Sharpening the echo: An iterative‐resonance model for short‐term recognition memory

We present an iterative‐resonance model for recognition memory. On successive iterations, the probe is compared against a feature‐by‐feature profile of the study set. Yes decisions depend on the similarity of the probe to the profile; No decisions depend on a count of elements in the probe that are not in the profile. Successive iterations sharpen the evidence, and response latency is a function of the number of iterations needed to obtain a sufficiently clear result. The model successfully simulates classic data as well as recent data problematic for alternate models.     

Decision making and confidence given uncertain advice

We study human decision making in a simple forced-choice task that manipulates the frequency and accuracy of available information. Empirically, we find that people make decisions consistent with the advice provided, but that their subjective confidence in their decisions shows 2 interesting properties. First, people's confidence does not depend solely on the accuracy of the advice. Rather, confidence seems to be influenced by both the frequency and accuracy of the advice. Second, people are less confident in their guessed decisions when they have to make relatively more of them. Theoretically, we develop and evaluate a type of sequential sampling process model—known as a self-regulating accumulator—that accounts for both decision making and confidence. The model captures the regularities in people's behavior with interpretable parameter values, and we show its ability to fit the data is not due to excessive model complexity. Using the model, we draw conclusions about some properties of human reasoning under uncertainty.     

Modeling Response Times for Two-Choice Decisions

The diffusion model for two-choice real-time decisions is applied to four psychophysical tasks. The model reveals how stimulus information guides decisions and shows how the information is processed through time to yield sometimes correct and sometimes incorrect decisions. Rapid two-choice decisions yield multiple empirical measures: response times for correct and error responses, the probabilities of correct and error responses, and a variety of interactions between accuracy and response time that depend on instructions and task difficulty. The diffusion model can explain all these aspects of the data for the four experiments we present. The model correctly accounts for error response times, something previous models have failed to do. Variability within the decision process explains how errors are made, and variability across trials correctly predicts when errors are faster than correct responses and when they are slower.     

Should metacognition be measured by logistic regression?

Are logistic regression slopes suitable to quantify metacognitive sensitivity, i.e. the efficiency with which subjective reports differentiate between correct and incorrect task responses? We analytically show that logistic regression slopes are independent from rating criteria in one specific model of metacognition, which assumes (i) that rating decisions are based on sensory evidence generated independently of the sensory evidence used for primary task responses and (ii) that the distributions of evidence are logistic. Given a hierarchical model of metacognition, logistic regression slopes depend on rating criteria. According to all considered models, regression slopes depend on the primary task criterion. A reanalysis of previous data revealed that massive numbers of trials are required to distinguish between hierarchical and independent models with tolerable accuracy. It is argued that researchers who wish to use logistic regression as measure of metacognitive sensitivity need to control the primary task criterion and rating criteria.     

Driven by power? Probe question and presentation format effects on causal judgment

It has been proposed that causal power (defined as the probability with which a candidate cause would produce an effect in the absence of any other background causes) can be intuitively computed from cause-effect covariation information. Estimation of power is assumed to require a special type of counterfactual probe question, worded to remove potential sources of ambiguity. The present study analyzes the adequacy of such questions to evoke normative causal power estimation. The authors report that judgments to counterfactual probes do not conform to causal power and that they strongly depend on both the probe question wording and the way that covariation information is presented. The data are parsimoniously accounted for by an alternative model of causal judgment, the Evidence Integration rule.     

Comparing perceptual and preferential decision making

Perceptual and preferential decision making have been studied largely in isolation. Perceptual decisions are considered to be at a non–deliberative cognitive level and have an outside criterion that defines the quality of decisions. Preferential decisions are considered to be at a higher cognitive level and the quality of decisions depend on the decision maker’s subjective goals. Besides these crucial differences, both types of decisions also have in common that uncertain information about the choice situation has to be processed before a decision can be made. The present work aims to acknowledge the commonalities of both types of decision making to lay bare the crucial differences. For this aim we examine perceptual and preferential decisions with a novel choice paradigm that uses the identical stimulus material for both types of decisions. This paradigm allows us to model the decisions and response times of both types of decisions with the same sequential sampling model, the drift diffusion model. The results illustrate that the different incentive structure in both types of tasks changes people’s behavior so that they process information more efficiently and respond more cautiously in the perceptual as compared to the preferential task. These findings set out a perspective for further integration of perceptual and preferential decision making in a single ramework.     

