Decision making

Cao, Y., Nobandegani, A., & Shultz, T. (2022). A Resource-Rational Process Model of Violation of Cumulative Independence. In J. Culbertson, A. Perfors, H. Rabagliati, & V. Ramenzoni (Eds.), Proceedings of the 44th Annual Conference of the Cognitive Science Society, (3051-3057). Toronto, ON: Cognitive Science Society. - Human decision-making is filled with numerous paradoxes and violations of rationality principles. A particularly notable example is violation of cumulative independence (VoCI). Recently, there has been a surge of interest in theorizing and developing a resource-rational foundation for many such phenomena. Here we ask whether VoCI also could be given a resource-rational basis, namely could it be explained in terms of the optimal use of limited cognitive resources? We examine VoCI through the lens of modern psychological theories of bounded rationality, presenting the first resource rational account of VoCI. We also discuss the implications of this work for risky decision-making, and more broadly, human rationality. pdf
Xia, F., Nobandegani, A., & Shultz, T., & Bhui, R. (2022). A Resource-Rational Process-Level Account of Violation of Stochastic Dominance. In J. Culbertson, A. Perfors, H. Rabagliati, & V. Ramenzoni (Eds.), Proceedings of the 44th Annual Conference of the Cognitive Science Society, (3133-3139). Toronto, ON: Cognitive Science Society. - Stochasitic dominance is widely considered a pillar of rational choice and has played a major role in the history of theorizing and developing models of human decision-making. A wealth of empirical evidence reveals that humans’ violation of dominance is both substantial and systematic. Here we as whether violation of dominance be given a rational basis? Specifically, could it be understood in terms of the optimal use of limited cognitive resources? We present the first resource-rational account of stochastic dominance, the most empirically studied version of dominance. We show that a resource-rational process model, sample-based expected utility (SbEU), provides a unified account of a broad range of empirical results on violation of stochastic dominance. We then discuss the implications of this work for risky decision-making, and more broadly, human rationality. pdf
Nobandegani, A., Shultz, T., & Rish, I. (2022). Cognitive Models as Simulators: The Case of Moral Decision-Making. In J. Culbertson, A. Perfors, H. Rabagliati, & V. Ramenzoni (Eds.), Proceedings of the 44th Annual Conference of the Cognitive Science Society, (3827). Toronto, ON: Cognitive Science Society. - To achieve desirable performance, current AI systems often require huge amounts of training data. This is especially problematic in domains where collecting data is both expensive and time-consuming, e.g., where AI systems require numerous interactions with humans, collecting feedback from them. We substantiate the idea of cognitive models as simulators, by having AI systems interact with, and collect feedback from, cognitive models instead of humans, thereby making their training process both less costly and faster. Here, we leverage this idea in the context of moral decision-making, by having reinforcement learning (RL) agents learn about fairness through interacting with a cognitive model of the Ultimatum Game (UG; Nobandegani, Destais, & Shultz, 2020), a canonical task in behavioral and brain sciences for studying fairness. Interestingly, these RL agents learn to rationally adapt their behavior depending on the emotional state of their simulated UG responder. pdf
Lizotte, M., Nobandegani, A. S., & Shultz, T. R. (2021). Emotions in Games: Toward a Unified Process-Level Account. In Proceedings of the 43rd Annual Conference of Cognitive Science Society, 43, 2630-2636). - Strategic decision-making is chiefly studied in behavioral economics using multi-agent games. Decades of empirical research has revealed that emotions play a crucial role in strategic decision-making, calling into question the “emotionless” homo economicus. Here we present a unified process-level account of a broad range of empirical findings on the effect of emotions in the two most studied games in behavioral sciences: the Prisoner’s Dilemma and the Ultimatum game. Under the empirically well-supported assumption that emotions modulate loss aversion, we show that Nobandegani et al.’s (2018) sample-based expected utility model can account for the effect of emotions on: (i) cooperation rate in Prisoner’s Dilemma, and (ii) the rejection rate of unfair offers in the Ultimatum game. This has implications for emotion research, and for developing a unified process-level account of the role of emotions in strategic decision-making. pdf
Nobandegani, A. S., Shultz, T. R., & Dubé, L. (2021). A Unified, Resource-Rational Account of the Allais and Ellsberg Paradoxes. In Proceedings of the 43rd Annual Conference of Cognitive Science Society, 43, 1208-1214. - Decades of empirical and theoretical research on human decision-making has broadly categorized it into two, separate realms: decision-making under risk and decision-making under uncertainty, with the Allais paradox and the Ellsberg paradox being a prominent example of each, respectively. Here we present the first unified, resource-rational account of these two paradoxes. Specifically, we show that Nobandegani et al.’s (2018) sample-based expected utility model provides a unified, process-level account of two variants of the Allais paradox (the common-consequence effect and the common-ratio effect) and the Ellsberg paradox. Our work suggests that the broad framework of resource-rationality permits a unified treatment of decision-making under risk and decision-making under uncertainty, thus approaching a unified account of human decision-making. pdf
