Applying Generative Artificial Intelligence to cognitive models of decision making.

Introduction: Generative Artificial Intelligence has made significant impacts in many fields, including computational cognitive modeling of decision making, although these applications have not yet been theoretically related to each other. This work introduces a categorization of applications of Generative Artificial Intelligence to cognitive models of decision making. Methods: This categorization is used to compare the existing literature and to provide insight into the design of an ablation study to evaluate our proposed model in three experimental paradigms. These experiments used for model comparison involve modeling human learning and decision making based on both visual information and natural language, in tasks that vary in realism and complexity. This comparison of applications takes as its basis Instance-Based Learning Theory, a theory of experiential decision making from which many models have emerged and been applied to a variety of domains and applications. Results: The best performing model from the ablation we performed used a generative model to both create memory representations as well as predict participant actions. The results of this comparison demonstrates the importance of generative models in both forming memories and predicting actions in decision-modeling research. Discussion: In this work, we present a model that integrates generative and cognitive models, using a variety of stimuli, applications, and training methods. These results can provide guidelines for cognitive modelers and decision making researchers interested in integrating Generative AI into their methods.

Minimap: An interactive dynamic decision making game for search and rescue missions.

Many aspects of humans' dynamic decision-making (DDM) behaviors have been studied with computer-simulated games called microworlds. However, most microworlds only emphasize specific elements of DDM and are inflexible in generating a variety of environments and experimental designs. Moreover, despite the ubiquity of gridworld games for Artificial Intelligence (AI) research, only some tools exist to aid in the development of browser-based gridworld environments for studying the dynamics of human decision-making behavior. To address these issues, we introduce Minimap, a dynamic interactive game to examine DDM in search and rescue missions, which incorporates all the essential characteristics of DDM and offers a wide range of flexibility regarding experimental setups and the creation of experimental scenarios. Minimap specifically allows customization of dynamics, complexity, opaqueness, and dynamic complexity when designing a DDM task. Minimap also enables researchers to visualize and replay recorded human trajectories for the analysis of human behavior. To demonstrate the utility of Minimap, we present a behavioral experiment that examines the impact of different degrees of structural complexity coupled with the opaqueness of the environment on human decision-making performance under time constraints. We discuss the potential applications of Minimap in improving productivity and transparent replications of human behavior and human-AI teaming research. We made Minimap an open-source tool, freely available at  https://github.com/DDM-Lab/MinimapInteractiveDDMGame .

Building Human-Like Artificial Agents: A General Cognitive Algorithm for Emulating Human Decision-Making in Dynamic Environments.

One of the early goals of artificial intelligence (AI) was to create algorithms that exhibited behavior indistinguishable from human behavior (i.e., human-like behavior). Today, AI has diverged, often aiming to excel in tasks inspired by human capabilities and outperform humans, rather than replicating human cogntion and action. In this paper, I explore the overarching question of whether computational algorithms have achieved this initial goal of AI. I focus on dynamic decision-making, approaching the question from the perspective of computational cognitive science. I present a general cognitive algorithm that intends to emulate human decision-making in dynamic environments, as defined in instance-based learning theory (IBLT). I use the cognitive steps proposed in IBLT to organize and discuss current evidence that supports some of the human-likeness of the decision-making mechanisms. I also highlight the significant gaps in research that are required to improve current models and to create higher fidelity in computational algorithms to represent human decision processes. I conclude with concrete steps toward advancing the construction of algorithms that exhibit human-like behavior with the ultimate goal of supporting human dynamic decision-making.

Fostering Collective Intelligence in Human-AI Collaboration: Laying the Groundwork for COHUMAIN.

Artificial Intelligence (AI) powered machines are increasingly mediating our work and many of our managerial, economic, and cultural interactions. While technology enhances individual capability in many ways, how do we know that the sociotechnical system as a whole, consisting of a complex web of hundreds of human-machine interactions, is exhibiting collective intelligence? Research on human-machine interactions has been conducted within different disciplinary silos, resulting in social science models that underestimate technology and vice versa. Bringing together these different perspectives and methods at this juncture is critical. To truly advance our understanding of this important and quickly evolving area, we need vehicles to help research connect across disciplinary boundaries. This paper advocates for establishing an interdisciplinary research domain-Collective Human-Machine Intelligence (COHUMAIN). It outlines a research agenda for a holistic approach to designing and developing the dynamics of sociotechnical systems. In illustrating the kind of approach, we envision in this domain, we describe recent work on a sociocognitive architecture, the transactive systems model of collective intelligence, that articulates the critical processes underlying the emergence and maintenance of collective intelligence and extend it to human-AI systems. We connect this with synergistic work on a compatible cognitive architecture, instance-based learning theory and apply it to the design of AI agents that collaborate with humans. We present this work as a call to researchers working on related questions to not only engage with our proposal but also develop their own sociocognitive architectures and unlock the real potential of human-machine intelligence.

