Expert performance was coined as a key construct for theorizing expert performance (Ericsson, 2007; Debarnot et al., 2014; Hashimoto et al., 2015). Experts are engaged in providing knowledge-intensive professional services for complex questions. Traditionally, a closed system perspective is adopted to interpret experts’ long-time superior performance (Ericsson, 2008; Ericsson and Harwell, 2019). The long-time superior performance is tied to the accumulation of deliberate practices and feedback that they encode and the refinery of inherent representations memorized for effective use in a determinate world (Ericsson and Kintsch, 1995). Similarly, novices may be likely to produce opinions and judgments as experts do for the same matters if they have enough knowledge and experience (Spence, 1993). As revealed in scientific judgment and the improvement of learning patterns (Krueger et al., 2012; Campitelli et al., 2015; Friedman and Korman, 2019), inherent cognition, including knowledge and experience, pertains to the generation of expert performance. In a closed system, experts’ performances are evaluated without considering the fluctuations in situations they are working with. Unexpected emergencies are precluded in the determination of expert performance.

However, the “closed system” perspective is subject to considerable flaws, as indicated by quantum physics and social and psychological science (Glimcher, 2005). Researchers have thus claimed to advance the closed system to embrace indeterminacy in the research area of brain and behavior (Gigerenzer et al., 2000; Hastie, 2001; Schall, 2004). An open system is consequently framed. Based on the tenet of the open-system philosophy, the determinants of expert performance are beyond the boundary of inherent cognition. For example, an expert’s performance does not necessarily exceed that of a novice, regardless of whether the problem is relatively simple or complicated. Chi (2006) and Spence (1993) indicated that inherent cognition might not realize better performance in judgment and decision making as experts must adapt to a new environment. Over the years, with the development of ecological cognitive science, Araújo et al. (2006) proposed an effective way to obtain superior expert performance by capturing specific environmental and perceptual information. Therefore, improving perceptions about environmental factors and owning privileged access to refined inherent representations are crucial to expert performance (Ericsson and Kintsch, 1995).

When decision-making time and support resources are stretched, the external environment becomes a predominant factor in the formation of expert performance. This is the case in competitive bidding in the construction sector, where experts are highly involved in evaluating submitted bids. As a result, experts must produce professional services to underpin the determination of winners in competitive bid evaluations. To address the issue, previous studies have proposed approaches and models to aid experts in evaluating bids (Liu et al., 2017; Semaan and Salem, 2017). However, they often face inadequate information and uncertain circumstances (Zhen-Song et al., 2021), suggesting that the interaction between inherent cognition and the external environment is not monotonic.

Previous studies have offered a few approaches to examining the links between inherent cognition and the external environment, such as measuring eye movements and verbal reports (Macmahon and McPherson, 2009; Afonso et al., 2012). The approaches address bid evaluation task constraints and behavior settings in experimental. However, laboratory-based simulations represent a real-life situation where experts are vulnerable to biases in decision-making (Araújo et al., 2006). Notwithstanding the complex factors of bid evaluation, expert performance has been paid extensive attention, Kardes (2006), Burgman et al. (2011), and Connor et al. (2020) proposed that it was better to redefine expert performance to account for those contextual and relational factors of expert performance.

According to the open-system concept, we outlined two dimensions of expert performance: inherent cognition and the external environment. A conceptual framework is proposed below to describe the mechanism of expert performance (Figure 1).

As Figure 1 tells, the formation mechanism of expert performance contains two parts: memory and ecological psychology. Inherent cognition is grounded on theories of memory, of which the main components are knowledge, experience, and qualifications (Ericsson, 2003). The formation of inherent cognition contains a loop in which knowledge, experience, and qualifications give a stimulus to memory. A changed memory enlarges the packet of knowledge, experience, and qualifications. This loop mirrors the formation of inherent cognition (Ericsson, 2007). To align with the enlargement, deliberate practice, encoding, and transformation transform external stimuli into meaningful representations (Kording and Wolpert, 2006). Meanwhile, ecological psychology is referred to explain the impacts of the external environment on expert performance, assuming a person–environment system (PES; Araújo et al., 2006). As Turvey et al. (1981) pointed out, actions, judgment, and performance are based on the lawful connections between individuals and the environment in which they act. Therefore, experts are engaged in social culture, rules, and regulations belonging to the external environment.