Nobandegani, A. S., & Shultz, R. (2020). A resource-rational, process-level account of the St. Petersburg Paradox. Topics in Cognitive Science, 12, 417-432. - The St. Petersburg paradox is a centuries‐old philosophical puzzle concerning a lottery with infinite expected payoff for which people are only willing to pay a small amount to play. Despite many attempts and several proposals, no generally accepted resolution is yet at hand. In this work, we present the first resource‐rational, process‐level explanation of this paradox, demonstrating that it can be accounted for by a variant of normative expected utility valuation which acknowledges cognitive limitations. We show that our metacognitively rational model, sample‐based expected utility (SbEU), can account for major experimental findings on this paradox. Our resolution is consistent with two empirically well‐supported assumptions: (a) People use only a few samples in probabilistic judgments and decision‐making, and (b) people tend to overestimate the probability of extreme events in their judgment. We frame the St. Petersburg gamble as a risky gamble whose process‐level explanation is consistent with a broader process‐level model of human decision‐making under risk. (An earlier version won Best paper award in Computational Modeling of Higher-level Cognition at the 41st Annual Conference of the Cognitive Science Society, 2019, Montreal. Another version was selected to represent the 2019 Cognitive Science conference at the February 2020 Sister AAAI Conference on Artificial Intelligence, NYC. pdf
Nobandegani, A. S., & Shultz, T. R. (2020). A resource-rational mechanistic account of human coordination strategies. In S. Denison, M. Mack, Y. Xu, & B. Armstrong (Eds.), Proceedings of the 42nd Annual Conference of the Cognitive Science Society, (3356-3362). Toronto, ON: Cognitive Science Society. - Humans often coordinate their actions in order to reach a mutually advantageous state. These circumstances are chiefly modeled by coordination games, a well-known class of games extensively studied in behavioral economics. In this work, we present the first resource-rational mechanistic approach to coordination games, showing that a variant of normative expected-utility maximization acknowledging cognitive limitations can account for several major experimental findings on human coordination behavior in strategic settings. Concretely, we show that Nobandegani et al.’s (2018) rational process model, sample-based expected utility, provides a unified account of (1) the effect of time pressure on human coordination, and (2) how systematic variations of riskvs. payoff-dominance affect coordination behavior. Importantly, Harsanyi and Selten’s (1988) theory of equilibrium selection fails to account for (1-2). As such, our work suggests that the optimal use of limited cognitive resources may lie at the core of human coordination behavior. We conclude by discussing the implication of our work for understanding human strategic behavior, moral decision-making, and human rationality. pdf
Nobandegani, A. S., Destais, C., & Shultz, T. R. (2020). A resource-rational process model of fairness in the ultimatum game. In S. Denison, M. Mack, Y. Xu, & B. Armstrong (Eds.), Proceedings of the 42nd Annual Conference of the Cognitive Science Society, (3205-3211). Toronto, ON: Cognitive Science Society. - Widely regarded as the cornerstone of justice (Rawls, 1971), fairness constitutes one of the pillars of human morality. The Ultimatum Game (UG), extensively studied in behavioral economics, is the canonical task for studying fairness. In sharp contrast to the predictions of normative standards in game theory, people typically reject low offers in UG. In this work, we present the first resource-rational process model of UG. By taking into account people’s expectations, we show that Nobandegani et al.'s (2018) resource-rational process model, sample-based expected-utility, provides a unified account of several experimental findings in UG: the effects of expectation, competition, and time pressure. Assuming that expectation serves as a reference point for subjective valuation of an offer, we show that the rejection of low offers in UG can arise from purely self-interested expected-utility maximization. We conclude by discussing the implication of our work for moral decision-making and human rationality. pdf
