A Bayesian analysis of human decision-making on bandit problems

The bandit problem is a dynamic decision-making task that is simply described, well-suited to controlled laboratory study, and representative of a broad class of real-world problems. In bandit problems, people must choose between a set of alternatives, each with different unknown reward rates, to maximize the total reward they receive over a fixed number of trials. A key feature of the task is that it challenges people to balance the exploration of unfamiliar choices with the exploitation of familiar ones. We use a Bayesian model of optimal decision-making on the task, in which how people balance exploration with exploitation depends on their assumptions about the distribution of reward rates. We also use Bayesian model selection measures that assess how well people adhere to an optimal decision process, compared to simpler heuristic decision strategies. Using these models, we make inferences about the decision-making of 451 participants who completed a set of bandit problems, and relate various measures of their performance to other psychological variables, including psychometric assessments of cognitive abilities and personality traits. We find clear evidence of individual differences in the way the participants made decisions on the bandit problems, and some interesting correlations with measures of general intelligence.     

Learning the form of causal relationships using hierarchical bayesian models

People learn quickly when reasoning about causal relationships, making inferences from limited data and avoiding spurious inferences. Efficient learning depends on abstract knowledge, which is often domain or context specific, and much of it must be learned. While such knowledge effects are well documented, little is known about exactly how we acquire knowledge that constrains learning. This work focuses on knowledge of the functional form of causal relationships; there are many kinds of relationships that can apply between causes and their effects, and knowledge of the form such a relationship takes is important in order to quickly identify the real causes of an observed effect. We developed a hierarchical Bayesian model of the acquisition of knowledge of the functional form of causal relationships and tested it in five experimental studies, considering disjunctive and conjunctive relationships, failure rates, and cross-domain effects. The Bayesian model accurately predicted human judgments and outperformed several alternative models.     

Exemplars, prototypes, similarities, and rules in category representation: an example of hierarchical bayesian analysis

This article demonstrates the potential of using hierarchical Bayesian methods to relate models and data in the cognitive sciences. This is done using a worked example that considers an existing model of category representation, the Varying Abstraction Model (VAM), which attempts to infer the representations people use from their behavior in category learning tasks. The VAM allows for a wide variety of category representations to be inferred, but this article shows how a hierarchical Bayesian analysis can provide a unifying explanation of the representational possibilities using 2 parameters. One parameter controls the emphasis on abstraction in category representations, and the other controls the emphasis on similarity. Using 30 previously published data sets, this work shows how inferences about these parameters, and about the category representations they generate, can be used to evaluate data in terms of the ongoing exemplar versus prototype and similarity versus rules debates in the literature. Using this concrete example, this article emphasizes the advantages of hierarchical Bayesian models in converting model selection problems to parameter estimation problems, and providing one way of specifying theoretically based priors for competing models.     

A Bayesian hierarchical model for the measurement of working memory capacity

Working memory is the memory system that allows for conscious storage and manipulation of information. The capacity of working memory is extremely limited. Measurements of this limit, and what affects it, are critical to understanding working memory. Cowan (2001) and Pashler (1988) suggested applying multinomial tree models to data from change detection paradigms in order to estimate working memory capacity. Both Pashler and Cowan suggested simple formulas for estimating capacity with these models. However, in many cases, these simple formulas are inadequate, and may lead to inefficient or biased estimation of working memory capacity. I propose a Bayesian hierarchical alternative to the Pashler and Cowan formulas, and show that the hierarchical model outperforms the traditional formulas. The models are easy to use and appropriate for a wide range of experimental designs. An easy-to-use graphical user interface for fitting the hierarchical model to data is available.     

A bayesian optimal foraging model of human visual search

Real-world visual searches often contain a variable and unknown number of targets. Such searches present difficult metacognitive challenges, as searchers must decide when to stop looking for additional targets, which results in high miss rates in multiple-target searches. In the study reported here, we quantified human strategies in multiple-target search via an ecological optimal foraging model and investigated whether searchers adapt their strategies to complex target-distribution statistics. Separate groups of individuals searched displays with the number of targets per trial sampled from different geometric distributions but with the same overall target prevalence. As predicted by optimal foraging theory, results showed that individuals searched longer when they expected more targets to be present and adjusted their expectations on-line during each search by taking into account the higher-order, across-trial target distributions. However, compared with modeled ideal observers, participants systematically responded as if the target distribution were more uniform than it was, which suggests that training could improve multiple-target search performance.     