Experts utilize external environmental factors to avoid biased decisions (Fischhoff et al., 1982; Ludger and Erik, 2016; Contreras-Pacheco et al., 2020). However, the factors also pose physical, goal-related, and task-related constraints to expert performance (Araújo et al., 2006). These constraints cause experts to feel physical discomfort (Liu et al., 2015), peer pressure (Bohannon, 2013), motivation deviation (Ericsson et al., 2009), loose supervision (Gonsalvez et al., 2017), and weak feedback (Kardes, 2006). Furthermore, given a specific external environment, expert performance may combine inherent cognition with the external environment to present a holistic “person–environment system.” Expert performance results from dynamic, continuous interaction within the PES, which is influenced by inherent cognition and external environmental perceptions.

According to ecological psychology theory, bid-evaluation experts’ inherent cognition and perception of the external environment are embedded in each other, emphasizing that the interaction between experts and the environment is an important motivation for expert performance. In the PES, bid-evaluation experts have many opportunities to address the opportunities or possibilities for actions, known as affordances, and seize a set of objective and physical external environmental factors to improve their performance (Costall, 1984). Regarding bid-evaluation experts, perceiving an “affordance” means perceiving how one can make decisions under specific bid evaluation conditions (Araújo et al., 2019).

In the bid evaluation process, experts are supposed to integrate multiple attributes into the overall description of bidders. This is a perceptual process in which bidders’ redundant and irrelevant attributes are removed, and their core attributes are framed. Finally, the experts form inherent cognitive “representations” of the bidders. Interweaving in the environment and perceptual processes (from reactions to relevant external objects and events), this cognitive process is dynamic. Therefore, a bid evaluation task is one in which “affordance” captures the interaction between experts and the environment (e.g., peer pressure, time pressure, supervision system, and opportunity cost) with a concatenation of interdependent decisions over time. At the end of bid evaluations, experts provide feedback by reflecting on or summarizing the results from the external environment to change inherent cognition. For example, whether a bid award decision is passable is determined not by its absolute attribution representation (whether measured in performance, qualification, or scale) but rather by how it relates to the substantive responses of an individual bidder to the rules, including technical requirements, economic indicators, and commitments.

We detected these factors’ dependency/driving power using cross-impact matrix multiplication, which is often applied to classification and interpretative structural modeling (MICMAC-ISM). Warfield (1974) established a computer-assisted learning process called interpretive structural modeling (ISM) to transform unclear and poorly articulated mental models into well-defined multi-level structural models through experts’ practical experience and knowledge. As many interrelated factors affect expert performance in bid evaluation, we selected ISM for our research methodology. The reason not only goes to a well-established methodology for identifying relationships among specific items but also to providing a fundamental understanding of complex situations. Therefore, this study established an ISM-based hierarchical structure model to clarify the dynamics of relationships among expert performance factors encountered in the bid evaluation practices. The technical route of the ISM is shown in Figure 2.

Following the literature review, we categorized expert performance dimensions into 16 factors. A questionnaire was designed, and 34 experts (from construction, design, cost consulting enterprises, research institutions, and government departments) were interviewed to test the reliability of these dimensions and factors. The interviewees were selected based on their bid-evaluation qualifications and experience, and a deep understanding of construction contracts. The profiles of the interviewees are given in Table 1. To allow them to understand the relationships among the factors, we conducted face-to-face interviews with all the experts. As a result, three factors were removed, and a new one was added. Finally, the remaining 14 bid-evaluation expert performance factors were compiled. A summary of these factors is presented in Table 2.

