arXiv subjects

Matthew O. Jackson

Publications and source records attributed to Matthew O. Jackson.

At least 37 records · Page 2Linked to original sources

arXiv2025

Be.FM: Open Foundation Models for Human Behavior

Despite their success in numerous fields, the potential of foundation models for modeling and understanding human behavior remains largely unexplored. We introduce Be.FM, one of the first open foundation models designed for human behavior modeling. Built upon open-source large language models and fine-tuned on a diverse range of behavioral data, Be.FM can be used to understand and predict human decision-making. We construct a comprehensive set of benchmark tasks for testing the capabilities of behavioral foundation models. Our results demonstrate that Be.FM can predict behaviors, infer characteristics of individuals and populations, generate insights about contexts, and apply behavioral science knowledge.

cs.AI↗

arXiv2024

How Different AI Chatbots Behave? Benchmarking Large Language Models in Behavioral Economics Games

The deployment of large language models (LLMs) in diverse applications requires a thorough understanding of their decision-making strategies and behavioral patterns. As a supplement to a recent study on the behavioral Turing test, this paper presents a comprehensive analysis of five leading LLM-based chatbot families as they navigate a series of behavioral economics games. By benchmarking these AI chatbots, we aim to uncover and document both common and distinct behavioral patterns across a range of scenarios. The findings provide valuable insights into the strategic preferences of each LLM, highlighting potential implications for their deployment in critical decision-making roles.

cs.AI↗

arXiv2024

A Network Formation Model Based on Subgraphs

We develop a new class of random graph models for the statistical estimation of network formation -- subgraph generated models (SUGMs). Various subgraphs -- e.g., links, triangles, cliques, stars -- are generated and their union results in a network. We show that SUGMs are identified and establish the consistency and asymptotic distribution of parameter estimators in empirically relevant cases. We show that a simple four-parameter SUGM matches basic patterns in empirical networks more closely than four standard models (with many more dimensions): (i) stochastic block models; (ii) models with node-level unobserved heterogeneity; (iii) latent space models; (iv) exponential random graphs. We illustrate the framework's value via several applications using networks from rural India. We study whether network structure helps enforce risk-sharing and whether cross-caste interactions are more likely to be private. We also develop a new central limit theorem for correlated random variables, which is required to prove our results and is of independent interest.

physics.soc-ph↗

arXiv2024

Content Quality vs. Attention Allocation: An LLM-Based Case Study in Peer-to-peer Mental Health Networks

With the rise of social media and peer-to-peer networks, users increasingly rely on crowdsourced responses for information and assistance. However, the mechanisms used to rank and promote responses often prioritize and end up biasing in favor of timeliness over quality, which may result in suboptimal support for help-seekers. We analyze millions of responses to mental health-related posts, utilizing large language models (LLMs) to assess the multi-dimensional quality of content, including relevance, empathy, and cultural alignment, among other aspects. Our findings reveal a mismatch between content quality and attention allocation: earlier responses - despite being relatively lower in quality - receive disproportionately high fractions of upvotes and visibility due to platform ranking algorithms. We demonstrate that the quality of the top-ranked responses could be improved by up to 39 percent, and even the simplest re-ranking strategy could significantly improve the quality of top responses, highlighting the need for more nuanced ranking mechanisms that prioritize both timeliness and content quality, especially emotional engagement in online mental health communities.

cs.SI↗

arXiv2024

General Covariance-Based Conditions for Central Limit Theorems with Dependent Triangular Arrays

We present a general central limit theorem with simple, easy-to-check covariance-based sufficient conditions for triangular arrays of random vectors when all variables could be interdependent. The result is constructed from Stein's method, but the conditions are distinct from related work. We show that these covariance conditions nest standard assumptions studied in the literature such as $M$-dependence, mixing random fields, non-mixing autoregressive processes, and dependency graphs, which themselves need not imply each other. This permits researchers to work with high-level but intuitive conditions based on overall correlation instead of more complicated and restrictive conditions such as strong mixing in random fields that may not have any obvious micro-foundation. As examples of the implications, we show how the theorem implies asymptotic normality in estimating: treatment effects with spillovers in more settings than previously admitted, covariance matrices, processes with global dependencies such as epidemic spread and information diffusion, and spatial process with Matérn dependencies.

stat.ME↗

arXiv2024

A Turing Test: Are AI Chatbots Behaviorally Similar to Humans?