COHUMAIN: Building the Socio-Cognitive Architecture of Collective Human-Machine Intelligence.

In recent years, we have experienced rapid development of advanced technology, machine learning, and artificial intelligence (AI), intended to interact with and augment the abilities of humans in practically every area of life. With the rapid growth of new capabilities, such as those enabled by generative AI (e.g., ChatGPT), AI is increasingly at the center of human communication and collaboration, resulting in a growing recognition of the need to understand how humans and AI can integrate their inputs in collaborative teams. However, there are many unanswered questions regarding how human-AI collective intelligence will emerge and what the barriers might be. Truly integrated collaboration between humans and intelligent agents may result in a different way of working that looks nothing like what we know now, and it is important to keep the essential goal of human societal well-being and prosperity a priority. In this special issue, we begin to scope out the underpinnings of a socio-cognitive architecture for Collective HUman-MAchine INtelligence (COHUMAIN), which is the study of the capability of an integrated human and machine (i.e., intelligent technology) system to achieve goals in a wide range of environments. This topic consists of nine papers including a description of the conceptual foundation for a socio-cognitive architecture for COHUMAIN, empirical tests of some aspects of this architecture, research on proposed representations of intelligent agents that can jointly interact with humans, empirical tests of human-human and human-machine interactions, and philosophical and ethical issues to consider as we develop these systems.

Beyond collective intelligence: Collective adaptation.

We develop a conceptual framework for studying collective adaptation in complex socio-cognitive systems, driven by dynamic interactions of social integration strategies, social environments and problem structures. Going beyond searching for 'intelligent' collectives, we integrate research from different disciplines and outline modelling approaches that can be used to begin answering questions such as why collectives sometimes fail to reach seemingly obvious solutions, how they change their strategies and network structures in response to different problems and how we can anticipate and perhaps change future harmful societal trajectories. We discuss the importance of considering path dependence, lack of optimization and collective myopia to understand the sometimes counterintuitive outcomes of collective adaptation. We call for a transdisciplinary, quantitative and societally useful social science that can help us to understand our rapidly changing and ever more complex societies, avoid collective disasters and reach the full potential of our ability to organize in adaptive collectives.

SpeedyIBL: A comprehensive, precise, and fast implementation of instance-based learning theory.

Instance-based learning theory (IBLT) is a comprehensive account of how humans make decisions from experience during dynamic tasks. Since it was first proposed almost two decades ago, multiple computational models have been constructed based on IBLT (i.e., IBL models). These models have been demonstrated to be very successful in explaining and predicting human decisions in multiple decision-making contexts. However, as IBLT has evolved, the initial description of the theory has become less precise, and it is unclear how its demonstration can be expanded to more complex, dynamic, and multi-agent environments. This paper presents an updated version of the current theoretical components of IBLT in a comprehensive and precise form. It also provides an advanced implementation of the full set of theoretical mechanisms, SpeedyIBL, to unlock the capabilities of IBLT to handle a diverse taxonomy of individual and multi-agent decision-making problems. SpeedyIBL addresses a practical computational issue in past implementations of IBL models, the curse of exponential growth, that emerges from memory-based tabular computations. When more observations accumulate over time, there is an exponential growth of the memory of instances that leads directly to an exponential slowdown of the computational time. Thus, SpeedyIBL leverages parallel computation with vectorization to speed up the execution time of IBL models. We evaluate the robustness of SpeedyIBL over an existing implementation of IBLT in decision games of increased complexity. The results not only demonstrate the applicability of IBLT through a wide range of decision-making tasks, but also highlight the improvement of SpeedyIBL over its prior implementation as the complexity of decision features the of agents increase. The library is open sourced for the use of the broad research community.

Human and machine similarity judgments in forensic firearm comparisons.

It is unclear whether humans assess similarity differently than automated algorithms in firearms comparisons. Human participants (untrained in firearm examination) were asked to assess the similarity of pairs of images (from 0 to 100). A sample of 40 pairs of cartridge casing 2D-images was used. The images were divided into 4 groups according to their similarity as determined by an algorithm. Humans were able to distinguish between matches and non-matches (both when shown the 2 middle groups, as well as when shown all 4 groups). Thus, humans are able to make high-quality similarity judgments in firearm comparisons based on two images. The humans' similarity scores were superior to the algorithms' scores at distinguishing matches and non-matches, but inferior in assessing similarity within groups. This suggests that humans do not have the same group thresholds as the algorithm, and that a hybrid human-machine approach could provide better identification results than humans or algorithms alone.

