Learning and transfer in dynamic decision environments

An important aspect of learning is the ability to transfer knowledge to new contexts. However, in dynamic decision tasks, such as bargaining, firefighting, and process control, where decision makers must make repeated decisions under time pressure and outcome feedback may relate to any of a number of decisions, such transfer has proven elusive. This paper proposes a two-stage connectionist model which hypothesizes that decision makers learn to identify categories of evidence requiring similar decisions as they perform in dynamic environments. The model suggests conditions under which decision makers will be able to use this ability to help them in novel situations. These predictions are compared against those of a one-stage decision model that does not learn evidence categories, as is common in many current theories of repeated decision making. Both models' predictions are then tested against the performance of decision makers in an Internet bargaining task. Both models correctly predict aspects of decision makers' learning under different interventions. The two-stage model provides closer fits to decision maker performance in a new, related bargaining task and accounts for important features of higher-performing decision makers' learning. Although frequently omitted in recent accounts of repeated decision making, the processes of evidence category formation described by the two-stage model appear critical in understanding the extent to which decision makers learn from feedback in dynamic tasks. Faison (Bud) Gibson is an Assistant Professor at College of Business, Eastern Michigan University. He has extensive experience developing and empirically testing models of decision behavior in dynamic decision environments.     

Is It Better to Forget? Stimulus-Response, Prediction, and the Weight of Past Experience in a Fast-Paced Bargaining Task

Decision makers in dynamic environments such as air traffic control, firefighting, and call center operations adapt in real-time using outcome feedback. Understanding this adaptation is important for influencing and improving the decisions made. Recently, stimulus-response (S-R) learning models have been proposed as explanations for decision makers' adaptation. S-R models hypothesize that decision makers choose an action option based on their anticipation of its success. Decision makers learn by accumulating evidence over action options and combining that evidence with prior expectations. This study examines a standard S-R model and a simple variation of this model, in which past experience may receive an extremely low weight, as explanations for decision makers' adaptation in an evolving Internet-based bargaining environment. In Experiment 1, decision makers are taught to predict behavior in a bargaining task that follows rules that may be the opposite of, congruent to, or unrelated to a second task in which they must choose the deal terms they will offer. Both models provide a good account of the prediction task. However, only the second model, in which decision makers heavily discount all but the most recent past experience, provides a good account of subsequent behavior in the second task. To test whether Experiment 1 artificially related choice behavior and prediction, a second experiment examines both models' predictions concerning the effects of bargaining experience on subsequent prediction. In this study, decision models where long-term experience plays a dominating role do not appear to provide adequate explanations of decision makers' adaptation to their opponent's changing response behavior.     

基于前景理论的两阶段DEA交叉效率评价模型

针对两阶段系统交叉效率值不唯一的问题,现有解决方法大多假设决策者完全理性,忽略了决策者心理因素对评价过程的影响。本文构建以平均值为新的参考点,分别在集中和分散决策环境下建立基于前景理论的两阶段DEA交叉效率评价模型,通过实际算例验证模型的可行性和有效性。     

An Experimental Assessment of Coleman's Linear System of Action for Supporting Policy Negotiations

We report on an experimental assessment of the applicability of Coleman's Linear System of Action (LSA) to policy negotiations. In LSA, policy negotiations are modeled as exchange of control over issues. LSA allows one to compute the equilibrium control distribution for a group of decision makers, based on their distributions of preferences for, and control over, the issues at stake. The LSA theory, however, does not address the process of how the decision makers should exchange control over the issues to arrive at the equilibrium control. We test in an experimental setting partial and full LSA-based support vs. no support, for three different social structures among the decision makers engaging in repeated bilateral negotiation rounds. The results of the laboratory experiment indicate that the level of LSA support affects the negotiation process and the efficiency of reaching the equilibrium, while the influence of the decision makers' social structure is less clear. These results suggest that Coleman's LSA concept can be employed to support policy negotiations in a practical setting.     

社会网络环境下基于“信任—知识”模型的风险性大群体应急决策方法

针对大群体应急决策专家之间信任关系及其传递引发的决策风险,以及由于大群体中个体偏好差异较大导致生成独立聚集等问题。首先,提出一个“信任—知识模型”对决策专家之间的信任关系进行集成和传递,并根据决策专家的信任风险偏好得出决策专家之间的信任知识度网络;其次,利用Louvain算法对信任知识度网络进行聚类,高效快速的获得若干个聚集,并用社会网络分析技术确定每个决策者和聚集的权重;然后对每个聚集中的决策者偏好进行集结,并综合决策者给出的信息对备选决策方案进行排序。最后,通过案例分析和对比验证了所提方法的合理性与有效性。     

On the application of TRIMAP to problems with multiple decision makers

In this paper the potentialities of TRIMAP to provide decision support in multiobjective problems with multiple decision makers are exploited. TRIMAP is an interactive three-objective linear programming package which enables a progressive and selective learning of the nondominated solution set. The aim is to aid the opposing parties in exploring their own preferences and to explore the dynamic nature of the negotiation process.     