The effect of expertise on collaborative problem solving

Why do some groups succeed where others fail? We hypothesise that collaborative success is achieved when the relationship between the dyad's prior expertise and the complexity of the task creates a situation that affords constructive and interactive processes between group members. We call this state the zone of proximal facilitation in which the dyad's prior knowledge and experience enables them to benefit from both knowledge-based problem-solving processes (e.g., elaboration, explanation, and error correction) andcollaborative skills (e.g., creating common ground, maintaining joint attention to the task). To test this hypothesis we conducted an experiment in which participants with different levels of aviation expertise, experts (flight instructors), novices (student pilots), and non-pilots, read flight problem scenarios of varying complexity and had to identify the problem and generate a solution with either another participant of the same level of expertise or alone. The non-pilots showed collaborative inhibition on problem identification in which dyads performed worse than their predicted potential for both simple and complex scenarios, whereas the novices and experts did not. On solution generation the non-pilot and novice dyads performed at their predicted potential with no collaborative inhibition on either simple or complex scenarios. In contrast, expert dyads showed collaborative gains, withdyads performing above their predicted potential, but only for the complex scenarios. On simple scenarios the expert dyads showed collaborative inhibition and performed worse than their predicted potential. We discuss the implications of these results for theories of collaborative problem solving.     

The effects of member expertise on group decision-making and performance

This study assesses the effects of member expertise on group decision-making and group performance. Three-person cooperative groups and three independent individuals solved either an easy or moderately difficult version of the deductive logic game Mastermind. Experimental groups were given veridical performance information, i.e., the members' rankings on prior individual administrations of the task. Control groups were not provided with this information. Results supported the predictions of this study: (1) groups gave more weight to the input of their highest performing members with the group decision-making process being best approximated by post hoc “expert weighted” social decision schemes and (2) groups performed at the level of the best of an equivalent number of individuals.     

Cognitive model decomposition of the BART: Assessment and application

The Balloon Analogue Risk Task, or BART, aims to measure risk taking behavior in a controlled setting. In order to quantify the processes that underlie performance on the BART, Wallsten, Pleskac, and Lejuez (2005) proposed a series of mathematical models whose parameters have a clear psychological interpretation. Here we examine a 2-parameter simplification of Wallsten et al.’s preferred 4-parameter model. A parameter recovery study shows that — with plausible restrictions on the number of participants and trials — both parameters (i.e., risk taking γ⁺ and response consistency β) can be estimated accurately. To demonstrate how the 2-parameter model can be used in practice, we implemented a Bayesian hierarchical version and applied it to an empirical data set in which participants performed the BART following various amounts of alcohol intake.     

High school students’ career decision-making process: A longitudinal study of one choice

This study investigated high school students’ process of choosing a study in higher education, using a longitudinal design. A sample of 535 adolescents participated at the beginning, middle, and end of Grade 12. Latent curve modeling revealed evidence for a mean developmental increase in the career decisional tasks of orientation, exploration, decisional status, and commitment. In general, the growth trajectories were in line with theoretical models in which orientation and broad exploration are important at the beginning of the career decision-making process whereas in-depth exploration and decisional status are considered important later on. Furthermore, the results showed substantial variability between adolescents in their initial levels of career decisional tasks and in the rate of change. Finally, growth in orientation related to growth in environmental and self-exploration, and growth in environmental exploration related to progress in commitment and decisional status.     

Servant Leadership Behaviour Scale: A hierarchical model and test of construct validity

Servant leadership is widely believed to be a multidimensional construct. However, existing measures of servant leadership typically suffer from highly correlated dimensions, raising concerns over discriminant validity. We set out in this study to examine the dimensionality of the hypothesized six-factor Servant Leadership Behaviour Scale (SLBS) and validate a hierarchical model of servant leadership. Using structural equation modelling, convergent and discriminant validity of the higher-order servant leadership construct was examined in relation to measures of character-related behaviour, Machiavellian leadership orientation, and social desirability. Our findings suggest that the SLBS assesses a single, higher-order construct of servant leadership with evidence of construct validity and absence of social desirability bias. The high correlation between servant leadership and character-related behaviour supports our expectation that servant leadership reflects in large part the character strength of the leader.     