The relationships among the 14 factors were confirmed via interviews with 34 experts. Using ISM, a “leads to” (one factor leading to another) contextual-type relationship was chosen to identify the interacting position of each factor for analysis. An adjacency 14*14 matrices of the identified impact factor elements (1 = Yes; 0 = No) was developed as an option for paired comparison between elements. The presence or absence of a relationship directed from element i to j was indicated by placing “1” or “0,” respectively, in the corresponding intersection of row i and column j. After that, information was sent to experts for comments. Based on their responses, we used the Delphi technique to obtain common views from the participants. When 80% of the interviewees agreed upon the relation, it was thought to have a consensus. The threshold value for the reliability of factors was 0.80. Adjacency reachability matrix A, indicating the relationship between elements, is presented in Table 3.

From the adjacency reachability matrix A, redundant relationships between elements were eliminated through an iterative process. A diagram established the final relationship level between elements in a hierarchical form. As the adjacency reachability matrix A is a Boolean matrix, we calculated it by applying the following Boolean algorithm: let I ( n × n ) = [ 1 0 ⋯ 0 0 1 ⋯ 0 ⋮ ⋮ ⋱ ⋮ 0 0 ⋯ 1. ] (1) R = ( A + I ) ( n + 1 ) = ( A + I ) n ≠ ⋯ ≠ ( A + I ) 2 ≠ ( A + I )

where R = ( A + I ) n is the reachability matrix R of matrix A.

According to adjacency reachability matrix I of the bid-evaluation expert performance, the calculation process was implemented in MATLAB (Lindfield and Penny, 2012). When n = 5 , the equation M = ( A + I ) ( n + 1 ) = ( A + I ) n ≠ ⋯ ≠ ( A + I ) 2 ≠ ( A + I ) held. The derived result and the reachability matrix are shown in Table 4.

The reachability matrix obtained above was partitioned by deriving the reachability set and the antecedent set to establish the hierarchy model of the ISM. The reachability set for each element represented a set of elements (i.e., several risk elements, including itself) upon which the current element had an impact. For example, in the ith horizontal row Si of reachability matrix R, if S i j = 1 ( j = 1 , 2 , ⋯ , n ) , the element Sij is placed in the reachability set, expressed as Si. The antecedent set reflects a set of elements that affected the current element. Along the same lines, in the jth column Sj of reachability matrix R, if R i j = 1 i = 1 , 2 , ⥂ ⋯ , n , the element Rij is placed in the antecedent set, expressed as Aj. The intersection of these sets S i ∩ A j was derived for all elements. If Si is a complete subset of Aj, the element or elements were moved from the reachability matrix and assigned a specific level.

After the iteration, the reachability set for the elements Expertise (S1) and Number of bid evaluations (S5) was a complete subset of the antecedent set; therefore, it was considered the highest level of the elements removed from the reachability matrix. On the other hand, the elements Stress situations (S10), Natural Environment (S11), and Distance (S12) occupied the reachability sets and the antecedent sets on their own, which implies that these factors were isolated from other elements. They were, therefore, removed from the reachability matrix and placed on the first level. The iterative processes were continued in this manner to identify the different levels. Consequently, the hierarchy model of the ISM consisted of all these levels. The results of the final iteration are given in Table 5.

A diagram explains the contextual relationship between an impact factor element and its hierarchy. First, as shown in Table 5, the elements taken from the reachability matrix in the previous step were placed at the highest level of the hierarchy. Thus, Expertise (S1), Number of bid evaluations (S5), Stress situations (S10), Natural environment (S11), and Distance (S12) appeared at the top. Next, the elements of Academic ability (S2), Motivation preference (S3), Years qualified as an expert (S4), and Rewards (S14) were removed before the next partition at the second level and placed below the top level. This process was repeated until all the elements were rearranged, creating a four-layer hierarchical structural diagram of the expert performance factors in bid evaluation (Figure 3).

Using MICMAC analysis, it is necessary to prove the relationships among impact factor elements based on the attained diagram. We used this approach to analyze a factor’s dependency and driving power. The analysis complements experienced users’ impressions drawn from the visual analysis of influence structures. To better understand the significance of certain elements and their influence on others, we used the MICMAC to classify the factors into four clusters (autonomous, dependent, linkage, and independent) according to their driving and dependence power. The driving-and-dependence power of an element was computed via the summation of the corresponding rows. Similarly, the dependence power was computed via the summation of the corresponding rows and columns, respectively. Finally, each element was plotted on the driving-dependence power matrix (Figure 4).