The sugar production factory—A dynamic decision task

This article describes the Sugar Production Factor and its structural equivalent, the Personal Interaction task. These are two simple, individual dynamic decision-making tasks in which subjects make interdependent decisions to reach a goal, and receive feedback on the outcome of their efforts along the way. An important result from human learning experiments using these two tasks and their variants is that subjects reliably improve their ability to reach the goal over a moderate number of training trials (40–90) but do not show consistent improvement in other measures of task knowledge. These other measures focus on subjects' ability to accurately predict the task environment's response to their actions and subjects' ability to produce useful heuristics. This pattern of results runs counter to the idea that decision makers' performance in dynamic decision tasks depends critically on the predictive accuracy their internal models of the task environment. Variants of both tasks have been used to manipulate this pattern of results and explore more deeply the nature of the internal models that subjects form of the task environment. These variants are discussed in the context of other relevant findings in the dynamic decision making literature.     

A note on system modelling and control in the case of conflicting objectives

The paper is concerned with the problem of system modelling in the case of many decision makers with conflicting goals. The suggested approach, preserving much of the tractability of the standard LQG problems, allows to take into account decision makers' risk attitudes. This revised version was published online in June 2006 with corrections to the Cover Date.     

A multiple criteria decision making approach to manure management systems in the Netherlands

The intensification of livestock operations in the last few decades has resulted in an increased social concern over the environmental impacts of livestock operations and thus making appropriate manure management decisions increasingly important. A socially acceptable manure management system that simultaneously achieves the pressing environmental objectives while balancing the socio-economic welfare of farmers and society at large is needed. Manure management decisions involve a number of decision makers with different and conflicting views of what is acceptable in the context of sustainable development. This paper developed a decision-making tool based on a multiple criteria decision making (MCDM) approach to address the manure management problems in the Netherlands. This paper has demonstrated the application of compromise programming and goal programming to evaluate key trade-offs between socio-economic benefits and environmental sustainability of manure management systems while taking decision makers’ conflicting views of the different criteria into account. The proposed methodology is a useful tool in assisting decision makers and policy makers in designing policies that enhance the introduction of economically, socially and environmentally sustainable manure management systems.     

Spatial multicriteria decision support for robust land-use suitability: The case of landfill site selection in Northeastern Greece

Multicriteria spatial decision support systems (MC-SDSS) have emerged as an integration of geographical information systems (GIS) and multiple criteria decision aid (MCDA) methods for incorporating conflicting objectives and decision makers’ preferences into spatial decision models. In this paper, we present spatial UTASTAR (S-UTASTAR), a raster-based MC-SDSS for land-use suitability analysis. The multicriteria component of the system is based on the UTA-type disaggregation-aggregation approach. S-UTASTAR is applied in a raster-based case study concerning land-use suitability analysis to identify appropriate municipal solid waste landfill (MSW) sites in Northeast Greece. Moreover, robustness analysis tools are implemented to guarantee robust decision support results. More specifically, during the aggregation phase, the Stochastic Multiobjective Acceptability Analysis (SMAA) is used to indicate the frequency at which a site achieves the best ranking positions within a large set of alternative landfill sites.     

Bad numbers: coping with flawed decision support

Evidence is accumulating that many spreadsheet-based decision support systems contain errors. These errors can result in bad numbers, which in turn could lead to bad decisions. We review the literature on the origins and consequences of bad (erroneous) numbers produced by models and/or decision support systems built around them. Then, we present a case study in which an experiment in visual support for a sequential decision-making task was tainted by bad numbers. Both the literature review and the experiment indicate a robust human ability to overcome flawed decision support. We conclude with questions that need to be addressed in order to better understand the capabilities of humans to deal with erroneous results from decision support systems.     

Allocation planning under service-level contracts

Motivated by the practical limitations of current demand fulfillment systems, this paper addresses the problem of allocation planning under service-level contracts in a multi-period setting. We provide a formal definition of the allocation planning problem under a type of service-level contract that is particularly relevant to manufacturing industries and formulate a corresponding stochastic dynamic program. Based on a rigorous formal analysis of the dynamic program, we derive the requirements a “good” allocation policy should meet and use them to evaluate the heuristic policies proposed in the literature and to derive new allocation policies that may enhance the performance of allocation planning under service-level contracts. After detailed characterization and discussion of these new policies, we present the results of an extensive numerical study that allow us to quantify and compare allocation policies’ performance and to derive recommendations for decision makers in practice.     

The impact of DSS use and information load on errors and decision quality

This paper uses a laboratory experiment to examine the effect of DSS use on the decision maker’s error patterns and decision quality. The DSS used in our experiments is the widely used Expert Choice (EC) implementation of the Analytic Hierarchy Process. Perhaps surprisingly, our experiments do not provide general support for the often tacit assumption that the use of a DSS such as EC improves decision quality. Rather, we find that, whereas a DSS can help decision makers develop a better understanding of the essence of a decision problem and can reduce logical error (especially if the information load is high), it is also susceptible to introducing accidental effects such as mechanical errors. In some cases, as in our study, the accidental errors may outweigh the benefits of using a DSS, leading to lower quality decisions.     

