Optimality in Choice

Forensic lineups are used to test a witness’ memory for the perpetrator of a crime. While they often take the form of a photo array presented simultaneously to the witness, in many jurisdictions, the lineup items are presented sequentially. Although decision rules vary, in its simplest form, the witness makes a decision concerning the identity of each item in the sequence before being presented with the next. Of both applied and theoretical interest is whether the location of the perpetrator (or target) in the lineup affects the probability of correct identification. To answer this question, it is necessary to develop, test, and evaluate a model of the sequential lineup task. We outline this model and apply it to data we recently collected as well as data reported by Wilson, Donnelly, Christenfeld, and Wixted (2019, Journal of Memory and Language, 104, 108-125). The two sets of data reveal similar results. There are little or no target position effects on discriminability but substantial effects on decision criteria including a short-lived increase following a failure to detect the target. We discuss the implications of these results for the interpretation of sequential lineup identifications.

Human decisions often deviate from economic rationality and are influenced by cognitive biases. One such bias is thememory bias according to which people prefer choice options they have a better memory of—even when the options’ utilities arecomparatively low. Although this phenomenon is well supported empirically, its cognitive foundation remains elusive. Here we test two conceivable computational accounts of the memory bias against each other. On the one hand, a single-process account explains the memory bias by assuming a single biased evidence-accumulation process in favor of remembered options. On the contrary, a dual-process account posits that some decisions are driven by a purely memory-driven process and others by a utility-maximizing one. We show that both accounts are indistinguishable based on choices alone as they make similar predictions with respect to the memory bias. However, they make qualitatively different predictions about response times. We tested the qualitative and quantitative predictions of both accounts on behavioral data from a memory-based decision-making task. Our results show that a single-process account provides a better account of the data, both qualitatively and quantitatively. In addition to deepening our understanding of memory-based decision making, our study provides an example of how to rigorously compare single- versus dual-process models using empirical data and hierarchical Bayesian estimation methods.

A central question in the psychology of reasoning is whether people use similar functions and processes to make deductive and inductive inferences, or whether deduction and induction are fundamentally distinct forms of thinking. Dual process theories distinguish two distinct types of thinking: Type 1 is said to be intuitive, heuristic and independent of working memory, and Type 2 deliberative, analytic, and reliant on working memory. Type 2 thinking is often considered uniquely qualified to compute the functions that define deduction. In contrast, single process theories argue that the dual process distinction is unnecessary, and point to Bayesian probabilistic logic as a shared basis for deduction and induction. How can these approaches be differentiated empirically? Much research in this area has relied on verbal theories and the measurement of dissociations in experiments. But these provide only limited information. More recent work has begun to formalise the theories e.g. in a signal detection theory framework, to compare them using more advanced modelling techniques. We present four signal detection theory models of reasoning within a Bayesian framework: two single process models, one with independent and the other with dependent decision criteria for deduction and induction, and two dual process models, again one with independent and the other with dependent decision criteria. We then assess the descriptive adequacy of the models across three different distributional assumptions: normal, binomial and beta, following the idea of a multiverse or sensitivity analysis. We discuss the implications of our results for the single-dual process theory debate.

Sequential sampling models describe the cognitive mechanisms at play in preferential decision making. These models can predict attention, choice, and response time in simple choices, but are currently unable to specify how decision-makers deliberate in more complex settings, such as those involving multi-branch gambles. To make rational decisions for such gambles, decision-makers need to compute interactions between payoffs and probabilities. Current sequential sampling models, which model information sampling as sequentially independent, are unable to capture within-branch attribute interactions, and thus make absurd predictions for multi-branch gambles. This is analogous to the feature binding problem in cognitive science, which involves the integration of the perceptual properties of objects. In this paper, we propose a solution to the feature binding problem for risky choice. Specifically, we propose that attribute sampling in multi-branch gambles is sequentially non-independent and that decision-makers are more likely to sample the probability of a branch if they observe a highly desirable payoff in that branch (and vice versa). We show that such a non-independent sampling process allows sequential sampling models to make utility-maximizing predictions. We test our model on data from four existing Mouselab and eye-tracking experiments, and two novel Mouselab experiments, and find that most participants display the non-independent attribute sampling proposed by our model. Additionally, we show that participants who display stronger non-independent sampling are also less likely to deviate from expected value/utility maximization. Overall, our results show how feature binding implemented in existing sequential sampling models can be used to predict sophisticated risky choice behavior.

Trait impulsivity—defined by strong preference for immediate over delayed rewards and difficulties inhibiting prepotent behaviors—is observed in all externalizing disorders, including substance use disorders. Many laboratory tasks have been developed to identify cognitive mechanisms and correlates of impulsive behavior, but convergence between task measures and self-reports of impulsivity are consistently low. Longstanding theories of personality and decision-making predict that neurally mediated individual differences in sensitivity to reward cues versus punishment cues (frustrative non-reward) interact to affect behavioral tendencies. Such interactions obscure 1:1 correspondences between single personality traits and task performance. Here, we develop models within the framework of decision theory that provide an explanation for how impulsive and anxious valuation may interact at the mechanistic level to produce observed interactions at the level of behavior. We then use hierarchical Bayesian analysis to fit these models to three samples with differing levels of substance use, trait impulsivity, and state anxiety (total N=967). Our findings: (1) reveal cognitive mechanisms through which anxiety can modulate impulsive valuation and subsequently attenuate impulsive behavior, and (2) demonstrate benefits of decision theory and hierarchical Bayesian analysis over traditional approaches for testing theories of psychopathology spanning levels of analysis.