Research on judgment under uncertainty has revealed the importance of heuristic strategies that people adopt for making inferences rapidly. One such heuristic is representativeness: The perceiver decides how likely it is that an object is a member of some category on the basis of whether the object’s features resemble the essential features of the category. Another heuristic is availability, which estimates frequency or likelihood by how quickly instances or associations come to mind. The simulation heuristic uses mental rehearsal of events to determine what outcome is likely, given a set of circumstances. The anchoring heuristic enables the perceiver to use an already-existing reference point and adjust from that reference point to reach an estimate to answer some new problem. Decision framing affects inferences, especially whether a decision is framed in terms of gains or losses.
Other shortcuts go beyond these heuristics. People often overlook good base-rate information describing population characteristics, in favor of less reliable, but seemingly more relevant, case history or anecdote. People are particularly poor at using and combining probabilistic information, often manifesting great confidence in the truth of unlikely events. When pre-existing expectations or theories guide social inference, a positive test case strategy is often adopted that selectively looks for instances that test the expected inference.
Some sources of error affect our ability to combine information into a judgment. Covariation, the estimate of the degree of association between two events, concatenates many of the previously described skills (e.g., detecting relevancy, sampling, combining information). When data are already collected and clearly summarized, instructions are clear, relevance to the statistical model is clear, and no a priori theory about degree of covariation exists, the social perceiver does fairly well at estimating covariation. However, in the absence of these factors, covariation estimates are heavily biased by prior expectations.
Temporal factors clearly guide the inference process, such as the fact that present rewards are typically valued more highly than larger, more distant rewards. In addition, abstract representative features govern decisions made for the future, whereas consideration of concrete details plays a stronger role in judgments about the present. In learning from past behavior or errors, people usually overestimate what could have been foreseen.