A Dynamic Perspective of Differences Between Cognitive Maps

To complement current methods used in measuring the differences between cognitive maps, the application of neural network tools is suggested. The neural network approach aids decision makers in the identification of dynamic aspects of target systems which are represented by the cognitive maps to be compared. A real data application of this method shows that the proposed technique can be used to enhance the understanding of a decision environment.     

Leader-follower strategy via a sliding mode approach

In this paper, the decision making problem in continuoustime dynamic systems is considered for the situation with two decision makers and a hierarchical decision structure. The leader-follower strategy is studied. To implement the leader's strategy, we propose to use a sliding mode approach, which allows the leader to constrain the state of the system within some manifold of the state space and forces the follower to choose the strategy preferable for the leader. The corresponding sliding manifolds are derived from the classical variational problem formulation for a class of systems whose right-hand side is affine with respect to the two control inputs. Numerical examples are considered with simulations to illustrate the technique.The authors wish to express their thanks to Dr. Vadim Utkin for his helpful discussions.     

MOIRA: A decision support system for decision making on aquatic ecosystems contaminated by radioactive fallout

Interventions to restore radionuclide contaminated aquatic ecosystems may reduce individual and collective radiation doses, but may also result in detrimental ecological, social and economic effects. Decision makers must carefully evaluate possible impacts before choosing a countermeasure, hence decision analysis methods constitute an important aid to rank intervention strategies after the contamination of an aquatic ecosystem. We describe MOIRA, a decision support system for the identification of optimal remedial strategies to restore water systems after accidental introduction of radioactive substances. MOIRA includes an evaluation module based on a multi-attribute value model to rank alternatives and a module to perform multiparametric sensitivity analyses, both with respect to weights and values, to allow us to gain insights into the problem. The problem is under certainty since the validation of models used to quantify countermeasure impacts suggests little uncertainty in policy effects.The system is implemented in a PC based decision support system which allows the inclusion of all relevant information.     

Learning in Dynamic Decision Making: The Recognition Process

The apparent difficulty that humans experience when asked to manage dynamic complexity might be related to their inability to discriminate among familiar classes of objects (i.e., flawed recognition). In this study we examined the change in individuals' recognition ability, as measured by the change in the similarity of decisions they made when confronted repeatedly with consistent dynamic situations of varying degrees of similarity. The study generated two primary findings. First, decisions became increasingly similar with task practice, a result that suggests gradually improving discrimination by the participants. Second, the similarity was determined by the interaction of many task features rather than individual task features. The general principles highlighted by this study are applicable to dynamic situations. For example, with practice, decision makers should be able to learn to identify the time at which to intervene to achieve the maximal effect during dynamic decision making.     

基于后悔理论的灰色应急决策方案动态调整方法

突发事件发展具有可变性、动态性、随机性等特点,这要求决策者能根据实际情况及时调整应急方案,在此过程中决策者往往表现出“有限理性”的心理特征。针对以上情形,提出一种基于后悔理论的决策方法,以解决不确定环境下考虑决策者心理因素的应急方案动态调整问题。该方法首先描述分析了基于灰数信息的应急调整方案的生成过程;然后从后悔规避的视角构造了调整方案集关于处置效果、调整成本、应对损失三方面的灰色感知效用矩阵;进一步,用转移概率矩阵预测突发事件的演化概率,计算各调整方案的灰色综合感知效用值以选出最佳调整方案;最后,通过实例验证了该方法的可行性和有效性。结果表明,该方法贴近决策实际、具有较强的实用性;能够为应急决策的方案调整问题提供方法指导和理论支持。     

Decision maker preferences for information in complex decision making: New directions for OR intervention

Operations research models are used in many business and non-business entities to support a variety of decision making activities, primarily well-defined, operational decisions. This is due to the traditional emphasis of these models on optimal solutions to pre-specified problems. Some attempts have been made to use OR models in support of more complex, strategic decision making. Traditionally, these models have been developed without explicit consideration for the information processing abilities and limitations of the decision makers, who interact with, provide input to, and receive output from such models.Research in judgement and decision making show that human decisions are influenced by a number of factors including, but not limited to, information presentation modes; information content, modes, e.g., quantitative versus qualitative; order effects such as primacy, recency; and simultaneous versus sequential presentation of data.This article presents empirical research findings involving executive business decision makers and their preferences for information in decision making scenarios. These preference functions were evaluated using OR techniques. The results indicate that decision makers view information in different ways. Some decision makers prefer qualitative, narrative, social information, whereas other prefer quantitative, numerical, firm specific information. Results also show that decision making tasks influence the preference structure of decision makers, but that in general, the preference are relatively stable across tasks.The results imply that for OR models to be more useful in support of non-routine decision making, attention needs to be focused on the information content and presentation effects of model inputs and outputs.