Nobandegani, A.S., da Silva-Castanheira, K., Shultz, T., & Otto, A.R. (2019). A resource-rational mechanistic approach to one-shot non-cooperative games: the case of prisoner’s dilemma. In A.K. Goel, C.M. Seifert, & C. Freksa (Eds.), Proceedings of the 41st Annual Conference of the Cognitive Science Society (pp. 863-869). Montreal, QC: Cognitive Science Society. - The concept of Nash equilibrium has played a profound role in economics, and is widely accepted as a normative stance for how people should choose their strategies in competitive environments. However, extensive empirical evidence shows that people often systematically deviate from Nash equilibrium. In this work, we present the first resource-rational mechanistic approach to one-shot, non-cooperative games (ONG), showing that a variant of normative expected-utility maximization acknowledging cognitive limitations can account for important deviations from the prescriptions of Nash equilibrium in ONGs. Concretely, we show that Nobandegani et al.’s (2018) metacognitively-rational model, sample-based expected utility, can account for purportedly irrational cooperation rates observed in one-shot, non-cooperative Prisoner’s Dilemma, and can accurately explain how cooperation rate varies depending on the parameterization of the game. Additionally, our work provides a resource-rational explanation of why people with higher general intelligence tend to cooperate less in OPDs, and serves as the first (Bayesian) rational, process-level explanation of a well-known violation of the law of total probability in OPDs, documented by Shafir and Tversky (1992), which has resisted explanation by a model governed by classical probability theory for nearly three decades. Surprisingly, our work demonstrates that cooperation can arise from purely selfish, expected-utility maximization subject to cognitive limitations. pdf
Nobandegani, A.S., da Silva-Castanheira, K., Shultz, T., & Otto, A.R. (2019). A resource-rational process-level account of the St. Petersburg paradox. In A.K. Goel, C.M. Seifert, & C. Freksa (Eds.), Proceedings of the 41st Annual Conference of the Cognitive Science Society (pp. 870-876). Montreal, QC: Cognitive Science Society. (Best paper award in Computational Modeling of Higher-level Cognition.) - The St. Petersburg paradox is a centuries-old philosophical puzzle concerning a lottery with infinite expected payoff, on which people are, nevertheless, willing to place only a small bid. Despite many attempts and several proposals, no generally-accepted resolution is yet at hand. In this work, we present the first resource-rational process-level explanation of this paradox, demonstrating that it can be accounted for by a variant of normative expected-utility-maximization which acknowledges cognitive limitations. Specifically, we show that Nobandegani et al.’s (2018) metacognitively-rational model, sample-based expected utility (SbEU), can account for major experimental findings on this paradox. Crucially, our resolution is consistent with two empirically well-supported assumptions: (1) people use only a few samples in probabilistic judgments and decision-making, and (2) people tend to overestimate the probability of extreme events in their judgment. pdf
Nobandegani, A.S. & Shultz, T. (2019). Toward a formal science of heuristics. In A.K. Goel, C.M. Seifert, & C. Freksa (Eds.), Proceedings of the 41st Annual Conference of the Cognitive Science Society (pp. 877-882). Montreal, QC: Cognitive Science Society. - Heuristics are simple, effective cognitive processes that deliberately ignore parts of information relevant to decision-making. Ecological rationality, as an essential part of the Adaptive Toolbox research program on heuristics, investigates the environmental conditions under which simple heuristics would outperform complex models of decision-making, thereby providing support for the surprising less-is-more effect. In this work, we present a new research program, dubbed formal science of heuristics (FSH), that complements the ecological rationality research, developing it into a much richer research program. Concretely, FSH sets to (i) mathematically delineate the broadest class of environmental conditions under which a heuristic is fully optimal, and (ii) formally investigate how deviations from those conditions would lead to degradation of performance, thereby allowing for a mathematically rigorous characterization of their robustness. As an instantiation of the FSH research program, we present several analytical results aiming to delineate the mildest conditions granting the optimality of a well-known heuristic: Take The Best. We conclude by discussing the implications that pursuit of FSH could have on the science of heuristics. pdf
Nobandegani, A.S., da Silva Castanheira, K., Shultz, T.R., & Otto, A.R. (2019). Decoy effect and violation of betweenness in risky decision making: a resource-rational mechanistic account. Proceedings of the 17th International Conference on Cognitive Modeling, 1-6. - A wealth of experimental evidence shows that, contrary to normative models of choice, people’s preferences are markedly swayed by the context in which options are presented. In this work, we present the first resource-rational, mechanistic account of the decoy effect—a major contextual effect in risky decision making. Our model additionally explains a related, well-known behavioral departure from expected utility theory: violation of betweenness. We demonstrate that, contrary to widely held views, these effects can be accounted for by a variant of normative expected-utility-maximization, which acknowledges cognitive limitations. Our work is consistent with two empirically well-supported hypotheses: (i) In probabilistic reasoning and judgment, a cognitive system accumulates information through sampling, and (ii) People engage in pairwise comparisons when choosing between multiple alternatives. pdf