Choice adaptation to changing environments: trends, feedback, and observability of change.

Making successful decisions in dynamic environments requires that we adapt our actions to the changing environmental conditions. Past research has found that people are slow to adapt their choices when faced with change, they tend to be over-reliant on initial experiences, and they are susceptible to factors such as feedback and the direction of change (trend). We build on these findings using two experiments that manipulate feedback and trend in a binary choice task, where decisions are made from experience. Feedback was either partial (providing only the outcome of the selected choice) or full (providing outcomes of the selected and the forgone choice) and the expected value of one option either increased, decreased, or remained constant. Crucially, although the two choice options had equal expected value averaged across all trials, their expected values on individual trials differed, and halfway through 100 choice trials the choice option with higher expected value switched, requiring participants to adapt their choices in order to maximize their outcomes. In Experiment 1, the probability of receiving the high-value outcome changed over time. In Experiment 2, the outcome value changed over time. Generally, we found that participants had trouble adapting to change: full feedback led to more maximization than partial feedback before the switch but did not make a difference after the switch, suggesting stickiness and poor adaptation. Slightly better adaptation was found for changing outcome values over changing probabilities, implying that the observability of the element of change influences adaptation.

Patterns of choice adaptation in dynamic risky environments.

An important aspect of making good decisions is the ability to adapt to changes in the values of available choice options, and research suggests that we are poor at changing behavior and adapting our choices successfully. The current paper contributes to clarifying the role of memory on learning and successful adaptation to changing decision environments. We test two aspects of changing decision environments: the direction of change and the type of feedback. The direction of change refers to how options become more or less rewarding compared to other options, over time. Feedback refers to whether full or partial information about decision outcomes is received. Results from behavioral experiments revealed a robust effect of the direction of change: risk that becomes more rewarding over time is harder to detect than risk that becomes less rewarding over time; even with full feedback. We rely on three distinct computational models to interpret the role of memory on learning and adaptation. The distributions of individual model parameters were analyzed in relation to participants' ability to successfully adapt to the changing conditions of the various decision environments. Consistent across the three models and two distinct data sets, results revealed the importance of recency as an individual memory component for choice adaptation. Individuals relying more on recent experiences were more successful at adapting to change, regardless of its direction. We explain the value and limitations of these findings as well as opportunities for future research.

Learning About the Effects of Alert Uncertainty in Attack and Defend Decisions via Cognitive Modeling.

OBJECTIVE: We aim to learn about the cognitive mechanisms governing the decisions of attackers and defenders in cybersecurity involving intrusion detection systems (IDSs). BACKGROUND: Prior research has experimentally studied the role of the presence and accuracy of IDS alerts on attacker's and defender's decisions using a game-theoretic approach. However, little is known about the cognitive mechanisms that govern these decisions. METHOD: To investigate the cognitive mechanisms governing the attacker's and defender's decisions in the presence of IDSs of different accuracies, instance-based learning (IBL) models were developed. One model (NIDS) disregarded the IDS alerts and one model (IDS) considered them in the instance structure. Both the IDS and NIDS models were trained in an existing dataset where IDSs were either absent or present and they possessed different accuracies. The calibrated IDS model was tested in a newly collected test dataset where IDSs were present 50% of the time and they possessed different accuracies. RESULTS: Both the IDS and NIDS models were able to account for human decisions in the training dataset, where IDS was absent or present and it possessed different accuracies. However, the IDS model could accurately predict the decision-making in only one of the several IDS accuracy conditions in the test dataset. CONCLUSIONS: Cognitive models like IBL may provide some insights regarding the cognitive mechanisms governing the decisions of attackers and defenders in conditions not involving IDSs or IDSs of different accuracies. APPLICATION: IBL models may be helpful for penetration testing exercises in scenarios involving IDSs of different accuracies.

Theory of Mind From Observation in Cognitive Models and Humans.

A major challenge for research in artificial intelligence is to develop systems that can infer the goals, beliefs, and intentions of others (i.e., systems that have theory of mind, ToM). In this research, we propose a cognitive ToM framework that uses a well-known theory of decisions from experience to construct a computational representation of ToM. Instance-based learning theory (IBLT) is used to construct a cognitive model that generates ToM from the observation of other agents' behavior. The IBL model of the observer distinguishes itself from previous models of ToM that make unreasonable assumptions about human cognition, are hand-crafted for particular settings, complex, or unable to explain a cognitive development of ToM compared to human's ToM. The IBL model learns from the observation of goal-directed agents' behavior in a gridworld navigation task, and it infers and predicts the behaviors of the agents in new gridworlds across different degrees of decision complexity in similar ways to the way human observers do. We provide evidence for the alignment of the IBL observer's predictions under various levels of decision complexity. We also advance the demonstration of the IBL predictions using a classic test of false beliefs (the Sally-Anne test), which is commonly used to test ToM in humans. We discuss our results and the potential of the IBL observer model to improve human-machine interactions.

