Strengthening Environmental Decision Making through Legislation: Insights from Cognitive Science and Behavioural Economics

Abstract The environmental assessment literature has neglected the distorting effect of cognitive and unconscious motivational biases (CUMB) in environmental assessment processes. This is problematic because CUMB are present in most, if not all, decision-making situations and can significantly distort decision-making processes. This article assesses how debiasing techniques are, or should be, incorporated in (supra)national environmental assessment legislation. The Dutch case study undertaken for this article shows that EU and Dutch legislation do not sufficiently incorporate debiasing techniques to ensure sound environmental decision making. Furthermore, the extent to which Dutch legislation incorporates debiasing techniques was found to be decreasing. Based on these findings, the article presents ways to incorporate debiasing techniques in environmental assessment legislation more generally, and in EU and Dutch legislation in particular.



The rationale behind environmental assessment (EA) 1 is that the impact of (potentially) environmentally harmful activities should be analyzed before authorization of an See, e.g., Convention on Wetlands of International Importance, Ramsar (Iran), 2 Feb. 1971, in force 21 Dec. 1975, available at: https://treaties.un.org/doc/Publication/UNTS/Volume%20996/volume-996-I-14583-English.pdf; United Nations Framework Convention on Climate Change (UNFCCC), New York, NY (United States (US)), 9 May 1992, in force 21 Mar. 1994, available at: https://unfccc. int; United Nations Convention on the Law of the Sea (UNCLOS), Montego Bay (Jamaica), 10 Dec. 1982, in force 16 Nov. 1994, available at: https://www.un.org/depts/los/convention_agreements/texts/ unclos/unclos_e.pdf; Convention on Environmental Impact Assessment in a Transboundary Context, Espoo (Finland), 25 Feb. 1991 bias (CUMB) when they process information, 9 including the kind of information contained in EAs. Cognitive biases are a systematic discrepancy between the 'correct' answer in a judgmental task, given by a formal normative rule, and the decision maker's actual answer to such task. Unconscious motivational biases are those in which judgments are unconsciously influenced by the (un)desirability of events, consequences, outcomes, or choices. 10 Furthermore, planning authorities are not the only ones involved in the EA process. Consultancy firms often write EA reports, sometimes in consultation with the public and advisers, to inform planning authorities of environmental impacts. 11 Research shows that these actors also suffer from CUMB, even when they are experts. 12 As a result, CUMB may, through various ways, distort the capacity of EAs to guide planning authorities to take environmental effects duly into account in decision making.
Public and judicial review are often regarded as the best ways to prevent planning authorities from making decisions that do not take environmental effects duly into account. 13 However, the public, as well as the judiciary, 14 have been found to be subject to CUMB as well. 15 As such, public and judicial review may not be sufficiently effective to prevent or correct decisions where environmental concerns are not duly taken into account. For example, empirical studies indicate that people systematically make decisions based on whether they are capable of constructing a coherent story, regardless of the amount and quality of the data on which the story is based. 16 As public and judicial review feed off the information generated by the EA and communicated in the EA report, the quality of EA processes needs to be as high as possible.
A rich body of scholarship already exists on how EA processes could be improved. 17 This literature often draws upon multidisciplinary research in related fields of planning, psychology, economics, policy analysis and political science in the interest of theory building. 18 The studies that draw from the broad field of behavioural science generally focus on research on bounded rationality. These studies argue that planning authorities face ambiguous and poorly defined problems, limited time, skills and resources, as well as incomplete information about alternatives, the baseline, consequences of supposed alternatives, range of values, preferences, and interests. 19 In this context, the concept of satisficing has been widely discussed in EA scholarship. 20 However, the EA literature has neglected more recent insights from behavioural science almost completely, most notably the insights on CUMB. This is problematic because CUMB are hard-wired and are therefore present in most, if not all, decisionmaking situations. 21 Nevertheless, many scholars emphasize the distorting effects of political bias in EA. 22 While political biases are indeed often found to be the predominant behavioural bias in major consequential decisions and projects, 23 experimental psychologists have shown that political bias directly amplifies CUMB. For example, powerful people are affected more strongly by CUMB than the less powerful. 24 Moreover, those affected by CUMB tend not to be aware of this and often blame unexpected outcomes on political bias or issues outside their control. 25 So far, only a few studies have focused on insights into CUMB in EA generally, and EA legislation more specifically. These studies provide insights into the practice of sustainability assessments, a deeper understanding of its effectiveness in trade-off decision making, 26 as well as how insights into CUMB could be integrated into the international law of EA. 27 This article takes a different approach by assessing how debiasing techniques are, or should be, incorporated at the (supra)national level to debias CUMB in the EA process.
