PKPeter D. Kvam

Peter D. Kvam

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Peter D. Kvam has a total of 135 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Moving on From Simple Operationalism: Anchoring Theoretical Constructs in Rich Relational Models

    by Peter D. Kvam, Abhay Alaukik

    Operational definitions have long been a core approach to measuring and relating observed data to theoretical constructs in psychological science. However, many contemporary modeling approaches conflict with the tacit assumptions of operationalism as it is implemented in psychology. Counterintuitively, these departures imbue model-based definitions of constructs with superior measurement properties, such as improved reliability and validity, when compared with their operationalizations based directly on observables. Instead of relying on operational definitions of constructs, we instead suggest that psychology should adopt relational definitions or relational operationalizations, representing constructs as latent variables in a multilevel generative model of behavior, self-report, or neuroimaging data. These model-based metrics can better reflect measurement error at multiple levels, account for the interactions between measurement devices (tasks, scales) and measurement objects (participants, processes), provide a holistic account of latent constructs and how they manifest across different measurements, facilitate convergence or discrimination tests among different tasks seeking to measure the same construct, and improve scientific communication by clarifying core psychological concepts. Relational definitions of important constructs should naturally emerge as researchers apply models more regularly, and these definitions and models will improve as researchers discover mathematical approaches that are suited to describing psychological processes.

  • Open Access

    Dissociating Reasoning and Planning through Markov and Quantum Metacognitive Dynamics

    by Ritesh K. Malaiya, Richard M. Golden, Michael J. Lundie, Peter D. Kvam, Stacie L. Warren

    Online metacognitive processes support learning and decision-making by monitoring and integrating heuristic cues, such as evidence variability and conflict, throughout deliberation. Certain computational approaches, such as Markov and Quantum random-walk models (MRM and QRM), describe these metacognitive processes as dynamical systems, enabling inference of latent metacognitive states from response choices and response times without relying on explicit metacognitive judgments. MRM and QRM characterize how uncertainty is represented and updated during decision-making, with MRM tracking transitions between definite latent states and QRM allowing simultaneous superposition over competing latent states. In the present study, we evaluated the effectiveness of MRM and QRM in capturing latent metacognitive dynamics using explicit confidence judgments in reasoning tasks and subsequent behavioral patterns in planning tasks. QRM provided greater predictive utility than MRM across difficulty levels and high-conflict reasoning trials. In planning, MRM tended to provide greater predictive utility across harder planning conditions. Overall, model-derived TECS was comparable to or outperformed commonly used empirical measures in predicting subsequent behavior. These results suggest that Quantum dynamics may better capture reasoning under conflict, where participants integrate evidence across uncertainty levels. In contrast, Markov dynamics may align more naturally with the planning of sequential steps, particularly in harder planning conditions involving the evaluation of potential action paths. In other words, reasoning and planning may differ in how uncertainty is represented, monitored, and resolved over time, and random-walk models provide a principled means of characterizing these differences in cognitive dynamics.

  • Cognitive mechanisms of subjective value in multiattribute pricing.

    by Anderson K. Fitch, Peter D. Kvam

    Understanding how people assign subjective value to outcomes with multiple attributes, such as risk and delay, is central to understanding the structure and manifestation of economic preferences. However, multiattribute preference has been primarily studied through binary choices. The price at which a person would buy, sell, or equate each prospect offers another measure of subjective value that may diverge from multiattribute choice. In both risky and intertemporal domains, choice and price preferences exhibit systematic preference reversals, where a smaller, sooner, or safer option is chosen while a larger, later, or riskier alternative is assigned a higher price. The present study takes a deep dive into how subjective value is assigned in each case in an attempt to reconcile these diverging measurements and methods of assessing value. To explain how and why preferences change across choice and price, the domains of gains and losses, price frames of buying and selling, and varying levels of time pressure, we develop a two-step neural network-based modeling approach. First, we tested cognitive mechanisms underlying value-based judgments and decisions using a switchboard model comparison. Next, we fit and evaluated individualized joint models, where all data from an individual are modeled using parameters and mechanisms that are specific to their best fitting model structure. While mechanisms like delay discounting and risk aversion are common to both models, our results suggest that anchoring and payoff sensitivity diverged between pricing and choice. Extensive differences across elicitation procedures indicate that a common representation of value may remain elusive. (PsycInfo Database Record (c) 2026 APA, all rights reserved).

  • Open Access

    Associations Between Steep Delay Discounting and Punishment Learning

    by Jeremy Haynes, Peter Kvam, Thomas Olino

    Reward valuation and reward and punishment learning are important constructs for understanding and intervening on mental illness. These constructs are frequently measured with behavioral tasks such as intertemporal choice tasks, measuring delay discounting, and the Iowa Gambling Task (IGT), measuring reward and punishment learning. Delay discounting refers to how people value delayed outcomes, with steep discounting (i.e., delayed outcomes holding little value) considered a form of impulsive decision-making. Reward and punishment learning refers to how people adjust their decision behavior in response to rewards and punishments, with deficits indicating atypical reward and punishment processing related to clinical outcomes such as substance abuse. In this study, we tested the association between delay discounting and computational measures of reward and punishment learning in adults (N = 299) that completed an intertemporal choice task and the play-or-pass IGT. We fit a joint model to the task data to simultaneously estimate parameters from the hyperbolic discounting model (for the intertemporal choice task) and a reinforcement learning model (for the play-or-pass IGT). Our results indicate that steep delay discounting is associated with lower performance on the IGT, specifically through lower punishment learning. That is, people who do not value delayed outcomes also show less learning from punishment, resulting in worse performance in the IGT. Implications of these findings are discussed in the context of improving assessment of reward valuation and reward and punishment learning as constructs relevant to mental illness.