Learning and Dynamic Decision Making.

Humans make decisions in dynamic environments (increasingly complex, highly uncertain, and changing situations) by searching for potential alternatives sequentially over time, to determine the best option at a precise moment. Surprisingly, the field of behavioral decision making has little to offer in terms of theoretical principles and practical guidelines on how people make decisions in dynamic situations. My research program aims to fill in this gap by developing theoretical understandings of decision processes as well as practical demonstrations of how these theoretical developments can improve human dynamic decision making. Throughout my research career, I have helped create, test, and improve a general theory of dynamic decision making, instance-based learning theory, IBLT. The methods I have used to contribute to IBLT are (1) laboratory experiments that rely on dynamic games in which humans make choices over time and space, individually and in teams, and from which we extrapolate robust phenomena and behavioral insights; and (2) computational, actionable cognitive models, which specify the decision-making process and the cognitive mechanisms involved into a computational algorithm. The combination of these methods spawned novel applications in areas such as cybersecurity, phishing, climate change, and human-machine interactions. In this paper, I will take you through my own intellectual exploratory experience of computational modeling of human decision processes, and how the integration of experimental work and cognitive modeling helped in discovering and uncovering the field of dynamic decision making.

A Social Interpolation Model of Group Problem-Solving.

How do people use information from others to solve complex problems? Prior work has addressed this question by placing people in social learning situations where the problems they were asked to solve required varying degrees of exploration. This past work uncovered important interactions between groups' connectivity and the problem's complexity: the advantage of less connected networks over more connected networks increased as exploration was increasingly required for optimally solving the problem at hand. We propose the Social Interpolation Model (SIM), an agent-based model to explore the cognitive mechanisms that can underlie exploratory behavior in groups. Through results from simulation experiments, we conclude that "exploration" may not be a single cognitive property, but rather the emergent result of three distinct behavioral and cognitive mechanisms, namely, (a) breadth of generalization, (b) quality of prior expectation, and (c) relative valuation of self-obtained information. We formalize these mechanisms in the SIM, and explore their effects on group dynamics and success at solving different kinds of problems. Our main finding is that broad generalization and high quality of prior expectation facilitate successful search in environments where exploration is important, and hinder successful search in environments where exploitation alone is sufficient.

How to use a multicriteria comparison procedure to improve modeling competitions: A comment on Erev et al. (2017).

Modeling competitions are a promising method for advancing psychological science. In this commentary to Erev et al. (Psychological Review, 2017, 124, p. 369), we highlight how this promise could be enhanced through modifying competition structures to produce insights more directly in line with the goals of promoting psychological knowledge. We argue that a single criterion on which models is compared limits the diversity of models entered into competitions, restricting the number and type of insights that can be gained consequently. We propose an alternative competition structure with multiple evaluative criteria and outline a quantitative selection method for choosing a winner. Our proposed competition structure has the advantages of (a) increasing the diversity of models entered, (b) incentivizing desirable qualities of models, (c) disambiguating competition winners, and (d) enhancing the impact and possible insights gained from competitions, all these while allowing flexibility for competition organizers to emphasize specific qualities of models. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Towards a Cognitive Theory of Cyber Deception.

This work is an initial step toward developing a cognitive theory of cyber deception. While widely studied, the psychology of deception has largely focused on physical cues of deception. Given that present-day communication among humans is largely electronic, we focus on the cyber domain where physical cues are unavailable and for which there is less psychological research. To improve cyber defense, researchers have used signaling theory to extended algorithms developed for the optimal allocation of limited defense resources by using deceptive signals to trick the human mind. However, the algorithms are designed to protect against adversaries that make perfectly rational decisions. In behavioral experiments using an abstract cybersecurity game (i.e., Insider Attack Game), we examined human decision-making when paired against the defense algorithm. We developed an instance-based learning (IBL) model of an attacker using the Adaptive Control of Thought-Rational (ACT-R) cognitive architecture to investigate how humans make decisions under deception in cyber-attack scenarios. Our results show that the defense algorithm is more effective at reducing the probability of attack and protecting assets when using deceptive signaling, compared to no signaling, but is less effective than predicted against a perfectly rational adversary. Also, the IBL model replicates human attack decisions accurately. The IBL model shows how human decisions arise from experience, and how memory retrieval dynamics can give rise to cognitive biases, such as confirmation bias. The implications of these findings are discussed in the perspective of informing theories of deception and designing more effective signaling schemes that consider human bounded rationality.

