Environments That Make Us Smart: Ecological Rationality

ABSTRACT— Traditional views of rationality posit general-purpose decision mechanisms based on logic or optimization. The study of ecological rationality focuses on uncovering the "adaptive toolbox" of domain-specific simple heuristics that real, computationally bounded minds employ, and explaining how these heuristics produce accurate decisions by exploiting the structures of information in the environments in which they are applied. Knowing when and how people use particular heuristics can facilitate the shaping of environments to engender better decisions.     

研究人类使用启发式之五原则(英文)

快速节俭启发式框架假定人们使用一套简捷的决策策略——启发式——进行推理、选择、评价及其他决策。这些启发式策略能顺应任务情境结构中的规律, 利用人类的基本认知能力。正基于此, 启发式成就了适应性行为。本文拟对启发式框架进行回顾, 并简要陈述引导研究者研究人类适应性工具箱的五条原则。我们强调, 启发式模型应(ⅰ)精确界定(ⅱ)对照检验(ⅲ)与策略选择理论相符(ⅳ)能评估其对新资料的预测力(ⅴ)能既在实验室又在现实世界中得以检验。     

Précis of Simple heuristics that make us smart

How can anyone be rational in a world where knowledge is limited, time is pressing, and deep thought is often an unattainable luxury? Traditional models of unbounded rationality and optimization in cognitive science, economics, and animal behavior have tended to view decision-makers as possessing supernatural powers of reason, limitless knowledge, and endless time. But understanding decisions in the real world requires a more psychologically plausible notion of bounded rationality. In Simple heuristics that make us smart (Gigerenzer et al. 1999), we explore fast and frugal heuristics--simple rules in the mind's adaptive toolbox for making decisions with realistic mental resources. These heuristics can enable both living organisms and artificial systems to make smart choices quickly and with a minimum of information by exploiting the way that information is structured in particular environments. In this précis, we show how simple building blocks that control information search, stop search, and make decisions can be put together to form classes of heuristics, including: ignorance-based and one-reason decision making for choice, elimination models for categorization, and satisficing heuristics for sequential search. These simple heuristics perform comparably to more complex algorithms, particularly when generalizing to new data--that is, simplicity leads to robustness. We present evidence regarding when people use simple heuristics and describe the challenges to be addressed by this research program.     

Visions of rationality

The classical view that equates rationality with adherence to the laws of probability theory and logic has driven much research on inference. Recently, an increasing number of researchers have begun to espouse a view of rationality that takes account of organisms' adaptive goals, natural environments, and cognitive constraints. We argue that inference is carried out using boundedly rational heuristics, that is, heuristics that allow organisms to reach their goals under conditions of limited time, information, and computational capacity. These heuristics are ecologically rational in that they exploit aspects of both the physical and social environment in order to make adaptive inferences. We review recent work exploring this multifaceted conception of rationality.     

The influence of information redundancy on probabilistic inferences

Information redundancy affects the accuracy of inference strategies. A simulation study illustrates that under high-information redundancy simple heuristics that rely on only the most important information are as accurate as strategies that integrate all available information, whereas under low redundancy integrating information becomes advantageous. Assuming that people exercise adaptive strategy selection, it is predicted that their inferences will more often be captured by simple heuristics that focus on part of the available information insituations ofhigh-information redundancy, especially when information search is costly. This prediction is confirmed in two experiments. The participants' task was to repeatedly infer which of two alternatives, described by several cues, had a higher criterion value. In the first experiment, simple heuristics predicted the inference process better under high-information redundancy than under low-information redundancy. In the second experiment, this result could be generalized to an inference situation in which participants had no prior opportunity to learn about the strategies' accuracies through outcome feedback. The results demonstrate that people are able to respond adaptively to different decision environments under various learning opportunities.     

Simple mechanisms for gathering social information

Social contexts are notoriously complex, yet decisions are nevertheless made by using simple strategies. We argue that the concept of fast and frugal heuristics provides a promising framework for understanding how we gather social information to make decisions in social environments. That is, we assume that under limitations of time, energy, and computational resources people use cognitively based shortcuts that rely on information from social environments to solve different types of real problems. We review three successful applications of heuristics to the social arena. We first introduce some commonly faced social inference problems (e.g., selecting a mate or deciding whether to cooperate with someone) and then discuss how individuals can use simple strategies to solve such problems. For each problem, we consider how social environments are structured, and how we take advantage of this structure when using heuristics to make inferences and decisions.     

How (far) can rationality be naturalized?

The paper shows why and how an empirical study of fast-and-frugal heuristics can provide norms of good reasoning, and thus how (and how far) rationality can be naturalized. We explain the heuristics that humans often rely on in solving problems, for example, choosing investment strategies or apartments, placing bets in sports, or making library searches. We then show that heuristics can lead to judgments that are as accurate as or even more accurate than strategies that use more information and computation, including optimization methods. A standard way to defend the use of heuristics is by reference to accuracy-effort trade-offs. We take a different route, emphasizing ecological rationality (the relationship between cognitive heuristics and environment), and argue that in uncertain environments, more information and computation are not always better (the ??less-can-be-more?? doctrine). The resulting naturalism about rationality is thus normative because it not only describes what heuristics people use, but also in which specific environments one should rely on a heuristic in order to make better inferences. While we desist from claiming that the scope of ecological rationality is unlimited, we think it is of wide practical use.     

