User:D.barkoczi/sandbox

An organisms repertoire of specialized cognitive mechanisms including fast and frugal heuristics is referred to as the adaptive toolbox. The basic idea of the adaptive toolbox is that different domains of thought require different specialized tools instead of one universal tool - just as a mechanic can use different specialized tools for different tasks. These specialized tools (heuristics) are effective when they exploit the structure of the information in the environment, that is when they are ecologically rational. They can handle situations of uncertainty involving limited time, computational resources and information. The content of the adaptive toolbox is shaped by evolution, learning, and culture for specific domains of inference and reasoning and changes across the life-span.

The adaptive toolbox includes:


 * 1) a specific group of rules or heuristics rather than a general-purpose decision-making algorithm. These heuristics are fast, frugal, and computationally cheap rather than consistent, coherent, and general . Classes and examples of heuristics that are likely to be in the adaptive toolbox of humans and some other animal species include:
 * 2) recognition-based heuristics: Examples:  Recognition heuristic, fluency heuristic
 * 3) one-reason decision-making: Examples: Take-the-best, Fast and Frugal Trees
 * 4) trade-off heuristics Examples: 1/N, Tallying
 * 5) satisficing heuristics
 * 6) social heuristics: Examples: tit for tat, imitate-the-majority, imitate-the-successful, default heuristic, social circle heuristic, averaging, choosing
 * 7) the collection of building blocks (search rules, stopping rules, decision rules) for constructing heuristics,
 * 8) core mental capacities that building blocks exploit (e.g. recognition memory, depth perception, frequency monitoring, object tracking, ability to imitate)

The extent to which humans and other species share heuristics depends on whether they face the same adaptive problems, environmental structures, and share core capacities. For example, “while the absence of language production from the adaptive toolbox of other animals means they cannot use name recognition to make inferences about their world, some animal species can use other capacities such as taste and smell recognition as input for the recognition heuristic”.

How are heuristics selected for a given problem?
The assumption that individuals are equipped with a repertoire of heuristics raises the question how they select strategies in a given context. Scholars have proposed two main mechanisms to explain how individuals select strategies from the adaptive toolbox: the cognitive niches approach and strategy selection learning theory.

According to the idea of cognitive niches, the applicability of specific heuristics is limited by the interplay between environmental structure, cognitive capacity, and strategy. Given the specific characteristics of the environment and the cognitive capacity of the decision maker, only a subset of all heuristics in the repertoire can be used, resulting in so called cognitive niches for different heuristics. Take for example the case of choosing a decision rule for purchasing a mobile phone: if a consumer only knows one of the available brands, she might use the recognition heuristic to make a purchasing decision and choose the brand she knows; however, if the consumer does not recognize any of the brands, she can either study all the features to compare two models and use tallying or in lack of time and cognitive capacity just make use of a few important features to compare the two models using take-the-best.

The strategy selection learning theory, in contrast, argues that people select appropriate strategies based on learning. It assumes that individuals form subjective expectations for the strategies they have, select strategies proportionally to these expectations and update their expectations after the use of the selected strategy. In the case of selecting a strategy for choosing among different options during the purchase of a mobile phone, strategy selection learning theory proposes that individuals first assess how good each of the available strategies would perform in terms of making the right decision. Then, based on this judgment, they would apply the decision rule with the highest expected outcome. After using a particular decision rule to make the purchase of the mobile phone, the choice outcome is evaluated and expectations about the performance of that executed decision rule are updated and reinforced to inform future purchase decisions.

Alternative views
The concept of the adaptive toolbox departs from the view that there is a single strategy that is universally superior as put forward by Gottfried Wilhelm Leibniz. Leibniz proposed to replace all reasoning with a universal logical language, the Universal Characteristic. "The multitude of simple concepts constituting Leibniz’s alphabet of human thought were all to be operated on by a single general-purpose tool such as probability theory” Today, a number of approaches exist that assume a universal strategy: for example rational choice theory, the Bayesian approach to cognition,  Parallel constraint satisfaction processes (PCS) , sequential-sampling process models such as the adaptive spanner perspective and decision field theory.