The Gap Between Descriptive And Empirical Decision-making Based On An Information Processing Perspective

In the traditional research paradigm of risky decision making,decision makers directly obtain explicit probability and outcome information about decision options and make choices.However,in some real-life situations,people do not have access to explicit probability and outcome information about decision options,and need to make choices based on their own experience.Researchers refer to decisions that provide explicit probability and outcome information as decision from description.The decision maker does not know the outcome and probability of each option,and needs to repeat the choice(information sampling)and feedback to obtain information about the distribution of the benefits of the options,and then make a decision in the form of decision-making is called the decision from experience(decision from experience).It has been found that there is a difference between descriptive and empirical decision making(referred to as the descriptive-empirical gap,D-E gap),which means that the same decision problem communicated to the decision maker in both descriptive and empirical ways will result in individuals exhibiting different behaviors:overestimation of small probability events in descriptive decision-making tasks,and underestimation of small probability events in empirical decision-making.However,the robustness of the description-experience gap is controversial due to the presence of sampling bias.Extreme values can also affect the robustness of the description-experience gap.Sampling bias refers to the fact that the skewed distribution of small probability events and the fact that people make decisions based on small sample sizes(5-10 samples for each option)can lead individuals to make decisions in which they do not encounter small probability events or are less likely to encounter small probability events.In addition,current research on empirical decision making has focused on its comparison with descriptive decision outcomes or classical effects,and the information processing mechanism of the description-empirical gap is unclear.Understanding the empirical-descriptive gap should compare the two types of decisions in terms of the entire information processing process,not just limited to comparing the two decision outcomes.Previous studies have briefly summarized the cognitive framework of decision making as a set of information processing processes such as information acquisition,memory,representation,and processing modalities.Empirical decision making focuses on information acquisition,and the explanation of the D-E gap must take into account the decision maker’s information-seeking pattern of small-sample sampling,which leads to sampling bias that occurs at the information acquisition stage.Based on the view that short-term memory capacity is limited,memory capacity limitation is also an important reason for the D-E gap.Memory affects how or how much to sample into the cognitive process,and information bias may occur in the process of transferring information from the sampling level to the cognitive level.Representations based on information in different formats affect how probabilistic information is processed and may contribute to the D-E gap.The use of heuristic and analytic strategies can lead to different decision outcomes.The characteristics of descriptive and empirical decision making in terms of information processing processes need to be further explored.The purpose of this paper is to test the robustness and boundary conditions of the description-experience gap after excluding sampling error,and to explore the mechanisms by which the description-experience gap occurs at different stages of the information processing process,based on the cognitive framework for decision making proposed by previous authors.This paper conducts the following three studies:Study 1 tests the robustness and bounds of the description-experience gap.Experiment 1 uses six classical decision problems that represent the description-experience gap and eliminates sampling bias at the information-gathering level by setting the number of samples to 40 and ensuring that the empirical probabilities are exactly equal to the descriptive probabilities.A one-factor design was used.The dependent variables were the subjects’ choice rate of item H(the option with the higher expectation)in both decisions and the rate of attention to small probability events.Experiment 2,based on Experiment 1,was set to have a sample size of 10,close to that used in natural sampling and consistent with memory constraints.The extremes of decision outcomes were also manipulated to examine the description-experience gap in the presence and absence of extremes.A 2(extremes:extremes in options,no extremes in options;within-subjects)x 2(decision type:descriptive,empirical decisions;between-subjects)mixed design was used,with dependent variables:H-item selection rate and small probability event detection rate.Study 1 showed that the description-experience gap is a robust systematic difference rather than sampling error:controlling for sampling error,the description-experience gap remained,and eye movement data showed that the rate of looking at small probability events was higher in descriptive than in empirical decision making(Experiment 1,Experiment 2);extreme values affected the description-experience gap,and the description-experience gap was larger in the absence of extreme values.Extreme values received more attention from decision makers in both empirical and descriptive decisions(Experiment 2).Study 2 examined whether there are differences in individuals’ representations of probability in descriptive and empirical decision making during the mental representation stage.In Study 2,two experiments were conducted to examine the format of individuals’ mental representations when acquiring information in a descriptive or empirical form in a small probability situation(Experiment 3)and a medium-large probability situation(Experiment 4),respectively.Experiment 3 used a 2(decision type:descriptive decision,empirical decision;between-subjects variable)x 2(estimation:percentage estimation,frequency estimation;within-subjects variable)mixed design.Dependent variable:individual correctness of probability estimates.The experimental design for Experiment 3 was the same as for Experiment 4.Study 2 showed that individuals characterize probability in different formats in descriptive and empirical decision making,with individuals tending to characterize probability in the form of percentage in descriptive decision making and frequency in empirical decision making:percentage estimation was as accurate as frequency estimation in descriptive decision making as well as frequency estimation in empirical decision making,whether in small or medium-large probability situations,and frequency estimation was more accurate than percentage estimation in empirical decision making(Experiment 3,Experiment 4).Study 3 examined the predominance of analytic and heuristic strategies in both decision types.Experiment 5 used a 2(decision type:descriptive decision,empirical decision;between-subjects variable)x 2(computational difficulty:high computational difficulty,low computational difficulty;within-subjects variable)mixed design.Dependent variables were H-item selection rate,pupil size,and average gaze duration at the fixation point.Experiment 6 used a 2(decision type:descriptive decision,empirical decision;between-subjects)x 2(difference in expectations between options:large,small;within-subjects)x 2(difference in outcomes:large,small;within-subjects)mixed design.The dependent variables were H-item selection rate,pupil size,and mean gaze duration at the gaze point.The results showed that there were no significant differences in information processing strategies between descriptive and empirical decision making:the computational difficulty of the options did not affect pupil size and mean fixation duration in empirical decision making,nor did it affect these metrics in descriptive decision making(Experiment 5).Differences in expectations and differences in outcomes between options did not affect pupil size and mean gaze duration in empirical decision making,nor did they affect this metric in descriptive decision making(Experiment 6).Individuals did not prefer the complexity of "weighted sums" in either decision,preferring the simplicity and speed of heuristic processing.In summary,the description-experience gap is a robust systematic difference,and extreme values are a boundary condition for the description-experience gap.In terms of mental representations,individuals use different formats to represent information under the two decisions,and both do not favor absolute analytical processing,preferring simple and fast heuristic processing.