This study is timely as jurisdictions have been or are currently reducing legal requirements in the EA process, known as 'streamlining'. 28 These streamlining efforts often reduce the requirements of three important debiasing techniques: (i) strengthening expertise, (ii) nominal group decision making, and (iii) frequent feedback loops. These debiasing techniques are important because they tackle a few instances of CUMB that frequently appear in decision-making processes, most notably: • WYSIATI ('what you see is all there is') refers to the observation that people construct opinions based on presented information, even if they are aware that that information is biased or one-sided. 29 To make matters worse, those confronted with biased or one-sided information are more confident about their conclusions than those confronted with balanced information. 30 This cognitive bias could, for example, lead planning authorities to decide in favour of a polluting activity presented positively by the initiator of a plan or project. • The overconfidence bias occurs when the decision maker provides estimates for a given parameter above the actual performance or when the range of variation they provide is too narrow. 31 This cognitive bias could occur, for example, at the EA preparation stage when an expert is asked to provide a probability distribution. • The confirmation bias occurs when there is a desire to confirm one's belief, leading to unconscious selectivity in the acquisition and use of evidence. 32 This motivational bias could, for example, unconsciously lead planning authorities to fail to take all relevant alternatives adequately into account in the EA process.
• The optimism bias occurs when the desirability of an outcome leads to an increase in the extent to which it is expected to occur. 33 This motivational bias could lead decision makers unconsciously to underestimate the environmental impact of an activity.
The efforts to streamline EA legislation are especially problematic when non-legal means are not sufficiently effective in coordinating human behaviour to overcome CUMB in the EA process. Introducing legal means, then, can be helpful in influencing individual and state behaviour as a result of possible sanctions imposed (such as fines, requirements for supplementary assessments, or reputation damage caused by transparency requirements). 34 Furthermore, even without sanctions, legal means can have an expressive effect. Empirical studies confirm that actors tend to internalize values expressed by law and obey out of internal respect for the law. 35 Legal means can also provide a focal point, creating a social norm around which actors can coordinate their behaviour. 36 This article adopts an exploratory research design, taking the Netherlands as a case study, to understand whether EA legislation at the (supra)national level incorporates sufficient debiasing techniques and whether further legislation is needed. The Netherlands served as a case study for three reasons. Firstly, formally introduced in 1987, 37 the EA process is well established in the Netherlands and is often considered a frontrunner in good EA practices and legislation. 38 Its requirements often exceeded the requirements of the EU Directives, which set the minimum requirements for EA in the EU. Secondly, the Netherlands, being a Member State of the EU, has to comply with the two main EU Directives regulating EA in the EU: the EIA and the SEA Directives. 39 As such, the case study provides insight into how the EU Directives are implemented at the Member State level. Thirdly, since 2005, a clear trend has been visible where the Netherlands is increasingly streamlining its legal requirements to make them similar to the minimum requirements set by EU legislation on EAs. 40 Many other jurisdictions have also streamlined legal requirements in the EA process. 41 A behavioural explanation for this is that minimum legal standards, such as those set by the EU, run the risk of being considered as 'anchors' to which no further effort is required, thereby possibly weakening Member State motivation to take additional measures. 42 Insights into how the EU could improve its EA legislation to ensure a baseline of good EA practices, leaving a role for Member States to put new and better practices into place, is therefore needed. As such, this case study provides preliminary insights into whether streamlining could adversely affect the emergence of CUMB in the EA process, and whether the minimum requirements set by EU legislation on EAs are sufficient to overcome CUMB.
To achieve the aim, this article, firstly, explains the methodology (Section 2). Section 3 then provides a brief overview of the various elements of an EA that are often mandated by EA legislation, focusing in particular on EU and Dutch legislation. Section 4 explains the three debiasing techniques studied in this article in depth in three subsections: expertise (4.1), group decision making (4.2), and feedback loops (4.3). Each subsection focuses on how these debiasing techniques are incorporated into EU and Dutch legislation as well as in practice (drawing from existing studies as well as from the case study). Recommendations on how to strengthen these debiasing techniquesthrough EA legislation more generally, and EU and Dutch law specifically are presented at the end of each subsection. Section 5 concludes and provides avenues for further research.