Toward Personalized Deceptive Signaling for Cyber Defense Using Cognitive Models.

Recent research in cybersecurity has begun to develop active defense strategies using game-theoretic optimization of the allocation of limited defenses combined with deceptive signaling. These algorithms assume rational human behavior. However, human behavior in an online game designed to simulate an insider attack scenario shows that humans, playing the role of attackers, attack far more often than predicted under perfect rationality. We describe an instance-based learning cognitive model, built in ACT-R, that accurately predicts human performance and biases in the game. To improve defenses, we propose an adaptive method of signaling that uses the cognitive model to trace an individual's experience in real time. We discuss the results and implications of this adaptive signaling method for personalized defense.

Cognitive Models in Cybersecurity: Learning From Expert Analysts and Predicting Attacker Behavior.

Cybersecurity stands to benefit greatly from models able to generate predictions of attacker and defender behavior. On the defender side, there is promising research suggesting that Symbolic Deep Learning (SDL) may be employed to automatically construct cognitive models of expert behavior based on small samples of expert decisions. Such models could then be employed to provide decision support for non-expert users in the form of explainable expert-based suggestions. On the attacker side, there is promising research suggesting that model-tracing with dynamic parameter fitting may be used to automatically construct models during live attack scenarios, and to predict individual attacker preferences. Predicted attacker preferences could then be exploited for mitigating risk of successful attacks. In this paper we examine how these two cognitive modeling approaches may be useful for cybersecurity professionals via two human experiments. In the first experiment participants play the role of cyber analysts performing a task based on Intrusion Detection System alert elevation. Experiment results and analysis reveal that SDL can help to reduce missed threats by 25%. In the second experiment participants play the role of attackers picking among four attack strategies. Experiment results and analysis reveal that model-tracing with dynamic parameter fitting can be used to predict (and exploit) most attackers' preferences 40-70% of the time. We conclude that studies and models of human cognition are highly valuable for advancing cybersecurity.

Understanding Cumulative Risk Perception from Judgments and Choices: An Application to Flood Risks.

Catastrophic events, such as floods, earthquakes, hurricanes, and tsunamis, are rare, yet the cumulative risk of each event occurring at least once over an extended time period can be substantial. In this work, we assess the perception of cumulative flood risks, how those perceptions affect the choice of insurance, and whether perceptions and choices are influenced by cumulative risk information. We find that participants' cumulative risk judgments are well represented by a bimodal distribution, with a group that severely underestimates the risk and a group that moderately overestimates it. Individuals who underestimate cumulative risks make more risk-seeking choices compared to those who overestimate cumulative risks. Providing explicit cumulative risk information for relevant time periods, as opposed to annual probabilities, is an inexpensive and effective way to improve both the perception of cumulative risk and the choices people make to protect against that risk.

Creative Persuasion: A Study on Adversarial Behaviors and Strategies in Phishing Attacks.

Success of phishing attacks depend on effective exploitation of human weaknesses. This research explores a largely ignored, but crucial aspect of phishing: the adversarial behavior. We aim at understanding human behaviors and strategies that adversaries use, and how these may determine the end-user response to phishing emails. We accomplish this through a novel experiment paradigm involving two phases. In the adversarial phase, 105 participants played the role of a phishing adversary who were incentivized to produce multiple phishing emails that would evade detection and persuade end-users to respond. In the end-user phase, 340 participants performed an email management task, where they examined and classified phishing emails generated by participants in phase-one along with benign emails. Participants in the adversary role, self-reported the strategies they employed in each email they created, and responded to a test of individual creativity. Data from both phases of the study was combined and analyzed, to measure the effect of adversarial behaviors on end-user response to phishing emails. We found that participants who persistently used specific attack strategies (e.g., sending notifications, use of authoritative tone, or expressing shared interest) in all their attempts were overall more successful, compared to others who explored different strategies in each attempt. We also found that strategies largely determined whether an end-user was more likely to respond to an email immediately, or delete it. Individual creativity was not a reliable predictor of adversarial performance, but it was a predictor of an adversary's ability to evade detection. In summary, the phishing example provided initially, the strategies used, and the participants' persistence with some of the strategies led to higher performance in persuading end-users to respond to phishing emails. These insights may be used to inform tools and training procedures to detect phishing strategies in emails.