  • The role of risk tolerance in navigation strategy decisions.

    by Eliany Perez, Peter Kvam, Rory McKemey, Steven M. Weisberg

    Everyday situations require us to face a trade-off to inform our decisions: exploit known information or explore for new information. Although both have risks, empirical research has not shown whether individuals prefer exploring or exploiting across contexts. In the present study, we examined the explore-exploit trade-off as a theoretical framework across two broad domains: decision making and spatial navigation. In this registered report, we applied computational modeling to human behavior on a novel version of the Iowa gambling task to predict behavior on a spatial navigation task in which the navigator must either exploit a learned, familiar route or explore a new shortcut. If the hypothesis that risk tolerance is a domain-general trait is correct, we predicted that explore-exploit patterns would correlate across these tasks. We also examine the predictive power of computational models for the Iowa gambling task on behavioral uncertainty and the role of confidence in spatial navigation strategy. Our findings suggest that, while there is some overlap in risk tolerance between spatial navigation and gambling, the influence of exploration and exploitation on navigational decision making is weaker than initially predicted. (PsycInfo Database Record (c) 2025 APA, all rights reserved).


  • Open Access

    Comparing likelihood-based and likelihood-free approaches to fitting and comparing models of intertemporal choice

    by Peter D. Kvam, Konstantina Sokratous, Anderson K. Fitch, Jasmin Vassileva

    Machine learning methods have recently begun to be used for fitting and comparing cognitive models, yet they have mainly focused on methods for dealing with models that lack tractable likelihoods. Evaluating how these approaches compare to traditional likelihood-based methods is critical to understanding the utility of machine learning for modeling and determining what role it might play in the development of new models and theories. In this paper, we systematically benchmark neural network approaches against likelihood-based approaches to model fitting and comparison, focusing on intertemporal choice modeling as an illustrative application. By applying each approach to intertemporal choice data from participants with substance use problems, we show that there is convergence between neural network and Bayesian methods when it comes to making inferences about latent processes and related substance use outcomes. For model comparison, however, classification networks significantly outperformed likelihood-based metrics. Next, we explored two extensions of this approach, using recurrent layers to allow them to fit data with variable stimuli and numbers of trials, and using dropout layers to allow for posterior sampling. We ultimately suggest that neural networks are better suited to fast parameter estimation and posterior sampling, applications to large data sets, and model comparison, while Bayesian MCMC methods should be preferred for flexible applications to smaller data sets featuring many conditions or experimental designs.


  • Open Access

    Decisions among shifting choice alternatives reveal option-general representations of evidence.

    by Peter D. Kvam, Konstantina Sokratous, Anderson K. Fitch

    Dynamic models of choice typically describe the decision-making process in terms of the degree or balance of support for available response options. However, these alternative-specific representations of support are liable to fail when the available options change during the course of a decision. We suggest that people may use alternative-general representations, where stimulus feature information-rather than option-specific support-is accumulated over time and mapped onto support for available options as they appear. We tested alternative-specific and alternative-general models of choice in two perceptual experiments where the available options could change during a trial. In the first study, we showed that changing the choice options partway through a trial resulted in no substantial difference in performance relative to a condition where the final options were always onscreen. This was supported by a quantitative model comparison that strongly favored an alternative-general (geometric) model over two alternative-specific models (diffusion and racing accumulator models). In the second study, the stimulus primed specific unavailable responses to test whether irrelevant support for unavailable options was integrated into the decision process. This study elicited a pattern of accuracy that could not have occurred unless participants accumulated support for options that were not yet available. Together, these experiments and modeling results indicate that the majority of participants rely on alternative-general representations of evidence during dynamic decisions among options that can change over time. Future work on decision behavior and its neural antecedents should explore the predictions of these alternative-general theories of choice. (PsycInfo Database Record (c) 2024 APA, all rights reserved).


  • Open Access

    The Tweedledum and Tweedledee of dynamic decisions: Discriminating between diffusion decision and accumulator models

    by Peter D. Kvam

    Theories of dynamic decision-making are typically built on evidence accumulation, which is modeled using racing accumulators or diffusion models that track a shifting balance of support over time. However, these two types of models are only two special cases of a more general evidence accumulation process where options correspond to directions in an accumulation space. Using this generalized evidence accumulation approach as a starting point, I identify four ways to discriminate between absolute-evidence and relative-evidence models. First, an experimenter can look at the information that decision-makers considered to identify whether there is a filtering of near-zero evidence samples, which is characteristic of a relative-evidence decision rule (e.g., diffusion decision model). Second, an experimenter can disentangle different components of drift rates by manipulating the discriminability of the two response options relative to the stimulus to delineate the balance of evidence from the total amount of evidence. Third, a modeler can use machine learning to classify a set of data according to its generative model. Finally, machine learning can also be used to directly estimate the geometric relationships between choice options. I illustrate these different approaches by applying them to data from an orientation-discrimination task, showing converging conclusions across all four methods in favor of accumulator-based representations of evidence during choice. These tools can clearly delineate absolute-evidence and relative-evidence models, and should be useful for comparing many other types of decision theories.


  • Open Access

    Using artificial intelligence to fit, compare, evaluate, and discover computational models of decision behavior.

    by Peter D. Kvam, Konstantina Sokratous, Anderson Fitch, Arend Hintze