A case study design was chosen as it allows an in-depth analysis of a specific phenomenon in its local context. 43 Furthermore, a case study design is especially suitable for exploratory research, in which understanding is the primary objective, and the phenomenon under investigation is difficult to quantify, not well understood and needs to be studied. 44 Data for the case study was gathered through (i) the analysis of academic literature, relevant law and policy documents, as well as authoritative court texts; and (ii) semi-structured interviews in December 2019 with 11 actors involved in the Dutch EA process. 45 The interviews were aimed at obtaining a deeper understanding of the EA process, especially whether non-legal debiasing techniques were used in practice to sufficiently overcome CUMB (arguably making further legislation redundant). Textual data was collected using a comprehensive search for articles and books through several databases, such as Web of Science, GoogleScholar, Curia, the jurisprudence database of the Netherlands Commission for EA (NCEA), and online libraries. Search words included 'environmental (impact) assessment', 'strategic environmental assessment', 'EIA Directive', 'SEA Directive', together with 'cognitive bias', 'motivational bias', 'bounded rationality', 'human decision making' and 'expert decision making', as well as the Dutch translation of these search terms. Snowballing was used to find additional literature, relevant law and policy documents, and court texts. 46 The interviewees were selected based on the parties mentioned in the EIA Directive, the SEA Directive, and Dutch EA legislation. Interviewees were identified using snowball sampling techniques, and were selected based on their expertise, role, and practical experience with the EU and Dutch EA processes. The interviewees were four government employees at the local, regional, and state levels (acting as authority or initiator), four consultants from consultancy firms of small, medium, and large capacity involved in EA report preparation; one government employee of the Directorate-General for the Environment of the European Commission responsible for EA implementation by the EU Member States, and two employees of the NCEA. The interviews were held one-on-one at the office of the interviewee and lasted for about 60 minutes.
An open-ended theory-building approach was used, which started during the data collection stage and involved several iterative cycles. 47 The data analysis involved coding, categorizing, and abstracting to higher-level concepts. 48 A concept-driven approach was used, where data is reported based on the relevance for understanding whether expertise, nominal group decision making, and feedback loops are sufficiently stimulated in the EA process. All data passages that were informative in terms of understanding where CUMB may play a role in the EU and Dutch EA processes or where debiasing techniques are used were coded in MAXQDA (See the Appendix in the online Supplementary Materials). During the data analysis, I constantly iterated back and forth between the textual and interview data.
An important limitation of qualitative research, such as case studies and semistructured interviews, is that of double hermeneutics: the researcher may influence the people or the topic being studied and the other way around. 49 Reflexivity is generally understood as awareness of this double hermeneutics. 50 To reduce the possibility that my biases and hypotheses influenced the interviewees, I avoided words such as 'psychology', 'cognitive', 'rational', 'tunnel vision' or 'behavioural' during the interviews and in any prior correspondence. Instead, I focused on prompting, probing and encouraging the interviewees to share their views of their experience with EA by 46  asking them to give concrete examples of EAs that they thought went well or not so well, followed by questions on why they thought this was the case.