The influence of discrete emotions on judgement and decision-making: A meta-analytic review

During the past three decades, researchers interested in emotions and cognition have attempted to understand the relationship that affect and emotions have with cognitive outcomes such as judgement and decision-making. Recent research has revealed the importance of examining more discrete emotions, showing that same-valence emotions (e.g., anger and fear) differentially impact judgement and decision-making outcomes. Narrative reviews of the literature (Lerner & Tiedens, 2006 Lerner, J. S. and Tiedens, L. Z. 2006. Portrait of the angry decision maker: How appraisal tendencies shape anger's influence on cognition. Journal of Behavioral Decision-Making, 19: 115–137. [Crossref], [Web of Science ®] , [Google Scholar]; Pham, 2007 Pham, M. T. 2007. Emotion and rationality: A critical review and interpretation of empirical evidence. Review of General Psychology, 11(2): 155–178. [Crossref], [Web of Science ®] , [Google Scholar]) have identified some under-researched topics, but provide a limited synthesis of findings. The purpose of this study was to review the research examining the influence of discrete emotions on judgement and decision-making outcomes and provide an assessment of the observed effects using a meta-analytic approach. Results, overall, show that discrete emotions have moderate to large effects on judgement and decision-making outcomes. However, moderator analyses revealed differential effects for study-design characteristics and emotion-manipulation characteristics by emotion type. Implications are discussed.     

A rational model of the effects of distributional information on feature learning

Most psychological theories treat the features of objects as being fixed and immediately available to observers. However, novel objects have an infinite array of properties that could potentially be encoded as features, raising the question of how people learn which features to use in representing those objects. We focus on the effects of distributional information on feature learning, considering how a rational agent should use statistical information about the properties of objects in identifying features. Inspired by previous behavioral results on human feature learning, we present an ideal observer model based on nonparametric Bayesian statistics. This model balances the idea that objects have potentially infinitely many features with the goal of using a relatively small number of features to represent any finite set of objects. We then explore the predictions of this ideal observer model. In particular, we investigate whether people are sensitive to how parts co-vary over objects they observe. In a series of four behavioral experiments (three using visual stimuli, one using conceptual stimuli), we demonstrate that people infer different features to represent the same four objects depending on the distribution of parts over the objects they observe. Additionally in all four experiments, the features people infer have consequences for how they generalize properties to novel objects. We also show that simple models that use the raw sensory data as inputs and standard dimensionality reduction techniques (principal component analysis and independent component analysis) are insufficient to explain our results.     

A computational model of early argument structure acquisition

How children go about learning the general regularities that govern language, as well as keeping track of the exceptions to them, remains one of the challenging open questions in the cognitive science of language. Computational modeling is an important methodology in research aimed at addressing this issue. We must determine appropriate learning mechanisms that can grasp generalizations from examples of specific usages, and that exhibit patterns of behavior over the course of learning similar to those in children. Early learning of verb argument structure is an area of language acquisition that provides an interesting testbed for such approaches due to the complexity of verb usages. A range of linguistic factors interact in determining the felicitous use of a verb in various constructions —associations between syntactic forms and properties of meaning that form the basis for a number of linguistic and psycholinguistic theories of language. This article presents a computational model for the representation, acquisition, and use of verbs and constructions. The Bayesian framework is founded on a novel view of constructions as a probabilistic association between syntactic and semantic features. The computational experiments reported here demonstrate the feasibility of learning general constructions, and their exceptions, from individual usages of verbs. The behavior of the model over the timecourse of acquisition mimics, in relevant aspects, the stages of learning exhibited by children. Therefore, this proposal sheds light on the possible mechanisms at work in forming linguistic generalizations and maintaining knowledge of exceptions.     

A tutorial on adaptive design optimization

Experimentation is ubiquitous in the field of psychology and fundamental to the advancement of its science, and one of the biggest challenges for researchers is designing experiments that can conclusively discriminate the theoretical hypotheses or models under investigation. The recognition of this challenge has led to the development of sophisticated statistical methods that aid in the design of experiments and that are within the reach of everyday experimental scientists. This tutorial paper introduces the reader to an implementable experimentation methodology, dubbed Adaptive Design Optimization, that can help scientists to conduct “smart” experiments that are maximally informative and highly efficient, which in turn should accelerate scientific discovery in psychology and beyond.     