Tracing memory-structure activation during person judgments

The purpose of the reported research was twofold: (a) to introduce a procedure for measuring concept activation during memory-based decisions and (b) to employ the procedure to investigate memory-activation processes in social judgments. Recent research has focused on subjects' reliance on memory for earlier inferred categorizations as the basis for judgments about persons. However, subjects have also been shown to rely on memory for factual information when making such decisions. To more fully understand how social judgments are made, methods are needed that are capable of tracing concept activation during the judgment process. The present study introduces a procedure that relies on probe recognition speed as a measure of concept activation. The procedure is used to examine alternative models of how subjects activate categorical and event memory when making contemplative impression judgments (i.e., judgments that they have to justify). The results favor a dependent memory-activation model that hypothesizes subjects activate both facts and earlier categorizations that they have made about a person when making subsequent memory-based judgments. Memory-structure activation was dependent in that facts relevant to forming the early categorizations were more likely to be activated in the service of a judgment than category-irrelevant facts. Advantages and limitations of the probe procedure as a measure of memory-structure activation during decisions are discussed.     

More than meets the eye: Context effects in word identification

The influence of semantic context on word identification was examined using masked target displays. Related prime words enhanced a signal detection measure of sensitivity in making lexical decisions and in determining whether a probe word matched the target word. When line drawings were used as primes, a similar benefit was obtained with the probe task. Although these results suggest that contextual information affects perceptual encoding, this conclusion is questioned on the grounds that sensitivity in these tasks may be determined by independent contributions of perceptual and contextual information. The plausibility of this view is supported by a simulation of the experiments using a connectionist model in which perceptual and semantic information make independent contributions to word identification. The model also predicts results with two other analytic methods that have been used to argue for priming effects on perceptual encoding.     

Differential sensitivity of recent vs. remote memory traces to extinction in a water-maze task in rats

Extinction has mostly been studied in conditioning paradigms, more sparsely in spatial tasks, and never as a function of the age of a spatial memory. Using rats, we compared the time-course of extinction of a recent (5 days) vs. remote (25 days) spatial memory in a water maze, over three probe trials. When the trials were set 24h apart, performance in the remote memory group was significantly worse on the first probe trial and significantly better on the third probe trial, as compared to the recent memory group, thereby showing differences between cognitive operations underlying recent vs. remote memory extinction. In contrast, when trials were given consecutively, both groups showed a similar profile of extinction. Furthermore, in a room with overly-salient cues providing a strong remote memory trace, no difference between groups was observed when the spaced extinction paradigm was used. These results might be related to a balance between reconsolidation and extinction processes occurring after a first retrieval experience, of which the outcome may depend on the extinction protocol, and on the age and strength of a memory.     

Spelling-sound effects in reading: Time-course and decision criteria

In these studies, we examined predictions of the time-course model of word recognition (Seidenberg, 1985b; Seidenberg, Waters, Barnes, & Tanenhaus, 1984). The model suggests that effects of irregular spelling or pronunciation should be specific to more slowly recognized words, such as lower frequency items, as shown in previous studies and replicated here. The model also explains why effects of irregular pronunciation are more robust in naming than in lexical decisions: Only the effects in lexical decisions depend on subjects’ response criteria. We show that these criteria are affected by the composition of the stimuli in an experiment (i.e., whether irregularly spelled words are present) and by pressure to respond quickly. In contrast to the dual-route model of word recognition, the time-course model accounts for these task differences without assuming that subjects strategically control access to phonology.     

