Judging Cumulative Risk1

The accuracy of young adults' perceptions of how cumulative risks to life, health, and property change over time was tested by asking subjects to judge the long-term probabilities associated with different periods of exposure to risks that are very small in the short term. Process analyses revealed evidence that strategy choice and associated accuracy depended on context and framing variables. Subjects asked to judge conjunctive probabilities adopted a variety of strategies, all of which failed to yield consistently accurate long-term probability judgments. However, subjects asked to judge disjunctive probabilities often reframed the probability questions into questions about the expected number of times the hazard would strike, which they could answer more accurately. Implications of the research for promoting public understanding of the long-term implications of cumulative risks are discussed.     

Probability judgments of agency: Rational or irrational?

We studied how people attribute action outcomes to their own actions under conditions of uncertainty. Participants chose between left and right keypresses to produce an action effect (a corresponding left or right light), while a computer player made a simultaneous keypress decision. In each trial, a random generator determined which of the players controlled the action effect at varying probabilities, and participants then judged which player had produced it. Participants’ effect control ranged from 20% to 80%, varied blockwise, and they could use trial-by-trial feedback to optimize the accuracy of their agency judgments. Participants tended to attribute action effects to themselves (agency bias), probably reflecting a rational guessing strategy of always naming the more likely player. However, participants systematically neglected information favoring the computer player as the agent, even under conditions where this bias could only harm judgment accuracy. We conclude that agency biases have both rational and irrational components.     

Errors in the quantification of uncertainty: A product of heuristics or minimal probability knowledge base?

The use of intuitive heuristics (e. g. representativeness and availability) has been put forward as an explanation for people's assignment of probabilities (Tversky and Kahneman, 1971). This phenomenon is seen as robust since experts as defined by education (professional psychologists), despite advanced training in statistics and methodology, rely on the same heuristics as novices (lay people). Both experts and novices as defined by education were studied in a series of experiments and further classified as experts and novices according to their probability knowledge base, prior to receiving (or not) a brief (15-minute) training session. Immediately following training, subjects completed a probability test which consisted of ten Tversky and Kahneman (e. g. 1974) problems. The training significantly increased the number of problems correctly solved on the probability test and eliminated the expert/novice education classification. The results of a follow-up test 5 weeks after the experiment indicated that the training group maintained its superior performance. It is proposed that failure to use proper methods of probability assignment may not be due to intrinsic human inference biases or heuristics, but is a result of a minimal probability knowledge base.     

Errors in the quantification of uncertainty: A product of heuristics or minimal probability knowledge base?

The use of intuitive heuristics (e. g. representativeness and availability) has been put forward as an explanation for peoples' assignment of probabilities (Tversky and Kahneman, 1971). This phenomenon is seen as robust since experts as defined by education (professional psychologisis), despite advanced training in statistics and methodology, rely on the same heuristics as novices (lay people). Both experts and novices, as defined by education, were studied in a series of experiments and further classified as experts and novices according to their probability knowledge base, prior to receiving (or not) a brief (15-minute) training session. Immediately following training, subjects completed a probability test which consisted of ten Tversky and Kahneman (e. g. 1974) problems. The training significantly increased the number of problems correctly solved on the probability test and eliminated the expert/novice education classification. The results of a follow-up test 5 weeks after the experiment indicated that the training group maintained its superior performance. It is proposed that failure to use proper methods of probability assignment may not be due to intrinsic human inference biases or heuristics, but is a result of a minimal probability knowledge base.     

Sensitivity to alternatives in social hypothesis-testing

Past research has documented a hypothesis-testing strategy wherein evidence is sought to the extent that it is probable under the hypothesis. This strategy may yield nondiagnostic information and even biased confirmation of the hypothesis if the simultaneous probability of the evidence under the alternatives is disregarded. The results of three experiments demonstrated that hypothesis-testers were in fact sensitive to the probability of the evidence under the alternatives. In the first experiment, subjects tested a hypothesis under which two kinds of personal features, A-features and B-features, were highly probable. Subjects could test their hypothesis by selecting questions from a list of questions about A-features and B-features. The rerults showed that subjects' questions depended on the probability of the features under the alternative. Specifically, when the hypothesis shared A-features with the alternative, subjects preferred questions about B-features, but when the hypothesis shared B-features with the alternative, subjects preferred questions about A-features. Experiment 2 extended these findings to self-generated questions about a broader range of hypotheses and alternatives. Experiment 3 found that subjects who were provided with a specific alternative asked more diagnostic questions than subjects who were not provided with a specific alternative. Together, these results suggest that the process of generating and evaluating alternatives plays a crucial role in social hypothesis-testing and categorization.     