      
Most EA legislation includes requirements on screening, scoping, EA report preparation, review, decision making, and follow-up. These requirements will be explained briefly below, including more specific requirements at the EU and Dutch levels. A simplified overview of the requirements common to EU and Dutch legislation is provided in Figure 1 and below.
The step that precedes most EAs, screening, serves to determine whether a proposed activity requires further identification and prediction of environmental impacts through an EA. 51 Most EA legislation, including EU and Dutch law, include two approaches to screening: case-by-case examination, and thresholds. The planning authority is responsible for undertaking one of these approaches depending on the characteristics of the proposed activity.
Case-by-case examination involves the appraisal of the characteristics of activities against a checklist of criteria. 52 The use of thresholds involves placing activities in categories and setting thresholds for each activity (for example, relating to scale, anticipated impacts, or location). 53 Dutch legislation specifies an exhaustive list, based on the mandatory lists provided by the EU, 54 for activities that meet certain thresholds which always require an EA. 55 Dutch law requires a case-by-case examination for those activities not meeting the thresholds of the positive lists by the authority. 56 If an EA is warranted, the process may start with scoping, which may be undertaken by the initiator and the planning authority. During scoping, the content and extent of the matters to be covered in the EA are specified before the environmental impacts are predicted. 57 Following identification of possible environmental impacts (by the initiator of the plan or project, or through a scoping process) prediction of likely environmental impacts is required in preparing the EA report. Consultancy firms usually conduct the preparation of EA reports.
The public is typically empowered to review the final EA report. 58 Some jurisdictions, such as the Netherlands and Canada, also mandate expert review of the final EA report. The Canadian approach, whereby a panel appointed by the Ministry of the Environment is mandated to review the final EA report, inspired the establishment 51 Wood & Becker, n. 9 above, pp. 352-3; Craik, n. 2 above, p. 133. 52 Glasson & Therivel, n. 11 above, p. 86. 53 Ibid. 54 For projects: Annex I EIA Directive; for plans: Art. 3(2) SEA Directive. 55 Besluit Milieueffectrapportage 1987, n. 37 above, C and D lists. 56 Annex III EIA Directive; Annex II SEA Directive; Wet Milieubeheer (Wm), s. 7.16. 57 Glasson & Therivel, n. 11 above, p. 88. of the NCEA in the Netherlands. 59 The NCEA is a permanent independent foundation responsible for monitoring the scientific quality of EA reports in the Netherlands. Upon request, the NCEA assembles an ad hoc working group of university professors, research institutes, and consultancy firms to monitor the scientific quality of a specific EA report. 60 Once established, the working group may give, and is sometimes mandated to give non-binding advice throughout the EA process.
Thereafter, the EA report is communicated to the planning authority to decide on the authorization of the activity. The final step in the EA process is follow-up, which usually consists of monitoring, evaluation, management, and communication. 61 Monitoring involves the measuring and recording of physical, social, and economic variables associated with development impacts. 62 Evaluation compares impacts predicted in an EA with those that occur after development consent is given in order to assess whether the impact prediction performs satisfactorily. 63 Management refers to taking appropriate action in response to issues raised from the monitoring and auditing activities. 64  Research from cognitive science and behavioural economics shows that experts (those who by training and experience have more knowledge than the general population) 66 are superior to novices in nearly every aspect of cognitive functioning, from memory and learning to problem solving and reasoning. 67 Experts often show high, outstanding, and exceptional performance that is domain-specific, stable over time, and related to experience and practice. 68 When planning authorities are inexperienced with EAs they may have more difficulty in interpreting existing scientific evidence and they may be more likely to suffer from WYSIATI ('what you see is all there is') at all stages of the EA process. 69 This may lead planning authorities to conclude that an EA is not needed, even though it should be (or the other way around), or to conclude that a biased EA report is adequate to inform decision making. In addition, when consultancy firms lack experience they may not deliver sound EA reports.

Requirements on expertise in EU and Dutch legislation
EU legislation, most notably the EIA Directive, sets requirements for expertise, in that it obliges planning authorities to have (access to) sufficient expertise in examining the EA report. 70 However, it is not clarified what 'sufficient expertise' entails; hence, it is the Member States that decide when authorities have (access to) sufficient expertise. The Netherlands strengthens expertise through NCEA review, 71 which currently is required only for the final EA report for plans and complex projects. 72 Requirements on mandatory NCEA review will be further stripped down with the entry into force of the new Dutch legislation on EA (Figure 1), 73 which will no longer mandate NCEA review for complex projects. 74

Expertise in EA practice
Sufficient expertise is often lacking in practice. For example, EU Member States report challenges regarding the expertise of authorities in preparing EAs of good quality. 75 As was found in the empirical data collected for this study, a government employee of the European Commission said: If you are a municipality of medium level or a small region in a specific Member State, you may not have all the capacity to make a good decision. So, this is an issue in general, especially when Member States are not well equipped or if your personnel is cut and you don't you have dedicated persons to do something.
Interviews with the Dutch authorities confirmed this concern. For example, a governmental employee at the state level shared: An EA often fails because a specific EA is not conducted that often. If you only have an EA regarding that theme every other year it is all new and unknown, which makes the risk for mistakes higher. Also, the authority is often somebody who has to decide on an EA only very irregularly.
In addition to the lack of expertise on the part of planning authorities, the technical knowledge and experience of the experts who predict the environmental impacts (often consultancy firms) are also often lacking. 76 The interviewees also frequently questioned their expertise. For example, a governmental employee at the regional level noted the problems experienced by EA consultancy firms in delivering sound EAs based on a lack of expertise: Year after year, EA consultancy firms have more problems delivering EAs in which they see the linkages between the different components. This is not only because the world is becoming more complex; it is also because the consultants lack expertise.