We administer a Turing Test to AI Chatbots. We examine how Chatbots behave in a suite of classic behavioral games that are designed to elicit characteristics such as trust, fairness, risk-aversion, cooperation, \textit{etc.}, as well as how they respond to a traditional Big-5 psychological survey that measures personality traits. ChatGPT-4 exhibits behavioral and personality traits that are statistically indistinguishable from a random human from tens of thousands of human subjects from more than 50 countries. Chatbots also modify their behavior based on previous experience and contexts ``as if'' they were learning from the interactions, and change their behavior in response to different framings of the same strategic situation. Their behaviors are often distinct from average and modal human behaviors, in which case they tend to behave on the more altruistic and cooperative end of the distribution. We estimate that they act as if they are maximizing an average of their own and partner's payoffs.

cs.AI↗

arXiv2023

Behavioral Communities and the Atomic Structure of Networks

When people prefer to coordinate their behaviors with their friends -- e.g., choosing whether to adopt a new technology, to protest against a government, to attend university -- divisions within a social network can sustain different behaviors in different parts of the network. We define a society's `behavioral communities' via its network's `atoms': groups of people who adopt the same behavior in every equilibrium. We analyze how the atoms change with the intensity of the peer effects, and characterize the atoms in a prominent class of network models. We show that using knowledge of atoms to seed the diffusion of a behavior significantly increases diffusion compared to seeding based on standard community detection algorithms. We also show how to use observed behaviors to estimate the intensity of peer effects.

physics.soc-ph↗

arXiv2023

Credit Freezes, Equilibrium Multiplicity, and Optimal Bailouts in Financial Networks

We analyze how interdependencies between organizations in financial networks can lead to multiple possible equilibrium outcomes. A multiplicity arises if and only if there exists a certain type of dependency cycle in the network that allows for self-fulfilling chains of defaults. We provide necessary and sufficient conditions for banks' solvency in any equilibrium. Building on these conditions, we characterize the minimum bailout payments needed to ensure systemic solvency, as well as how solvency can be ensured by guaranteeing a specific set of debt payments. Bailout injections needed to eliminate self-fulfilling cycles of defaults (credit freezes) are fully recoverable, while those needed to prevent cascading defaults outside of cycles are not. We show that the minimum bailout problem is computationally hard, but provide an upper bound on optimal payments and show that the problem has intuitive solutions in specific network structures such as those with disjoint cycles or a core-periphery structure.

cs.GT↗

arXiv2022

Selecting the Most Effective Nudge: Evidence from a Large-Scale Experiment on Immunization

Policymakers often choose a policy bundle that is a combination of different interventions in different dosages. We develop a new technique -- treatment variant aggregation (TVA) -- to select a policy from a large factorial design. TVA pools together policy variants that are not meaningfully different and prunes those deemed ineffective. This allows us to restrict attention to aggregated policy variants, consistently estimate their effects on the outcome, and estimate the best policy effect adjusting for the winner's curse. We apply TVA to a large randomized controlled trial that tests interventions to stimulate demand for immunization in Haryana, India. The policies under consideration include reminders, incentives, and local ambassadors for community mobilization. Cross-randomizing these interventions, with different dosages or types of each intervention, yields 75 combinations. The policy with the largest impact (which combines incentives, ambassadors who are information hubs, and reminders) increases the number of immunizations by 44% relative to the status quo. The most cost-effective policy (information hubs, ambassadors, and SMS reminders but no incentives) increases the number of immunizations per dollar by 9.1% relative to status quo.

stat.AP↗

arXiv2021

Interacting Regional Policies in Containing a Disease

Regional quarantine policies, in which a portion of a population surrounding infections are locked down, are an important tool to contain disease. However, jurisdictional governments -- such as cities, counties, states, and countries -- act with minimal coordination across borders. We show that a regional quarantine policy's effectiveness depends upon whether (i) the network of interactions satisfies a balanced-growth condition, (ii) infections have a short delay in detection, and (iii) the government has control over and knowledge of the necessary parts of the network (no leakage of behaviors). As these conditions generally fail to be satisfied, especially when interactions cross borders, we show that substantial improvements are possible if governments are outward-looking and proactive: triggering quarantines in reaction to neighbors' infection rates, in some cases even before infections are detected internally. We also show that even a few lax governments -- those that wait for nontrivial internal infection rates before quarantining -- impose substantial costs on the whole system. Our results illustrate the importance of understanding contagion across policy borders and offer a starting point in designing proactive policies for decentralized jurisdictions.

physics.soc-ph↗

arXiv2021

Centrality Measures in Networks

We show that prominent centrality measures in network analysis are all based on additively separable and linear treatments of statistics that capture a node's position in the network. This enables us to provide a taxonomy of centrality measures that distills them to varying on two dimensions: (i) which information they make use of about nodes' positions, and (ii) how that information is weighted as a function of distance from the node in question. The three sorts of information about nodes' positions that are usually used -- which we refer to as "nodal statistics" -- are the paths from a given node to other nodes, the walks from a given node to other nodes, and the geodesics between other nodes that include a given node. Using such statistics on nodes' positions, we also characterize the types of trees such that centrality measures all agree, and we also discuss the properties that identify some path-based centrality measures.

physics.soc-ph↗

arXiv2020

Systemic Risk in Financial Networks: A Survey

We provide an overview of the relationship between financial networks and systemic risk. We present a taxonomy of different types of systemic risk, differentiating between direct externalities between financial organizations (e.g., defaults, correlated portfolios and firesales), and perceptions and feedback effects (e.g., bank runs, credit freezes). We also discuss optimal regulation and bailouts, measurements of systemic risk and financial centrality, choices by banks' regarding their portfolios and partnerships, and the changing nature of financial networks.