Everybody Will Win,and All Must Be Hired: Comparing Additivity Neglect with the Nonselective Superiority Bias

Two streams of research looking at referent‐dependent judgments from slightly different angles are subadditivity research and research on the nonselective superiority bias. Both biases violate basic formal constraints: the probabilities of a set of exclusive events cannot add up to more than 100%, and a set of attractive candidates cannot all be rated as superior to the group mean. We examine in three experiments how these two biases are related, by asking the same participants to perform both kinds of tasks on the same material. Both biases appear to be widespread, even for sets where all alternatives are presented together, but they differ in the way they are affected by response format and experimental setup. Thus, presenting participants with an unbiased set of ratings will reduce but not normalize their probability estimates of the same alternatives; while presenting them with an unbiased (additive) set of probabilities will make most alternatives appear inferior to the group mean, inverting the superiority bias. Self‐reports reveal that additivity neglect and the nonselective superiority bias can be based on two main response‐strategies: (i) considering each alternative independently or (ii) comparing alternatives, while neglecting their complementarity. In both cases, assessments will be the outcome of a compromise between the perceived “absolute” merits of each alternative, its standing relative to referents, and properties of the response scale. Copyright © 2015 John Wiley & Sons, Ltd.     

Seeking Confirmation Is Rational for Deterministic Hypotheses

The tendency to test outcomes that are predicted by our current theory (the confirmation bias) is one of the best‐known biases of human decision making. We prove that the confirmation bias is an optimal strategy for testing hypotheses when those hypotheses are deterministic, each making a single prediction about the next event in a sequence. Our proof applies for two normative standards commonly used for evaluating hypothesis testing: maximizing expected information gain and maximizing the probability of falsifying the current hypothesis. This analysis rests on two assumptions: (a) that people predict the next event in a sequence in a way that is consistent with Bayesian inference; and (b) when testing hypotheses, people test the hypothesis to which they assign highest posterior probability. We present four behavioral experiments that support these assumptions, showing that a simple Bayesian model can capture people's predictions about numerical sequences (Experiments 1 and 2), and that we can alter the hypotheses that people choose to test by manipulating the prior probability of those hypotheses (Experiments 3 and 4).     

Modeling the effects of payoff on response bias in a perceptual discrimination task: bound-change, drift-rate-change, or two-stage-processing hypothesis

Three hypotheses--the bound-change hypothesis, drift-rate-change hypothesis, and two-stage-processing hypothesis--are proposed to account for data from a perceptual discrimination task in which three different response deadlines were involved and three different payoffs were presented prior to each individual trial. The aim of the present research was to show (1) how the three different hypotheses incorporate response biases into a sequential sampling decision process, (2) how payoffs and deadlines affect choice probabilities, and (3) the hypotheses' predictions of response times and choice probabilities. The two-stage-processing hypothesis gave the best account, especially for the choice probabilities, whereas the drift-rate-change hypothesis had problems predicting choice probabilities as a function of deadlines.     

Information seeking on Bayesian conditional probability problems: A fuzzy-trace theory account

Recently, the ‘heuristics and biases’ approach to the study of decision making has been criticized, with a call for better integrated theory. Three experiments stemming from fuzzy-trace theory addressed information seeking on probability problems, and the cognitive representation of hit-rates, base-rates, and the contrapositive. As predicted by the fuzzy-trace principle of ‘denominator neglect’, many subjects exhibited ‘conversion errors’, confusing the hit-rate, P(A|B), with the answer, P(B|A). These subjects sought base-rates less often than other subjects. On causal problems, more subjects correctly represented base-rates, sought base-rates more often, and produced more accurate estimates than on non-causal problems. Subjects tutored on the meaning of the hit-rate sought the base-rate more often, and were more accurate than control subjects. Results are explained by fuzzy-trace theory principles of gist extraction, fuzzy processing preference, denominator neglect, and output interference.     