Recommendations for incorporating expertise requirements in EA legislation
Expertise can be strengthened by mandating a permanent independent foundation in each Member State that is responsible for reviewing the scientific quality of EA reports and to consult throughout the different stages of the EA process. Other options for strengthening expertise in EA processes through legislation is by requiring specified academic degrees or industry accreditation of authorities and consultancy firms to ensure that such skills are obtained, or requiring them to have certain levels of experience in relatively predictable EAs. 77 Legislation could also ensure expertise by allowing only consultants on a list of 75  approved consultants/consultancies to carry out EAs. 78 However, such a list may be problematic as it may increase rent seeking and set barriers to market entry. 79 It should be noted that the European Commission proposed to give Member States the option either of having competent accredited experts to draw up an EA report, or having a specific commission of experts in place responsible for checking the EA report. 80 However, the Member States did not wish to have this option in the Directives: further regulation on this was 'considered to increase the cost' for the Member States. As such, 'it is up to the Member State to say what type of diploma you need to have, how many years of experience', according to the Commission interviewee. However, research indicates that an independent foundation would be likely to be cost-efficient, given that a main cause of the increased total costs of the EA process is when the EA does not provide adequate or relevant data. An independent foundation involved in the EA process could spot such inadequacies, reducing the need for supplementary information and, possibly, costly judicial review. 81 The EU, therefore, should at least require Member States to install an independent foundation to check the EA report for plans and projects. Ideally, this foundation should be involved throughout the EA process to spot inadequacies at an early stage to avoid WYSIATI creeping into the process. The Netherlands could further improve its EA legislation by generally requiring review and consultation by the NCEA for all activities at all stages of the EA process.

Debiasing Technique 2: Strengthening Group Decision Making
Strengthening group decision making explained While strengthening the expertise of authorities and consultancy firms may reduce, to some extent, the degree to which they are susceptible to WYSIATI, cognitive insights show that expertise alone is not sufficient to reduce many other examples of CUMB in decision making. Most notably, the confirmation, the optimism, and the overconfidence biases are notoriously difficult to overcome by training. 82 As such, additional debiasing techniques are needed.
To reduce WYSIATI, the confirmation bias and the optimism bias, experiments show that groups are found to be well equipped to overcome these biases as groups tend to think more slowly. 83  found to provide more objective, unbiased, and more complete and professional information than any well-trained expert could ever do alone. 84 This does not mean that groups cannot arrive at poor decisions; groups have been shown to make disastrous decisions with deadly consequences. 85 However, experiments show that, to optimize group decision making, stimulating group interaction in nominal groups (where members generate ideas in isolation) is an effective debiasing technique of CUMB difficult to overcome by training. 86 Nominal decision making is especially effective in overcoming groupthink, a psychological phenomenon often occurring in highly cohesive decisionmaking groups that are insulated from experts, perform limited search and appraisal of information, operate under directed leadership, and experience high-stress conditions. 87 Research shows that initiating group interactions at the scoping stagethat is, before environmental impacts are predicted, and preferably including the initiator, the authority, other relevant public agencies, the public, and independent adviserscould reduce the time needed for an EA by ensuring that the process focuses on key issues and is carried out efficiently. 88 Furthermore, it reduces the risk of possible shortcomings in the final EA report that may be caused by and may exacerbate CUMB. Therefore, initiating group interactions at the scoping stage (and not only after the EA report has been submitted) reduces the risk of biased and one-sided EA reports, which can then aggravate CUMB, most notably WYSIATI, in the public, authorities, experts, and the judiciary. 89