The first cluster represents “autonomous factors” (weak driving and dependence power). It includes Stress situations (S10), Natural environment (S11), Distance (S12), Strength (S13), Rewards (S14), Academic ability (S2), Motivation preference (S3), Years qualified as an expert (S4), Morality (S6), and Objective (S7). This set of factors reflects complicated situations. The “dependent factors” belonged to the second cluster (weak driving, strong dependence power) and ranked highest in the importance levels of the ISM-based hierarchical model. Two factors were assigned: Expertise (S1) and the number of bid evaluations (S5). Their dependence indicates that they depended on other factors to be resolved during the bid-evaluation expert performance. Therefore, bid-evaluation experts should consider all other factors to achieve the dependent factors and retain superior expert performance. The third cluster specifies the “linkage factors” (strong driving and dependence power). Any change occurring to these factors would significantly impact other factors in return. No factors correlated with this category in our study, implying that none of them had strong driving and dependence power. The “independent factors” (strong driving, weak dependence power) in the fourth cluster formed the foundation of the ISM hierarchical model and should be given priority. The two factors in this set, Situation perception (S8) and Supervision system (S9), are vital, and bid-evaluation experts and policymakers should focus on them for superior bid-evaluation expert performance.

The two dimensions of internal cognition and the external environment were independent through theoretical deduction. In the PES, the interaction between the “perceptions” and “affordance” of experts is manifested as expert performance, which feeds back or verifies internal cognition and external environment, forming a “closed system” (Figure 1). However, in the context of engineering, 14 factors affect expert performance, seven of which are from the internal cognitive dimension and seven from the external environmental dimension. We could observe that the two-dimensional factors were scattered on four levels (Figure 3). Inherent cognitive factors influenced each other and were scattered at the 1–3 level. Most of them had strong dependence and were driven and influenced by external environmental dimension factors. Level 2–4 was PES, with perception (e.g., body perception) and supply (e.g., the regulatory system) at the lowest end, the most important driving force in the structural system that influences other factors to work together.

The three factors under the heading of the external environment, namely pressure, natural environment, and working distance, together with expertise and bid evaluation times (internal cognitive dimensions), were at the first level of the hierarchical structure system and were the most direct influencing factors. However, the three factors of the environmental dimension existed independently of any other factors, leaving the structure in an open form. These phenomena reflect the organic integration of memory and ecological cognition theories and completely describe the factors affecting expert performance in engineering. These influencing factors differ from the characteristic cognitive ability mentioned in the ecological dynamics theory of sports decision-making with the help of regularity and universality (Araújo et al., 2006). For expert-performance factors with engineering characteristics, inherent cognition has a dynamic influence on expert performance under the constraint of a unique external environment. In contrast, the external environmental factors are more strongly driven or have a direct and independent influence.

In the bid evaluation, considering independent factors (strong driving and weak dependence power from cluster four), situation perception reflects the experts’ feelings regarding the environmental properties, including seating, air humidity, and physical comfort. Negative feelings will hinder expert performance as perception, cognition, and behavior are integrated processes. The nature and type of cognition are influenced by the interaction between the body and the external environment. For example, situation perception, relating to concepts such as fatigue (Orazbayev, 2017), is a hidden expert-performance-influencing factor that few researchers focus on. The perception of “affordance” has a dynamic quality and can change (Fajen et al., 2009). Improved working conditions can maintain enhanced situation perceptions and effectively obtain sustainable and superior expert performance. The other independent factor with strong driving power, a sustainable supervision system rather than self-restriction, is also confirmed to improve expert performance. Environmental properties directly inform experts about what they can and cannot do in a performance context (Michaels, 2000). In essence, the confluence of constraints and perception determines the stability and instability of expert performance.

