A new model for confidence judgments in recognition memory is presented. In the model, the match between a single test item and memory produces a distribution of evidence, with better matches corresponding to distributions with higher means. On this match dimension, confidence criteria are placed, and the areas between the criteria under the distribution are used as drift rates to drive racing Ornstein-Uhlenbeck diffusion processes. The model is fit to confidence judgments and quantile response times from two recognition memory experiments that manipulated word frequency and speed versus accuracy emphasis. The model and data show that the standard signal detection interpretation of z-transformed receiver operating characteristic (z-ROC) functions is wrong. The model also explains sequential effects in which the slope of the z-ROC function changes by about 10% as a function of the prior response in the test list.

We evaluated age-related differences in the optimality of decision boundary settings in a diffusion model analysis. In the model, the width of the decision boundary represents the amount of evidence that must accumulate in favor of a response alternative before a decision is made. Wide boundaries lead to slow but accurate responding, and narrow boundaries lead to fast but inaccurate responding. There is a single value of boundary separation that produces the most correct answers in a given period of time, and we will refer to this value as the reward rate optimal boundary (RROB). Across a variety of decision tasks, we consistently found that older adults used boundaries that were much wider than the RROB value. Young adults used boundaries that were closer to the RROB value, although age differences in optimality were smaller with instructions emphasizing speed than with instructions emphasizing accuracy. Young adults adjusted their boundary settings to more closely approach the RROB value when they were provided with accuracy feedback and extensive practice. Older participants showed no evidence of making boundary adjustments in response to feedback or task practice, and they consistently used boundary separation values that produced accuracy levels that were near asymptote. Our results suggest that young participants attempt to balance speed and accuracy to achieve the most correct answers per unit time, whereas older participants attempt to minimize errors even if they must respond quite slowly to do so.To make good decisions, one must gather an appropriate amount of information before selecting an alternative. Gathering more information leads to more accurate but slower decisions, a situation referred to as the speed-accuracy tradeoff. Decision makers are faced with the dilemma of adopting a level of conservativeness that appropriately balances the speed and accuracy of their responding.Prior research shows that young and older adults differ in terms of how they balance speed and accuracy. Older adults make decisions slowly and avoid errors, whereas young adults decide more quickly and are more accepting of errors (Baron & Matilla, 1989;Hertzog, Vernon, & Rypma, 1993;Rabbitt, 1979;Salthouse, 1979;Smith & Brewer, 1985, 1995. Perhaps older adults are influenced by a lifetime of experience in which arriving at the correct answer is well worth the necessary deliberation. In this article, we explore the difference between young and older adults by examining the optimality of boundary settings using the diffusion model. As we explain in detail below, the diffusion model is a model of two-choice decision tasks that makes choices based on evidence accumulated over time (Ratcliff, 1978;Ratcliff & McKoon, 2008). One advantage of applying the model is that it accounts for speed-accuracy tradeoffs in terms of a single parameter representing response conservativeness: boundary separation. Boundary separation determines the amount of evidence that must accumulate in favor of a response alternative before a choice ...

Most data analyses rely on models. To complement statistical models, psychologists have developed cognitive models, which translate observed variables into psychologically interesting constructs. Response time models, in particular, assume that response time and accuracy are the observed expression of latent variables including 1) ease of processing, 2) response caution, 3) response bias, and 4) non-decision time. Inferences about these psychological factors hinge upon the validity of the models' parameters. Here, we use a blinded, collaborative approach to assess the validity of such model-based inferences. Seventeen teams of researchers analyzed the same 14 data sets. In each of these two-condition data sets, we manipulated properties of participants' behavior in a two-alternative forced choice task. The contributing teams were blind to the manipulations, and had to infer what aspect of behavior was changed using their method of choice. The contributors chose to employ a variety of models, estimation methods, and inference procedures. Our results show that, although conclusions were similar across different methods, these "modeler's degrees of freedom" did affect their inferences. Interestingly, many of the simpler approaches yielded as robust and accurate inferences as the more complex methods. We recommend that, in general, cognitive models become a typical analysis tool for response time data. In particular, we argue that the simpler models and procedures are sufficient for standard experimental designs. We finish by outlining situations in which more complicated models and methods may be necessary, and discuss potential pitfalls when interpreting the output from response time models.