Re-visions of rationality?

The appeal of simple algorithms that take account of both the constraints of human cognitive capacity and the structure of environments has been an enduring theme in cognitive science. A novel version of such a boundedly rational perspective views the mind as containing an 'adaptive toolbox' of specialized cognitive heuristics suited to different problems. Although intuitively appealing, when this version was proposed, empirical evidence for the use of such heuristics was scant. I argue that in the light of empirical studies carried out since then, it is time this 'vision of rationality' was revised. An alternative view based on integrative models rather than collections of heuristics is proposed.     

Decision making with the "adaptive toolbox": influence of environmental structure,intelligence, and working memory load

In 2 experiments with a total of 220 participants, the tendency to use simple heuristics such as the take the best heuristic in an adaptive manner was investigated. In a simulated stock market paradigm, the payoff structure of environments was varied, favoring either compensatory or noncompensatory decision strategies in terms of expected long-term payoff. In both experiments, the majority of participants were classified as using strategies that were adequate for the environment, supporting the notion of adaptive strategy selection. These strategy shifts were moderated by intelligence, as measured with common tests. Neither an additional learning phase (Experiment 1) nor working memory load or working memory capacity (Experiment 2) had additional effects on strategy selection.     

Contingency, causation, and adaptive inference

In contingency judgment tasks involving 2 event types, individuals weight the a and b cells of a 2 x 2 contingency table more than the c and d cells. Some theorists have argued that they can provide normative justifications for this weighting and that the weighting reflects simple heuristics that are adaptive in the real world. The authors show that, to avoid error, individual judgments about real contingencies should be more subtle than these supposedly adaptive heuristics allow.     

Models of ecological rationality: the recognition heuristic

One view of heuristics is that they are imperfect versions of optimal statistical procedures considered too complicated for ordinary minds to carry out. In contrast, the authors consider heuristics to be adaptive strategies that evolved in tandem with fundamental psychological mechanisms. The recognition heuristic, arguably the most frugal of all heuristics, makes inferences from patterns of missing knowledge. This heuristic exploits a fundamental adaptation of many organisms: the vast, sensitive, and reliable capacity for recognition. The authors specify the conditions under which the recognition heuristic is successful and when it leads to the counterintuitive less-is-more effect in which less knowledge is better than more for making accurate inferences.     

When Cognitive Diagnosis Meets Computerized Adaptive Testing: CD-CAT

Computerized adaptive testing (CAT) is a mode of testing which enables more efficient and accurate recovery of one or more latent traits. Traditionally, CAT is built upon Item Response Theory (IRT) models that assume unidimensionality. However, the problem of how to build CAT upon latent class models (LCM) has not been investigated until recently, when Tatsuoka (J. R. Stat. Soc., Ser. C, Appl. Stat. 51:337–350, 2002) and Tatsuoka and Ferguson (J. R. Stat., Ser. B 65:143–157, 2003) established a general theorem on the asymptotically optimal sequential selection of experiments to classify finite, partially ordered sets. Xu, Chang, and Douglas (Paper presented at the annual meeting of National Council on Measurement in Education, Montreal, Canada, 2003) then tested two heuristics in a simulation study based on Tatsuoka’s theoretical work in the context of computerized adaptive testing. One of the heuristics was developed based on Kullback–Leibler information, and the other based on Shannon entropy. In this paper, we showcase the application of the optimal sequential selection methodology in item selection of CAT that is built upon cognitive diagnostic models. Two new heuristics are proposed, and are compared against the randomized item selection method and the two heuristics investigated in Xu et al. (Paper presented at the annual meeting of National Council on Measurement in Education, Montreal, Canada, 2003). Finally, we show the connection between the Kullback–Leibler-information-based approaches and the Shannon-entropy-based approach, as well as the connection between algorithms built upon LCM and those built upon IRT models.     

Verbal cues flexibly transform spatial representations in human memory

ABSTRACT

Humans possess a unique ability to communicate spatially-relevant information, yet the intersection between language and navigation remains largely unexplored. One possibility is that verbal cues accentuate heuristics useful for coding spatial layouts, yet this idea remains largely untested. We test the idea that verbal cues flexibly accentuate the coding of heuristics to remember spatial layouts via spatial boundaries or landmarks. The alternative hypothesis instead conceives of encoding during navigation as a step-wise process involving binding lower-level features, and thus subsequently formed spatial representations should not be modified by verbal cues. Across three experiments, we found that verbal cues significantly affected pointing error patterns at axes that were aligned with the verbally cued heuristic, suggesting that verbal cues influenced the heuristics employed to remember object positions. Further analyses suggested evidence for a hybrid model, in which boundaries were encoded more obligatorily than landmarks, but both were accessed flexibly with verbal instruction. These findings could not be accounted for by a tendency to spend more time facing the instructed component during navigation, ruling out an attentional-encoding mechanism. Our findings argue that verbal cues influence the heuristics employed to code environments, suggesting a mechanism for how humans use language to communicate navigationally-relevant information.     