The structure of short-term memory scanning: an investigation using response time distribution models

A classic question in cognitive psychology concerns the nature of memory search in short-term recognition. Despite its long history of investigation, however, there is still no consensus on whether memory search takes place serially or in parallel or is based on global access. In the present investigation, we formalize a variety of models designed to account for detailed response time distribution data in the classic Sternberg (Science 153: 652-654, 1966) memory-scanning task. The models vary in their mental architectures (serial exhaustive, parallel self-terminating, and global access). Furthermore, the component processes within the architectures that make match/mismatch decisions are formalized as linear ballistic accumulators (LBAs). In fast presentation rate conditions, the parallel and global access models provide far better accounts of the data than does the serial model. LBA drift rates are found to depend almost solely on the lag between study items and test probes, whereas response thresholds change with memory set size. Under slow presentation rate conditions, even simple versions of the serial-exhaustive model provide accounts of the data that are as good as those of the parallel and global access models. We provide alternative interpretations of the results in our General Discussion.     

Hierarchical Bayesian parameter estimation for cumulative prospect theory

Cumulative prospect theory (CPT Tversky & Kahneman, 1992) has provided one of the most influential accounts of how people make decisions under risk. CPT is a formal model with parameters that quantify psychological processes such as loss aversion, subjective values of gains and losses, and subjective probabilities. In practical applications of CPT, the model’s parameters are usually estimated using a single-participant maximum likelihood approach. The present study shows the advantages of an alternative, hierarchical Bayesian parameter estimation procedure. Performance of the procedure is illustrated with a parameter recovery study and application to a real data set. The work reveals that without particular constraints on the parameter space, CPT can produce loss aversion without the parameter that has traditionally been associated with loss aversion. In general, the results illustrate that inferences about people’s decision processes can crucially depend on the method used to estimate model parameters.     

Are individual and group decisions dependent on the available information?

Abstract.— A choice-shift experiment explored the role of available information in individual and group decisions to one "risky" and one "cautious" problem. Individual pre-discussion decisions to both problems by Bristol University students were shown to be related to the relative number of arguments favouring risky and cautious decisions known by each student. This information index was related to group decisions on the cautious problem only. Normative group influence, as well, seems to be involved in group decisions, and it is suggested that the relative importance of informational and normative processes for group decision may depend on the nature of the problem discussed.     

Using confidence ratings to identify a target among foils

Sauer, Brewer, and Weber (2008) advanced a novel procedure for testing eyewitness recognition memory. Rather than providing a single decision (i.e., identifying a lineup member or rejecting the lineup as a whole), participants rated their confidence that each lineup member was the culprit. Classification algorithms determined when patterns of confidence ratings indicated suspect guilt or innocence. Across varied test stimuli, confidence-based classifications equalled or out-performed single decisions. However, Sauer et al.’s classification criteria were designed to optimize performance for the data to which they were applied. If effective classification using confidence ratings requires such idiosyncratic criteria, the applied utility of the confidence procedure is nil. We re-analysed the data from Sauer et al.’s two identification experiments and demonstrated that confidence-based classification performance exceeding that of a traditional lineup task did not depend on uniquely developed classification criteria. Confidence-rating lineups offer a potentially promising alternative to procedures requiring single decisions from witnesses.     

Retrieval plus scanning: Does it occur?

In the two experiments reported here, we tested the retrieval-plus-scan model of delayed probe recognition by adding a second probe 2 sec after subjects had responded to the first probe. According to this model, the list items should still be in consciousness from the first probe at the time of the second probe. Consequently, on tests in which the first probe had been tested immediately, we expected to find the same pattern of performance on both first and second probes. On tests in which the first probe came after a filled delay, we expected the first-probe data to show the effects of retrieval and proactive interference. These effects should not be present on the second probe. The results of the first-probe data in both experiments were consistent with the retrieval-plus-scan model. The second-probe data did not fit the model, however. In the first experiment, the second probe on what had been delayed first-probe trials still produced intercept differences, indicating the presence of retrieval. In the second experiment, the effects of proactive interference were still present on the second probe. It is suggested that these results not only invalidate the retrieval-plus-scan model for delayed probe recognition but that they, along with other results, cast doubt on a scanning operation in immediate probe recognition.     