Requirements for group decision making in EU and Dutch legislation
Both sets of legislation require group decision making at the screening, scoping, and review stages. At the screening stage, the EIA Directive considers it 'good administrative practice' to take 'unsolicited comments' into account, 90 and the European Commission stipulates the usefulness of consulting public authorities, stakeholders, experts or other planning authorities. 91 However, when consultations are required (that is, for plans at the EU level and, until the new legislation comes into force, also complex projects in the Netherlands), they are required only with governmental authorities. 92 Requirements at the scoping stage may enable nominal group decision making in the Netherlands: namely, the requirement for public participation when scoping is mandated. 93 However, scoping is required only for plans and complex projects at the Dutch level (under the forthcoming Dutch legislation only for plans). 94 In cases where scoping is not mandated initiators must request scoping themselves or, if the initiator does not make such a request, scoping can be initiated by the planning authority (known as scoping ex officio) 95 However, if scoping is not mandated, there are no requirements to include other relevant public agencies or independent advisers, which may strengthen group decision making. Furthermore, with the forthcoming Dutch legislation, planning authorities will no longer be able to initiate scoping ex officio, 96 leaving it for the initiator to decide whether scoping is undertaken for a project.
At the review stage, also, requirements exist that could enable nominal group decision making. Under both EU and Dutch EA legislation the public must have access to the EA report. 97 Furthermore, at the Dutch level NCEA review is required before it is sent to the authority to decide on authorization of the activity. However, this is required only for plans and, until the forthcoming legislation comes into force, complex projects. 98

Group decision making in EA practice
The case study indicates that, in practice, group decision making is not stimulated beyond legal requirements. This also follows from other research which indicates that Dutch EA tends to be carried out because there is a legal requirement to do so, not because actors choose to do so. 99 Generally, initiators fear the costs of scoping, 100 even though research indicates that sufficient scopingwhich ideally includes the initiator, the authority, other relevant public agencies, the public, and independent advisersreduces the total costs of EAs 101 and improves their quality. 102 As such, it seems unlikely that initiators are sufficiently aware of the benefits of adequate scoping to initiate it without mandatory legislation. In addition, while NCEA review can be requested by the initiator or the planning authority at every stage of the EA process, the NCEA often is not involved if it is not mandated by law. 103 Therefore, reducing the requirements for mandatory NCEA review of final EA reports in respect of all plans and complex projects to plans only (as is envisaged with the forthcoming Dutch legislation) is likely to affect negatively the extent to which the NCEA will be involved in EA review. This can be illustrated by an amendment in 2010 that made NCEA review of final EA reports voluntary for simple projects. 104 Figure 2 shows that since 2010 the number of mandatory review procedures has declined (from over 250 in 2009 to only 50 in 2019), while the number of voluntary review procedures (where the authority requests the advice) has remained relatively stable. Similar effects may be expected when the new legislation takes force.
Consultancy firms, who often conduct the EA report preparation, also do not seem to stimulate nominal group decision making. For example, an interviewed employee of a small consultancy firm said: 'Two sets of eyes always review the EA at our consultancy firm. This could be a senior or a medior employee, depending on the complexity of the assessment'. An interviewed employee at a large consultancy firm further explained that '[a]ll EAs will be read by a second reader'. While this, at least, adds one person to the decision-making process, it can hardly be considered a group. Furthermore, it should be noted that consultants are paid by the initiator. An NCEA employee noted: 'In practice, EAs are all outsourced to consultancy firms. And if you pay for something … "who pays the piper calls the tune", as we sometimes say in jest'. This may increase the risks of many political biases. However, it may also increase the risk of groupthink.
Furthermore, while both EU and Dutch EA legislation require the public to have access to the EA report, 105 such access can contribute to nominal decision making only when the public understands the report. 106 During the interviews, consultants seemed to be well aware of the importance of how the EA report is communicated, stating that 'you have to communicate at the right time the right amount of information' and you must be aware of 'your audience and how you approach them'. Furthermore, EA reports can be over 1,000 pages long, 107 and may result in information overload. 108 To overcome this overload, a non-technical summary of the EA report may help the public and the public authority to better understand the environmental impacts of an activity. In line with EU legislation, Dutch law requires a non-technical summary. 109 Consultancy firms generally write this summary after predicting the environmental impact. The scholarly literature shows that non-technical summaries often use a scoring grid to communicate the pros and cons of the proposed activity and the alternatives. 110 These pros and cons are often communicated in the form of symbols (plus and minus) or colours (green to indicate a positive environmental impact, amber to indicate neutral environmental impact, and red to indicate a negative environmental impact). Several interviewees confirmed that this also occurs in the Netherlands. For example, an employee of a small consultancy firm said: With every environmental concern, we write something off as a plus-plus in a scoring grid. If nothing is expected to happen to the environment [for that specific concern] we will write it down as a zero. And in between we use zero minus, and minus and you describe that as well. So, for example, zero minus means there is an effect on the environment but it can be easily mitigated. We also often work with colours. The reason we communicate like this is to create an overview. Sometimes an EA is a thousand pages long, and you want that all underlying information is available, but you also want an overview. These scoring grids are meant to create an overview.
While such a summary may be appealing to overcome information overload in the public and the public authority, it also runs the risk of oversimplifying or misrepresenting the issues and trade-offs involved. 111 As such, the technical summary may increase the risk of WYSIATIespecially in those who, because of limited expertise, rely more on the technical summary, which may include authorities, the public, and the judiciary. For example, a governmental employee at the local level said: One of the tricks often used is using different colours. If you use red, that is a signalling colour, it seems very serious. But if you use purple or blue … blue does not seem serious at all. Sometimes you see in these reports that serious environmental concerns are presented in blue or pastel. Nowhere is specified which colours you are allowed to use. It makes a difference whether alarm bells will ring or not, depending on which colours are used.
An interviewee with the NCEA confirmed this: Sometimes they use colours, or smileysa happy and a sad one, or plusses and minuses. But how they reach these plusses and minuses is a trade of its own. Sometimes the text does not correlate with the plusses or minuses at all. Sometimes they lump several issues together, most of them positive, and some of them negative. It creates a distorted picture when they communicate that with a plus. These are things that we pay attention to.
The public may underestimate environmental concerns when serious environmental concerns are presented in purple or blue, or lumped together with positive attributes with a symbol, as colours and symbols have a communication value, carrying different associations and meanings. 112 Colours are found to be the most influential. 113 Strong evidence is found of explicit and implicit associations of the colour red with danger, 114 the colour green with safety, 115 and blue with neutrality. 116 Approximately 65 to 90% of product and services assessments are built on colours only. 117 As such, while a non-technical summary may be a helpful tool to inform the public and the public authority, it becomes useless when visualizations falsely trigger certain associations which may increase the prevalence of CUMB.