From embodied to socially embedded agents – Implications for interaction-aware robots

First, this article proposes a minimal definition of embodiment that can be applied across animals and artefacts. We discuss the potential contributions of this operational definition with respect to assessing and measuring the degree of embodiment in different biological and artificial systems. Second, we outline how this definition can be extended to lead to the particular notion of social embeddedness. Socially embedded agents are structurally coupled with their social environment, in that their sensorimotor activity is grounded in the social environment that the agent is surrounded by. Lastly, based on research in the social sciences on human–human interaction, we discuss perceptual requirements for interaction-aware robotic agents—agents whose identification and interpretation of the (social) environment is facilitated by awareness of the structure of agent–agent interactions (including humans ‘in the loop’). We suggest relevant concepts and heuristics that can contribute to studies of degrees of embodiment of robots that interact with social environments. Manipulating and systematically investigating these heuristics permits variation of the degree of embodiment of such interaction-aware robots.     

Coherence shifts in probabilistic inference tasks

The fast‐and‐frugal heuristics approach to probabilistic inference assumes that individuals often employ simple heuristics to integrate cue information that commonly function in a non‐reciprocal fashion. Specifically, the subjective validity of a certain cue remains stable during the application of a heuristic and is not changed by the presence or absence of another cue. The parallel constraint satisfaction (PCS) model, in contrast, predicts that information is processed in a reciprocal fashion. Specifically, it assumes that subjective cue validities interactively affect each other and are modified to coherently support the favored choice. Corresponding to the model's simulation, we predicted the direction and the size of such coherence shifts. Cue validities were measured before, after (Experiment 1), and during judgment (Experiments 2 and 3). Coherence shifts were found in environments involving real‐world cue knowledge (weather forecasts) and in a domain for which the application of fast‐and‐frugal heuristics has been demonstrated (city‐size tasks). The results indicate that subjective cue validities are not fixed parameters, but that they are interactively changed to form coherent representations of the task. Copyright © 2009 John Wiley & Sons, Ltd.     

Heuristic and linear models of judgment: matching rules and environments

Much research has highlighted incoherent implications of judgmental heuristics, yet other findings have demonstrated high correspondence between predictions and outcomes. At the same time, judgment has been well modeled in the form of as if linear models. Accepting the probabilistic nature of the environment, the authors use statistical tools to model how the performance of heuristic rules varies as a function of environmental characteristics. They further characterize the human use of linear models by exploring effects of different levels of cognitive ability. They illustrate with both theoretical analyses and simulations. Results are linked to the empirical literature by a meta-analysis of lens model studies. Using the same tasks, the authors estimate the performance of both heuristics and humans where the latter are assumed to use linear models. Their results emphasize that judgmental accuracy depends on matching characteristics of rules and environments and highlight the trade-off between using linear models and heuristics. Whereas the former can be cognitively demanding, the latter are simple to implement. However, heuristics require knowledge to indicate when they should be used.     

Perceptual constraints and the learnability of simple grammars

Cognitive processes are often attributed to statistical or symbolic general-purpose mechanisms. Here we show that some spontaneous generalizations are driven by specialized, highly constrained symbolic operations. We explore how two types of artificial grammars are acquired, one based on repetitions and the other on characteristic relations between tones ("ordinal" grammars). Whereas participants readily acquire repetition-based grammars, displaying early electrophysiological responses to grammar violations, they perform poorly with ordinal grammars, displaying no such electrophysiological responses. This outcome is problematic for both general symbolic and statistical models, which predict that both types of grammars should be processed equally easily. This suggests that some simple grammars are acquired using perceptual primitives rather than general-purpose mechanisms; such primitives may be elements of a "toolbox" of specialized computational heuristics, which may ultimately allow constructing a psychological theory of symbol manipulation.     

Action-Perception Patterns Emerge From Coupling and Adaptation

Organisms function in particular environments, the properties of which are reflected in the structure of their nervous systems. Therefore, sensory information may directly specify motor behaviors. We argue that such specification involves the coupling of sensory information into appropriately structured control systems that generate action. The nature of this coupling as well as the structure of the control systems reflect properties of the environment. This is most dramatically demonstrated when adaptive processes adjust the underlying control system in response to changes in the environment. Experimental and modeling work on posture in perturbed visual and haptic environments is reviewed to provide evidence for these arguments. Theoretical modeling and autonomous robotics work that goes beyond the posture example of perception-action coupling is briefly discussed, primarily to point out that the integration of multiple behavioral constraints is a nontrivial problem that has received too little attention.