At level III of the ISM, the factors of morality, objective, and strength precede motivation preference (Figure 3, Level II), while motivation preference, academic ability, and years qualified as an expert (all level II) affect each other bilaterally. Regarding the level II factor, rewards are influenced by the factor of strength. Therefore, these factors with higher driving power in the autonomous cluster (such as Strength) can impact each other. Therefore, their promotion should be given priority in attaining superior expert performance. In contrast, the factors with higher dependence on power are influenced by other factors. Consistent with current studies, academic ability, motivation preference, and years qualified as an expert (Figure 3, Level III) are key factors determining expert performance as acquired skills and experience in bid evaluation.

More factors from inherent cognition and the external environment are plotted at the top of the four-level diagram (Figure 3, Level IV), indicating that factors from the two dimensions collectively determine expert performance. These factors include the inherent cognition dimension (expertise, number of bid evaluations) and the external environment dimension (stress situations, natural environment, and distance).

Previous studies have demonstrated that expertise and exceptional performance are highly reproducible regarding inherent cognition. There is no doubt that unfailed successful learning is necessary for experts to achieve an improved or higher level of performance (Ericsson and Lehmann, 1996). In this regard, of all the factors at the top level, expertise and the number of bid evaluations, both belonging to the inherent cognition dimension, are viewed as the most natural elements influencing expert performance. The other factors: stress situations, natural environment, and distance, which belong to the external environment, are isolated and have no relationship with other factors. This phenomenon shows that the factors affecting experts’ performance are in an open system, and some external environmental factors will directly impact them in some situations. Especially in the bid evaluation process, the pressure from peers or the physical impact of the natural environment on experts (e.g., fatigue caused by long-distance travel to the workplace or bad weather) will make expert performances deviate.

The situation perception and supervision system factors (Figure 3, Level IV) may forego morality, objective, and strength (Figure 3, Level III), determining expert performance in the bid evaluation. These are key issues to be addressed to realize superior expert performance. Moreover, Ericsson and Lehmann (1996) showed that situation perception affects labor intensity, whereas a strict supervision system initiates moral restraint. Thus, morality and some other factors (including procedural fairness, technical competence, and general reputation) impact and interact with expert performance. Still, illegalities often occur despite the vital importance of limiting moral deviation during expert behavior. For instance, bid-evaluation experts and bidders collude to manipulate the bidding of private customers (forcing the client to pay high prices). In addition, they colluded on certain municipal contracts, causing huge economic losses to New York taxpayers (TargetedNewsService, 2018). It has been reported that experts’ antagonistic feelings may hinder objectivity and obscure experts’ fairness (Lindfield and Penny, 2012).

In particular, three environmental constraint factors (stress situation, natural environment, and distance), located in the top layer, are mainly manifested as physical perception in the natural environment. They do not participate in the PES due to no correlation with any factor in internal cognition and other factors in the external environmental dimension. They make the system open, and these factors show that they directly, solely, and openly affect expert performance in engineering. This contrasts with the traditional information processing methods for decision-making in the “open system.” When decision-makers calculate and select options in the psychological or neural model, the influence of more uncertain external environment factors should be considered to maximize performance effectiveness.

Our results show that inherent cognitive and external environmental factors are characterized by high dependence and strong driving power, respectively (Figure 4). For example, Figure 3 Level IV shows that situation perception and supervision system (external environment dimension) are antecedent variables driving expert performance factors and should not be precluded first. However, the classification of factors (Figure 4) shows that morality and objective (internal cognitive dimension) also have higher driving and lower dependence power. This finding suggests that these two special factors bridge the inherent cognitive and external environmental dimensions. Moreover, morality and objectivity (Figure 3, Level III), controlled by the supervision mechanism (Figure 3, Level IV), influence inherent cognition dimension factors.

We confirmed the relationship and hierarchical structure of the two dimensions, suggesting that environmental constraint factors other than inherent cognition contribute to individual expert performance differences in bid evaluation. Furthermore, they show that external environmental factors strongly influence the externalization of inherent cognition. At a practical level, our findings on the two dimensions affirm existing research, supporting the notion that expert performance depends on professional knowledge and the environment in engineering. Furthermore, the findings also provide a basis for exploring ecological cognition based on the idea that the “brain–body-environment system is embedded and embodied.”