An investigation of the relationships between problem solving strategies, representation and memory

There has been considerable debate in recent years about the status of “imagery” in problem solving. The present experiment attempts to show that while subjects may employ representational strategies when they first encounter a class of problems, they abandon such strategies as they gain experience with the problems. It does this by asking subjects to answer unexpected questions which are based upon the information which they have just used to solve a problem. The hypothesis, which is supported by the results, is that increasing experience with problems will be paralleled by a decreasing ability to answer unexpected questions. The experiment also shows that such effects are not attributable to a build-up in proactive interference.     

Doing philosophy is doing its history

While Curley argues that we need to know the history of philosophy so as not to avoid important alternatives to contemporary proposals, I argue that philosophy is an essentially historical enterprise. Unlike science, philosophy cannot forget its history. Not to know the history of philosophy is not to understand why the questions we seek to answer are worth answering or asking.     

Information selection and use in hypothesis testing: What is a good question,and what is a good answer?

The process of hypothesis testing entails both information selection (asking questions) and information use (drawing inferences from the answers to those questions). We demonstrate that although subjects may be sensitive to diagnosticity in choosing which questions to ask, they are insufficiently sensitive to the fact that different answers to the same question can have very different diagnosticities. This can lead subjects to overestimate or underestimate the information in the answers they receive. This phenomenon is demonstrated in two experiments using different kinds of inferences (category membership of individuals and composition of sampled populations). In combination with certain information-gathering tendencies, demonstrated in a third experiment, insensitivity to answer diagnosticity can contribute to a tendency toward preservation of the initial hypothesis. Results such as these illustrate the importance of viewing hypothesis-testing behavior as an interactive, multistage process that includes selecting questions, interpreting data, and drawing inferences.     

Dual Space Search During Scientific Reasoning

The purpose of the two studies reported here was to develop an integrated model of the scientific reasoning process. Subjects were placed in a simulated scientific discovery context by first teaching them how to use an electronic device and then asking them to discover how a hitherto unencountered function worked. To do this task, subjects had to formulate hypotheses based on their prior knowledge, conduct experiments, and evaluate the results of their experiments. In the first study, using 20 adult subjects, we identified two main strategies that subjects used to generate new hypotheses. One strategy was to search memory and the other was to generalize from the results of previous experiments. We described the former group as searching an hypothesis space, and the latter as searching an experiment space. In a second study, with 10 adults, we investigated how subjects search the hypothesis space by instructing them to state all the hypotheses that they could think of prior to conducting any experiments. Following this phase, subjects were then allowed to conduct experiments. Subjects who could not think of the correct rule in the hypothesis generation phase discovered the correct rule only by generalizing from the results of experiments in the experimental phase.
Both studies provide support for the view that scientific reasoning can be characterized as search in two problem spaces. By extending Simon and Lea's (1974) Generalized Rule Inducer, we present a general model of Scientific Discovery as Dual Search (SDDS) that shows how search in two problem spaces (an hypothesis space and an experiment space) shapes hypothesis generation, experimental design, and the evaluation of hypotheses. The model also shows how these processes interact with each other. Finally, we interpret earlier findings about the psychology of scientific reasoning in terms of the SDDS model.     

Unknown probabilities

From a point of view like de Finetti's, what is the judgmental reality underlying the objectivistic claim that a physical magnitude X determines the objective probability that a hypothesis H is true? When you have definite conditional judgmental probabilities for H given the various unknown values of X, a plausible answer is sufficiency, i.e., invariance of those conditional probabilities as your probability distribution over the values of X varies. A different answer, in terms of conditional exchangeability, is offered for use when such definite conditional probabilities are absent.     

