value of knowledge co-creation for solving circular economy challenges.

Knowledge Brokers [60] is a rare example of a serious game that attempts to express the importance of knowledge architecture. The difficulty lies in introducing game mechanics that facilitate not only seeing but also experiencing how learning and teaching practices can have a real impact on problem resolution outside of academic conditions. Players respond here to unexpected events by assigning fictional staff to perform socio-economic interventions [61]. While not being connected to environmental issues, the game features a knowledge co-creation loop, enabling cooperation for better understanding of the rules and pursuit of a shared goal. This lets participants experience, through practice, that brokering knowledge is an effective way to solve operational problems.

Finally, we want to distinguish between games as a form of knowledge co-creation and games about knowledge co-creation. One of the prime examples of the former is Santonen et al.’s game [43], which is used as a tool for facilitating co-creation in the business sector. However, it represents only one method of co-creation and does not attempt to argue for the broader application of this practice. The game is highly specific and practical, but it does not facilitate discussion of different knowledge architectures or examine whether knowledge co-creation may be more effective than other strategies. While Santonen et al.’s game is a dynamic first-order system of co-creation, Closing the Loop is a second-order system of knowledge co-creation that targets knowledge architecture and cooperation, in which participants learn about learning [62].

The game studies discussed here [36,43,59,60] focus on the qualitative assessment of learning outcomes. In contrast, our main point of interest was the influence of the “Ivory Tower” setting on the resolution of common-pool resource problems. To facilitate this, we introduced incentives for both securing individual profits and achieving a common victory, which are absent from the other games. Another novel aspect of our design was the presence of two sets of rules for the two teams, which highlighted the necessity of social learning to establish the best strategy. In this way, we avoided simulating how two communities learn the formal rules needed to solve the problem successfully. Despite the limited set of possible actions and clearly stated payoff conditions, player exchange still managed to improve knowledge and strategy co-creation. This mechanism led to significantly better resolution of the presented problem. This result, together with tentative data on the game’s influence on attitudes and perceptions, allows us to conclude that our game not only presents the concept of knowledge co-creation but also demonstrates, through experience, how practicing it helps solve common problems.

Knowledge co-creation remains a widely misunderstood concept, often perceived more as a data-collection practice than a joint effort in producing unique insights. The advantages of knowledge co-creation extend beyond more detailed observations or more accurate descriptions of specific phenomena to include the adoption of common goals and acknowledgment of diverse needs and operational modes [63]. However, coupling science and business is not without risk, as it can lead to conflicts of interest, as seen in the biomedical field [63]. Seemingly sustainable governance may serve as a pretext for obscuring other unsustainable actions. Indirect funding streamed through independent institutions that promote business–science cooperation without creating co-dependence might mitigate this risk. The use of a common-pool resource game with a collective failure trigger in this study prevented such outcomes, highlighting the benefits of collaboration without associated drawbacks. However, implementing a more competitive game design could reveal interesting interdependencies.

Values play a significant role in resolving social dilemmas, even when only approximated at the group level. Although individual value profiles, like moral foundations, may seem secondary in large-scale problems, their impact on decision-makers remains important [64]. Environmentally oriented organizations might therefore incorporate value-based criteria when forming managerial teams. While values and deeply held attitudes are usually stable, they can be shaped by priming—creating memory links that increase their accessibility for behavior—or by other interventions, especially during adolescence [65]. Serious simulation games may be effective for such interventions among youth, particularly when used over time and/or in their direct social context [66]. We showed that Closing the Loop can influence attitudes, but its capacity to affect values remains uncertain.

Limitations

The main limitation of our study was the choice of the final value of the World State as the dependent variable. Although necessary from the perspective of the research design, this choice generated two interconnected problems.

Firstly, we needed to recruit a large cohort of participants to maintain statistical power. However, potential subjects were reluctant to participate. Playing was viewed as an activity of lesser importance, which led to lower interest in participation. Therefore, we had to settle on a convenience sample, which consisted entirely of pre-existing student groups of similar age from a WEIRD population (white, educated, industrialized, rich, and democratic) [67]. Pre–post question results confirmed this limitation. Future research could study a more diverse population or examine the reactions and behavior of actual scientists and business representatives.

Secondly, we had to limit the scope of the study, placing less focus on testing its educational effects. We are aware that, in order to influence students’ attitudes and perceptions, we would have to expose them to more game sessions [66]. To demonstrate it, a more robust design would be needed, such as the Solomon four-group method [68]. Only then would we be able to account for such problems as Hawthorne effects, novelty effects, and desirability bias [69,70]. However, such an approach deserves separate study.

5. Conclusions

We have found that successful in-game resolution of the common-pool resource dilemma correlates with the introduced mechanism of knowledge co-creation and Self-Transcendence values. The limitations of this study prevent broad generalization. However, both in the game and in the organization, social learning can be enhanced or impeded by the rules in use—rules that structure participation, access to information, communication, feedback, and the consequences of collective decisions [71,72]. The game is used to test this mechanism under controlled conditions; whether the observed effect transfers to organizational practice is an empirical question for another study.

Another aim of this study was to assess whether Closing the Loop has the potential to educate users about the value of knowledge co-creation and collaboration in addressing environmental and social dilemmas, such as single-use plastic pollution. The results indicate that the implemented personnel exchange mechanism helps address this issue through knowledge co-creation, which stays in line with previous findings [45]. By providing real-time observable outcomes and rules that foster social learning, the game facilitates the construction of knowledge during gameplay [73]. Further research is needed to determine the persistence of this knowledge.

We wish to end by addressing the broader question of why knowledge co-creation should be taught at all. Our answer is that knowledge co-creation is an effective strategy for collaborative problem-solving, a claim supported by the findings of our study. In this context, experiential learning is not only a process of acquiring testable knowledge but, more importantly, an insider’s exposure to systemic processes. Soft skills and deeply rooted attitudes are not easily influenced by knowledge-based interventions or direct confrontation unless a supportive setting is provided [74,75,76]. A serious simulation game like Closing the Loop can provide such conditions, encouraging individual reflection on one’s approach. As one participant noted: “If I were to play again, I would know what to do”—and hopefully this time the game will be for real.