q-fin.RM↗

arXiv2020

Learning through the Grapevine: The Impact of Noise and the Breadth and Depth of Social Networks

We examine how well people learn when information is noisily relayed from person to person; and we study how communication platforms can improve learning without censoring or fact-checking messages. We analyze learning as a function of social network depth (how many times information is relayed) and breadth (the number of relay chains accessed). Noise builds up as depth increases, so learning requires greater breadth. In the presence of mutations (deliberate or random) and transmission failures of messages, we characterize sharp thresholds for breadths above which receivers learn fully and below which they learn nothing. When there is uncertainty about mutation rates, optimizing learning requires either capping depth, or if that is not possible, limiting breadth by capping the number of people to whom someone can forward a message. Limiting breadth cuts the number of messages received but also decreases the fraction originating further from the receiver, and so can increase the signal to noise ratio. Finally, we extend our model to study learning from message survival: e.g., people are more likely to pass messages with one conclusion than another. We find that as depth grows, all learning comes from either the total number of messages received or from the content of received messages, but the learner does not need to pay attention to both.

physics.soc-ph↗

arXiv2020

Why understanding multiplex social network structuring processes will help us better understand the evolution of human behavior

Social scientists have long appreciated that relationships between individuals cannot be described from observing a single domain, and that the structure across domains of interaction can have important effects on outcomes of interest (e.g., cooperation).1 One debate explicitly about this surrounds food sharing. Some argue that failing to find reciprocal food sharing means that some process other than reciprocity must be occurring, whereas others argue for models that allow reciprocity to span domains in the form of trade.2 Multilayer networks, high-dimensional networks that allow us to consider multiple sets of relationships at the same time, are ubiquitous and have consequences, so processes giving rise to them are important social phenomena. The analysis of multi-dimensional social networks has recently garnered the attention of the network science community.3 Recent models of these processes show how ignoring layer interdependencies can lead one to miss why a layer formed the way it did, and/or draw erroneous conclusions.6 Understanding the structuring processes that underlie multiplex networks will help understand increasingly rich datasets, giving more accurate and complete pictures of social interactions.

econ.GN↗

arXiv2019

A Typology of Social Capital and Associated Network Measures

I provide a typology of social capital, breaking it down into seven more fundamental forms of capital: information capital, brokerage capital, coordination and leadership capital, bridging capital, favor capital, reputation capital, and community capital. I discuss how most of these forms of social capital can be identified using different network-based measures.

physics.soc-ph↗

arXiv2017

Diffusion in Networks and the Unexpected Virtue of Burstiness

Whether an idea, information, infection, or innovation diffuses throughout a society depends not only on the structure of the network of interactions, but also on the timing of those interactions. Recent studies have shown that diffusion can fail on a network in which people are only active in "bursts", active for a while and then silent for a while, but diffusion could succeed on the same network if people were active in a more random Poisson manner. Those studies generally consider models in which nodes are active according to the same random timing process and then ask which timing is optimal. In reality, people differ widely in their activity patterns -- some are bursty and others are not. Here we show that, if people differ in their activity patterns, bursty behavior does not always hurt the diffusion, and in fact having some (but not all) of the population be bursty significantly helps diffusion. We prove that maximizing diffusion requires heterogeneous activity patterns across agents, and the overall maximizing pattern of agents' activity times does not involve any Poisson behavior.

physics.soc-ph↗

arXiv2017

The Friendship Paradox and Systematic Biases in Perceptions and Social Norms

The "friendship paradox" (Feld1991) refers to the fact that, on average, people have strictly fewer friends than their friends have. I show that this over-sampling of the most popular people amplifies behaviors that involve complementarities. People with more friends experience greater interactive effects and hence engage more in socially influenced activities. Given the friendship paradox, people then perceive more engagement when sampling their friends than exists in the overall population. Given the complementarities, this feeds back to amplify average engagement. In addition, people with the greatest innate benefits from a behavior also tend to be the ones who choose to interact the most, leading to further feedback and amplification. These results are consistent with studies finding that people consistently overestimate peer consumption of alcohol, cigarettes, and drugs; and, can help explain problems with adolescent abuse of drugs and binge-drinking, as well as other behaviors. I also discuss how these results change in cases of strategic substitutes, where individuals overestimate free-riding by peers.

physics.soc-ph↗

arXiv2017

Pricing and Referrals in Diffusion on Networks

When a new product or technology is introduced, potential consumers can learn its quality by trying the product, at a risk, or by letting others try it and free-riding on the information that they generate. We propose a dynamic game to study the adoption of technologies of uncertain value, when agents are connected by a network and a monopolist seller chooses a policy to maximize profits. Consumers with low degree (few friends) have incentives to adopt early, while consumers with high degree have incentives to free ride. The seller can induce high-degree consumers to adopt early by offering referral incentives - rewards to early adopters whose friends buy in the second period. Referral incentives thus lead to a `double-threshold strategy' by which low and high-degree agents adopt the product early while middle-degree agents wait. We show that referral incentives are optimal on certain networks while inter-temporal price discrimination (i.e., a first-period price discount) is optimal on others, and discuss welfare implications.

econ.GN↗