Nobandegani, A.S., da Silva Castanheira, K., O’Donnell, T.J., & Shultz, T.R. (2019). On robustness: an undervalued dimension of human rationality. Proceedings of the 17th International Conference on Cognitive Modeling. 1-6. - Human rationality is predominantly evaluated by the extent to which the mind respects the tenets of normative formalisms like logic and probability theory, and is often invoked by appealing to the notion of optimality. Drawing mainly on Simon’s bounded rationality principle, there has been a surge in the understanding of human rationality with respect to the limited computational and cognitive resources the mind is faced with. In this work, we focus on another fairly underappreciated yet crucial facet of rationality, robustness: insensitivity of a model’s performance to miscalculations of its parameters. We argue that an integrative pursuit of three facets (optimality, efficient use of limited resources, and robustness) would be a fruitful approach to understanding the extent of human rationality. We present several novel formalizations of robustness and discuss a recently proposed metacognitively-rational model of risky choice which is surprisingly robust to under- and over-estimation of its focal parameter, accounting for well-known framing effects in human decision-making under risk. We close by highlighting the ubiquitous presence of robustness in natural as well as artificial realms, and the implications of our work for rationalistic approaches to understanding human cognition at the algorithmic level of analysis. pdf
Nobandegani, A.S., Campoli, W, & Shultz, T.R. (2019). Bringing order to the cognitive fallacy zoo. Proceedings of the 17th International Conference on Cognitive Modeling. 1-6. - Investigations into human reasoning, judgment, and decision-making have found numerous cognitive biases and fallacies, with new ones continually emerging, leading to a state of affairs which can fairly be characterized as the cognitive fallacy zoo. In this work, we formally present a principled way to bring order to this zoo. We introduce the idea of establishing implication relationships (IRs) between cognitive fallacies, formally characterizing how one fallacy implies another. IR is analogous to, and partly inspired by, the fundamental concept of reduction in computational complexity theory. We present several examples of IRs involving experimentally well-documented cognitive fallacies: base-rate neglect, availability bias, conjunction fallacy, decoy effect, framing effect, and Allais paradox. We conclude by discussing how our work: (i) allows for identifying those pivotal cognitive fallacies whose investigation would be the most rewarding research agenda, and (ii) permits a systematized, guided research program on cognitive fallacies, motivating influential theoretical as well as experimental avenues of future research. pdf
Nobandegani, A. S., da Silva Castanheira, K., Otto, A. R., & Shultz, T. R. (2018). Over-representation of extreme events in decision-making: A rational metacognitive account. In T. T. Rogers, M. Rau, X. Zhu, & C. W. Kalish (Eds.), Proceedings of the 40th Annual Conference of the Cognitive Science Society (pp. 2391-2396). Austin, TX: Cognitive Science Society. - The Availability bias, manifested in the over-representation of extreme eventualities, is a well-known cognitive bias, and is generally taken as evidence of human irrationality. In this work, we present the first rational, metacognitive account of the Availability bias, formally articulated at Marr’s algorithmic level of analysis. Concretely, we present a normative, metacognitive model of how a cognitive system should overrepresent extreme eventualities, depending on the amount of time available for decision-making. Our Sample-based Expected Utility model also accounts for two well-known framing effects in human decision-making under risk—the fourfold pattern of risk preferences in outcome probability (Tversky & Kahneman, 1992) and in outcome magnitude (Markovitz, 1952)—thereby providing the first metacognitively-rational basis for the aforementioned effects. Empirical evidence confirms an important prediction of our model. Surprisingly, our model is strikingly robust with respect to its focal parameter. We discuss the implications of our work for studies on human decision-making, and conclude by presenting a counterintuitive prediction of our model, which, if confirmed, would have intriguing implications for human decision-making under risk. To our knowledge, our model is the first metacognitive, resource-rational process model of cognitive biases in decision-making. Notably, our work also contributes to the fields of artificial intelligence and computational statistics, by presenting a previously unknown proposal distribution, with firm rational grounds, broadly applicable to the influential subfield of importance sampling Monte Carlo methods. pdf
Helfer, P., & Shultz, T. R. (2014). The effects of nutrition labeling on consumer food choice: a psychological experiment and computational model. Annals of the New York Academy of Sciences, 1331(1), 174-185. - Quantitative, single-attribute nutrition labels have greater usability than multiattribute and binary labels, with or without time pressure. A decision-field-theory neural network simulates these psychological results and provides explanatory insights into them. pdf