Recommendations for incorporating requirements on group decision making in EA legislation
Ideally, nominal decision making should be required at all stages of the EA process ( Figure 1). This includes nominal group interactions earlier in the EA process, including whether an activity requires further identification and prediction (screening), which alternatives should be examined (scoping), and the methods to be used for preparing the EA report.
EU and Dutch legislation on EA should include at least the requirement to consult independent actors who are knowledgeable about the decision in nominal groups at the screening, scoping and EA report preparation stages. Legislation could further specify which actors should be consulted; preferably these should include the initiator, relevant stakeholders, relevant administrative bodies, and independent advisers such as the NCEA.
Furthermore, to improve early and effective public participation, legislation should require independent review of the EA report for all activities, given the crucial importance of independent review to overcome the possible CUMB of those reading the report. Alternatively, a list added as an annex to EA legislation on good communication practices and/or requirements regarding visualizations in EAs could, for example, improve the quality of the information provided to the public. This, in turn, could improve the quality, comprehensiveness, effectiveness, and legitimacy of environmental decision making.

Strengthening feedback loops explained
An important CUMB that is remarkably difficult to overcome, even among experts, is the overconfidence bias. 118 This could result in those involved in the EA to underestimate the environmental concerns of a given activity. To reduce overconfidence in experts, cognitive science and behavioural economics show that frequent feedback is helpful. For example, research shows that the aspect of daily feedback makes weather forecasters remarkably accurate at weather prediction. 119 EA follow-upsuch as monitoring, evaluation, management, and communication (see Section 3)may create such feedback loops.