A bio-inspired model of behavior considering decision-making and planning,spatial attention and basic motor commands processes

Cognitive architectures (CA) are an IA approach to implement computer systems with human-like behavior. Fundamental exhibited human capabilities include planning and decision-making. In that regard, numerous AI systems successfully exhibit human-like behavior but are limited to either achieving specific objectives or are restrained to too heavily constrained environments, which makes them unsuitable in the presence of unforeseen situations where autonomy is required. To try to alleviate the problem, we present a bio-inspired computational model to solve the autonomous navigation problem of a computational entity in a controlled context. This proposal is the result of the interaction between planning and decision-making, spatial attention and the motor cognitive functions. The proposed model is based on neuroscientific evidence concerning the involved cognitive functions and is part of a more general cognitive architecture. In the case study developed to validate our idea, we can see that the processes previously identified play an important role to accomplish spatial navigation. In the case study presented, an agent achieves the navigation over an unexplored maze from an initial to a final position successfully. The reunited results motivate us to continue improving our model considering attentional information to influence the agent’s motor behavior.     

Intra‐sentential context effects on the interpretation of logical metonymy⋆

Verbs such as enjoy in the student enjoyed the book exhibit logical metonymy: enjoy is interpreted as enjoy reading. Theoreticalwork [Computational Linguistics 17 (4) (1991) 409; The Generative Lexicon, MIT Press, Cambridge, MA, 1995] predicts that this interpretation can be influenced by intra‐sentential context, e.g., by the subject of enjoy. In this article, we test this prediction using a completion experiment and find that the interpretation of a metonymic verb is influenced by the semantic role of its subject. We present a Bayesian model that accounts for the interpretation of logical metonymy and achieves a good fit on our experimental data. We show that the parameters of the model can be estimated from completion data or from corpus data.     

A tutorial on Bayes factor estimation with the product space method

The Bayes factor is an intuitive and principled model selection tool from Bayesian statistics. The Bayes factor quantifies the relative likelihood of the observed data under two competing models, and as such, it measures the evidence that the data provides for one model versus the other. Unfortunately, computation of the Bayes factor often requires sampling-based procedures that are not trivial to implement. In this tutorial, we explain and illustrate the use of one such procedure, known as the product space method (Carlin & Chib, 1995). This is a transdimensional Markov chain Monte Carlo method requiring the construction of a “supermodel” encompassing the models under consideration. A model index measures the proportion of times that either model is visited to account for the observed data. This proportion can then be transformed to yield a Bayes factor. We discuss the theory behind the product space method and illustrate, by means of applied examples from psychological research, how the method can be implemented in practice.     

Exploring the effects of congruence and Holland’s personality codes on job satisfaction: An application of hierarchical linear modeling techniques

This study applied hierarchical linear modeling to investigate the effect of congruence on intrinsic and extrinsic aspects of job satisfaction. Particular focus was given to differences in job satisfaction by gender and by Holland’s first-letter codes. The study sample included nationally represented 1462 female and 1280 male college graduates who held full-time jobs in 2000. Congruence was measured by using actual occupational and academic major codes. After controlling for the differences across three-letter Holland personality codes, results indicate that job satisfaction was largely a function of individual attributes rather than disparities in personality types. Holland’s congruence was most effective in explaining intrinsic dimensions of job satisfaction for females, and less effective for extrinsic job satisfaction for both genders. Social females and males were less satisfied with extrinsic satisfaction, while Conventional and Social females were more satisfied with intrinsic features in their vocational environments.     

A model of human performance on the traveling salesperson problem

A computational model is proposed of how humans solve the traveling salesperson problem (TSP). Tests of the model are reported, using human performance measures from a variety of 10-, 20-, 40-, and 60-node problems, a single 48-node problem, and a single 100-node problem. The model provided a range of solutions that approximated the range of human solutions and conformed closely to quantitative and qualitative characteristics of human performance. The minimum path lengths of subjects and model deviated by average absolute values of 0.0%, 0.9%, 2.4%, 1.4%, 3.5%, and 0.02% for the 10-, 20-, 40-, 48-, 60-, and 100-node problems, respectively. Because the model produces a range of solutions, rather than a single solution, it may find better solutions than some conventional heuristic algorithms for solving TSPs, and comparative results are reported that support this suggestion.