Diagnosticity,confidence, and the need for information

Recent research has suggested that people prefer to use the most diagnostic available information as the basis for their choices and decisions, and are most confident in those decisions when information is highly diagnostic. However, the effect of information diagnosticity on the need for additional information has yet to be investigated; that is, in an optional stopping task, will the amount of information requested depend upon information diagnosticity? Three models of the role of diagnosticity in information use were examined; expected value, a confidence criterion, and information cost. Subjects attempted to categorize stimuli with the aid of information of varying costs and diagnosticity levels. They requested more information when it was obtained at a low cost. More importantly, across cost conditions, subjects consistently requested greater amounts of information when that information was of a low diagnosticity. These data seem most consistent with use of a confidence criterion that is adjusted for information costs.     

Will commute drivers switch to park-and-ride under the influence of multimodal traveler information? A stated preference investigation

The knowledge about en-trip mode switching behavior with presence of multimodal traveler information is very limited so far. This study investigated the impacts on commute drivers’ en-trip mode switch decisions of smartphone multimodal traveler information systems (SMTIS) which integrate dynamic information of auto-drive and subway park-and-ride (P&R). This is based on data collected from a stated preference survey in Shanghai, China. A panel mixed probit model which accounts for potential correlations of observations among a same driver and heterogeneity in preferences for travel time savings and comfort level of subway car was developed. The panel model has a much better goodness of fit than a model without consideration of panel effect and heterogeneity. The results show that SMTIS have significant impacts on commuter drivers’ decision about switching from auto drive to P&R; the impacts depend on personal attributes including gender, age, education level, income, and P&R use experience; the sensitivity to time savings in the case non-incident induced delays, and the sensitivity to comfort level of subway, both vary significantly among the driver sample.     

On the construction of weak orders from fragmentary information

An iterative method is proposed for constructing a weak order from a partial order on a set of stimuli that is based on individual pairwise comparison data. The method generalizes Duncan Luce's construction of the weak order induced by a semiorder. Various aspects of the iterative procedure are discussed, including its rationale, the number of iterations required to obtain a weak order, and the extent to which the data support additions to the initial partial order as a function of the number of iterations performed before the additions occur.This work was supported in part by a grant from the Alfred P. Sloan Foundation to The Institute for Advanced Study.The computer simulation programs for this study were written and run by William Gehrlein. I am grateful to him for his assistance.     

Profiles of Dialogue for Evaluating Arguments from Ignorance

This investigation uses the technique of the profile of dialogue as a tool for the evaluation of arguments from ignorance (also called lack-of-evidence arguments, negative evidence, ad ignorantiam arguments and ex silentio arguments). Such arguments have traditionally been classified as fallacies by the logic textbooks, but recent research has shown that in many cases they can be used reasonably. A profile of dialogue is a connected sequence of moves and countermoves in a conversational exchange of a type that is goal-directed and can be represented in a normative model of dialogue. Selected case studies are used to probe special features of using the profile technique as applied to arguments from ignorance of a kind that occur frequently in everyday conversational exchanges. One of these special features is the use of Gricean implicature. Another is the need to use negative profiles of argument.     

Towards statistical multicriteria decision modelling: a first approach

Many real life situations result from decisions taken by a very large number of decision makers. Among them, we may cite road traffic congestion, crowding during shopping, equity market behaviour, distribution of holiday destinations, etc. Furthermore, these decisions often depend on the optimisation of several conflicting criteria. In this paper, we introduce a new multicriteria tool based on Markov chains to model and manage these macroscopic phenomena. Finally, the road traffic congestion problem will be considered to illustrate the applicability of our approach. Copyright © 2003 John Wiley & Sons, Ltd.