Confidence in judgments is a fundamental aspect of decision making, and tasks that collect confidence judgments are an instantiation of multiple-choice decision making. We present a model for confidence judgments in recognition memory tasks that uses a multiple-choice diffusion decision process with separate accumulators of evidence for the different confidence choices. The accumulator that first reaches its decision boundary determines which choice is made. Five algorithms for accumulating evidence were compared, and one of them produced proportions of responses for each of the choices and full response time distributions for each choice that closely matched empirical data. With this algorithm, an increase in the evidence in one accumulator is accompanied by a decrease in the others so that the total amount of evidence in the system is constant. Application of the model to the data from an earlier experiment (Ratcliff, McKoon, & Tindall, 1994) uncovered a relationship between the shapes of z-transformed receiver operating characteristics and the behavior of response time distributions. Both are explained in the model by the behavior of the decision boundaries. For generality, we also applied the decision model to a 3-choice motion discrimination task and found it accounted for data better than a competing class of models. The confidence model presents a coherent account of confidence judgments and response time that cannot be explained with currently popular signal detection theory analyses or dual-process models of recognition.

We tested two explanations for why the slope of the z-transformed receiver operating characteristic (zROC) is less than 1 in recognition memory: the unequal-variance account (target evidence is more variable than lure evidence) or the dual-process account (responding reflects both a continuous familiarity process and a threshold recollection process). These accounts are typically implemented in signal detection models that do not make predictions for response time (RT) data. We tested them using RT data and the diffusion model. Participants completed multiple study/test blocks of an “old”/”new” recognition task with the proportion of targets on the test varying from block to block (.21, .32, .50, .68, or .79 targets). The same participants completed sessions with both speed-emphasis and accuracy-emphasis instructions. zROC slopes were below one for both speed and accuracy sessions, and they were slightly lower for speed. The extremely fast pace of the speed sessions (mean RT = 526) should have severely limited the role of the slower recollection process relative to the fast familiarity process. Thus, the slope results are not consistent with the idea that recollection is responsible for slopes below 1. The diffusion model was able to match the empirical zROC slopes and RT distributions when between-trial variability in memory evidence was greater for targets than for lures, but missed the zROC slopes when target and lure variability were constrained to be equal. Therefore, unequal variability in continuous evidence is supported by RT modeling in addition to signal detection modeling. Finally, we found that a two-choice version of the RTCON model could not accommodate the RT distributions as successfully as the diffusion model.

In two-choice decision tasks, Starns and Ratcliff (Psychology and Aging 25: 377-390, 2010) showed that older adults are farther from the optimal speed-accuracy trade-off than young adults. They suggested that the age effect resulted from differences in task goals, with young participants focused on balancing speed and accuracy and older participants focused on minimizing errors. We compared speed-accuracy criteria with a standard procedure (blocks that had a fixed numbers of trials) to a condition in which blocks lasted a fixed amount of time and participants were instructed to get as many correct responses as possible within the time limit-a goal that explicitly required balancing speed and accuracy. Fits of the diffusion model showed that criteria differences persisted in the fixed-time condition, suggesting that age differences are not solely based on differences in task goals. Also, both groups produced more conservative criteria in difficult conditions when it would have been optimal to be more liberal.

Using the retrieval-practice paradigm (Anderson, R. A. Bjork, & E. L. Bjork, 1994), we tested whether or not retrieval-induced forgetting could be found in item recognition tests. In Experiment 1, retrieval practice on items from semantic categories depressed recognition of nonpracticed items from the same categories. Similar results were found in Experiment 2 in a more stringent source test for practiced, nonpracticed, and new items. These results conceptually replicate those of previous retrieval-induced forgetting studies done with cued recall (e.g., Anderson et al., 1994). Our findings are inconsistent with the hypothesis that item-specific cues during retrieval will eliminate retrieval interference in the retrieval-practice paradigm (Butler, Williams, Zacks, & Maki, 2001). We discuss our results in relation to other retrieval interference and inhibition effects in recall and recognition.

We report three experiments investigating source memory for words that were called "new" on a recognition test. In each experiment, participants could accurately specify the source of words that they failed to recognize. Results also demonstrated that source memory for unrecognized items varied with the bias to respond "old" in recognition decisions: Participants displayed unrecognized source memory when they were told that 25% of the recognition test words were old (promoting conservative responding) but not when they were told that 75% of the test words were old (promoting liberal responding). Our results were successfully predicted by a multivariate signal detection approach to recognition/source memory.