Quantifiers and question answering in young children

Two hypotheses were tested about how young children answer questions with the quantifiers all and some: (a) that children use syntactic cues in determining which noun phrase is quantified, and (b) that children evaluate a some-statement as part of evaluating an all-statement. To test these hypotheses, the same group of 60 4- to 7-year-olds were asked four contrasting types of quantitative questions. The results indicated that children can use syntactic cues under some presentation conditions. However, there was no evidence for an asymmetry between the all-and some-questions. A model of how young children might answer quantitative questions was then considered.     

Gambles vs. quasi-realistic scenarios: expectations to find probability and risk-defusing information

The paper investigates predecisional information search in risky decisions, specifically information concerning the probability of a negative outcome and whether a risk-defusing operator (RDO) is available. Experiment 1 (54 participants) tested the hypothesis that search for such information is triggered by expectations that it can be obtained in the situation. Cues for the availability of information were manipulated. It was predicted that cues mentioning possible information sources raise expectations and consequently increase search activity. Furthermore, gambles were expected to differ from other real world contexts, with lower expectations for RDOs and higher ones for probabilities. The Method of Active Information Search was employed. The number of questions asked about probability and RDOs in different conditions confirmed the hypotheses. Experiment 2 (36 participants) ruled out the alternative interpretation that the expectation to actually find favorable probabilities or applicable RDOs respectively, rather than the expectation to obtain information, determined information search.     

Jealousy and the nature of beliefs about infidelity: Tests of competing hypotheses about sex differences in the United States,Korea, and Japan

The different adaptive problems faced by men and women over evolutionary history led evolutionary psychologists to hypothesize and discover sex differences in jealousy as a function of infidelity type. An alternative hypothesis proposes that beliefs about the conditional probabilities of sexual and emotional infidelity account for these sex differences. Four studies tested these hypotheses. Study 1 tested the hypotheses in an American sample (N = 1,122) by rendering the types of infidelity mutually exclusive. Study 2 tested the hypotheses in an American sample (N = 234) by asking participants to identify which aspect of infidelity was more upsetting when both forms occurred, and by using regression to identify the unique contributions of sex and beliefs. Study 3 replicated Study 2 in a Korean sample (N = 190). Study 4 replicated Study 2 in a Japanese sample (N = 316). Across the studies, the evolutionary hypothesis, but not the belief hypothesis, accounted for sex differences in jealousy when the types of infidelity are rendered mutually exclusive; sex differences in which aspect of infidelity is more upsetting when both occur; significant variance attributable to sex, after controlling for beliefs; sex-differentiated patterns of beliefs; and the cross-cultural prevalence of all these sex differences.     

A further test of sequential-sampling models that account for payoff effects on response bias in perceptual decision tasks

Recently, Diederich and Busemeyer (2006) evaluated three hypotheses formulated as particular versions of a sequential-sampling model to account for the effects of payoffs in a perceptual decision task with time constraints. The bound-change hypothesis states that payoffs affect the distance of the starting position of the decision process to each decision bound. The drift-rate-change hypothesis states that payoffs affect the drift rate of the decision process. The two-stage-processing hypothesis assumes two processes, one for processing payoffs and another for processing stimulus information, and that on a given trial, attention switches from one process to the other. The latter hypothesis gave the best account of their data. The present study investigated two questions: (1) Does the experimental setting influence decisions, and consequently affect the fits of the hypotheses? A task was conducted in two experimental settings--either the time limit or the payoff matrix was held constant within a given block of trials, using three different payoff matrices and four different time limits--in order to answer this question. (2) Could it be that participants neglect payoffs on some trials and stimulus information on others? To investigate this idea, a further hypothesis was considered, the mixture-of-processes hypothesis. Like the two-stage-processing hypothesis, it postulates two processes, one for payoffs and another for stimulus information. However, it differs from the previous hypothesis in assuming that on a given trial exactly one of the processes operates, never both. The present design had no effect on choice probability but may have affected choice response times (RTs). Overall, the two-stage-processing hypothesis gave the best account, with respect both to choice probabilities and to observed mean RTs and mean RT patterns within a choice pair.