Requirements on feedback loops in EU and Dutch legislation
EU and Dutch EA legislation contain very limited requirements for feedback loops. The SEA Directive requires Member States to monitor the significant environmental effects of the implementation of plans. 120 The EIA Directive stipulates that Member States shall determine the procedures regarding the monitoring of significant adverse effects of projects on the environment, 121 thus allowing Member States to decide whether monitoring is needed (if no other legislation exists that mandates monitoring). Dutch law requires the planning authority to monitor the environmental impacts of plans and complex projects, 122 although it is up to the planning authority to decide when and how such monitoring will take place. For simple projects, the initiator is responsible for the monitoring if the planning authority so requires. 123 Current Dutch law stipulates that, if monitoring has taken place for an activity, the monitoring report (which is required to be written by the relevant public authority) 124 needs to be communicated to the initiator of the project, the governing bodies, and administrative bodies designated by law or by various ministries to advise on reaching the decision. 125 A report that concerns the environmental impact of a plan or complex project must also be communicated to the NCEA and the public. 126 Forthcoming Dutch legislation will no longer require planning authorities to write a monitoring report. 127

Feedback loops in practice
Research on EA more generally indicates a lack of interest or awareness among those involved in the EA process of the value of follow-up: the EA process is often considered to cover only the period before and until the planning authority makes its final decision. 128 This also followed from the interviews, where only one interviewee, a governmental employee at the Dutch local level, mentioned the importance of monitoring and the lack of monitoring requests: Monitoring is incredibly important. And in the Netherlands, it is still optional, not mandatory [for projects]. Nobody looks at it either. Nobody demands it. There is no environmental inspector that comes to me and asks 'Where is your latest monitoring report?' It just doesn't happen.
Furthermore, the interviews indicated that, in practice, when monitoring does happen, consultancy firms are not in the feedback loop. This is problematic as research indicates a vital need to introduce feedback to learn from experience and to avoid the constant 'reinventing of the wheel' in EAs. 129

Recommendations for incorporating feedback loops in EA legislation
Legislation could be strengthened by providing further requirements for the communication of follow-up, 130 most importantly by requiring a monitoring report (which is not required at the EU level and will no longer be required in the Netherlands with the new legislation). 131 Importantly, this monitoring report should be communicated to expert groups, such as the NCEA, and to consultancy firms, as they are often involved in the prediction of environmental impact.


While (bounded) rationality and political bias have been discussed extensively in the EA literature, recent insights on CUMB have not yet been fully explored. This article provided a selective review of how three debiasing techniquesexpertise, nominal group decision making, and frequent feedback loopsare, or should be, incorporated into (supra)national EA legislation. This article shows that neither EU nor Dutch legislation (yet) incorporates these debiasing techniques sufficiently. Specifically, the article underlines the importance of action at the EU level, as Member States, including the Netherlands, are increasingly reducing their requirements on debiasing techniques to bring their legislation more in line with the minimum requirements set by the EU. The article, therefore, recommends that EA legislation generally, and EU and Dutch legislation in particular, mandate (i) an independent foundation to check the EA report for plans and projects; (ii) nominal decision making at all stages of the EA process; and (iii) the communication of a monitoring report to expert groups and consultancy firms.
These recommendations are expected to improve environmental decision making; increase the impact of EAs on the design of initiatives and their modifications; and reduce the costs of delay, judicial review, and environmental harm. Furthermore, these recommendations should be incorporated, given that the emergence of CUMB in the EA process reduces the effectiveness of public and judicial review as the enforcement mechanism of sound environmental decision making, for two reasons. Firstly, if authorities have the political will and the power to make sound environmental decisions, they may fail to do so as a result of CUMB in the EA process. This may exacerbate CUMB in the public and the judiciary, in particular the WYSIATI bias. Secondly, initiators, as well as authorities that do not have the political will to make sound environmental decisions, may misuse the CUMB of the public and the judiciary to promote activities that harm the environment.
This article provides a better understanding of how insights from cognitive science and behavioural economics could be used to inform EA legislation. However, despite the impressive empirical evidence accumulated on CUMB and their potential debiasing techniques, cognitive research has been criticized for its lack of precision, refinement, and progress at the theoretical level. 132 Future research, therefore, is needed to develop theoretical frameworks and to refine hypotheses through empirical studies. In the context of this article, further research should experimentally test whether and to what extent CUMB may appear among those specifically involved in the EA process, as well as the role of different types of debiasing technique (such as voluntary or mandatory requirements) in affecting human behaviour.
Supplementary material. The supplementary material for this article can be found at: https://doi.org/10.1017/S2047102523000031.