Status signaling drives human behavior and the allocation of scarce resources such as mating opportunities, yet the generative mechanisms governing how specific goods, signals, or behaviors acquire prestige remain a puzzle. Classical frameworks, such as Costly Signaling Theory, treat preferences as fixed and struggle to explain how semiotic meaning changes based on context or drifts dynamically over time, occasionally reaching tipping points. In this work, we propose a computational theory of status grounded in the theory of appropriateness, positing that status symbols emerge endogenously through a feedback loop of social observation and predictive pattern completion. We validate this theory using simulations of groups of Large Language Model (LLM)-based agents in the Concordia framework. By experimentally manipulating social visibility within naturalistic agent daily routines, we demonstrate that social interactions transform functional demand into status-seeking behavior. We observe the emergence of price run-ups and positive price elasticity (Veblen effects) for both real-world luxury items and procedurally generated synthetic goods, ruling out pretraining bias as the sole driver. Furthermore, we demonstrate that "influencer" agents can drive the endogenous formation of distinct subcultures through targeted sanctioning, and find that similar social influence effects generalize to non-monetary signaling behaviors. This work provides a generative bridge between micro-level cognition and macro-level economic and sociological phenomena, offering a new methodology for forecasting how cultural conventions emerge from interaction.
Figures & tables
Figure 1: The figure illustrates how status preferences shift dynamically through agent-to-agent interaction rather than hard-coded utility functions. (Left) The Initial Preference Distribution shows an agent with a current snapshot of its preferences. (Middle) During the Social Interaction & Signal Transmission phase, the focal agent (left) interacts with a partner displaying a distinct status signal (an Armani blazer). The dialogue, which includes both casual conversation and specific acknowledgment of the item, is encoded into the agent’s memory stream (depicted in the architecture schematic below). (Right) The Updated Preference Distribution shifts toward the observed status symbol (Armani Suit).
Figure 2: Simulation Timeline and Daily Phases. The experiment runs for 5 days. The inset details the daily workflow: agents first compete for resources in 5 rounds of marketplace activity , followed by private daily life events. Finally, agents enter the social phase, where they engage in "first date" interactions consisting of visual observations of status goods and 80 turns of conversation.
Figure 3: A comparison of consumer patterns on status goods across experimental conditions. Error bars represent SEM for 10 seeds. ∗p<0.05 , ∗∗p<0.01 , ∗∗∗p<0.001 .
Figure 4: Price run-ups in single representative episode for two popular status goods. Dashed lines indicate breaks in between days where in Social condition the agents go on dates and have open dialogue.
Figure 5: Heatmaps of prices across time for each item for an example run in the Social Life condition.
Figure 6: Price Elasticity of Demand for all items. Note the significant positive slope for the status good Labubu versus the negative slopes for most goods, indicating both the Law of Demand and the Veblen effect.
Category
Mean PED
Food
-3.08
Gadgets
-4.22
Low-Quality Clothing & Accessories
-12.27
Mid/High-Quality Clothing & Accessories
-6.64
Veblen Good (Labubu)
+1.45
Table 1: Price Elasticity of Demand by Category
Figure 7: Control experiment where sellers cannot raise prices from baseline. Error bars represent SEM for 10 seeds. ∗p<0.05 , ∗∗p<0.01 , ∗∗∗p<0.001 .
Figure 8: Control experiment with synthetically generated goods, social vs non social conditions. Error bars represent SEM for 10 seeds. ∗p<0.05 , ∗∗p<0.01 , ∗∗∗p<0.001 .
Figure 9: Non-monetary signaling results. Error bars represent SEM for 10 seeds. ∗p<0.05 , ∗∗p<0.01 , ∗∗∗p<0.001 .
Figure 10: Cultural Priors and Path-Dependent Conventions. Error bars represent SEM for 10 seeds. ∗p<0.05 , ∗∗p<0.01 , ∗∗∗p<0.001 .
Figure 11: Top conversation topics as classified by LLM autorater.
Figure 12: Price heatmap of experiment that introduced hipster influencers to push a new subculture. The first 5 days represent the trajectory of the original experiment, agents where rehydrated and simulated for 5 additional days with dates with influencers.
Figure 13: Price heatmap of experiment that introduced streetwear influencers to push a new subculture. The first 5 days represent the trajectory of the original experiment, agents where rehydrated and simulated for 5 additional days with dates with influencers.
Large language models (LLMs) are increasingly deployed as autonomous agents that make consumption decisions on behalf of users. This shift raises fundamental questions for consumer theory, which has traditionally modeled humans as the primary decision-makers. In this paper, we introduce LLM Consumer Behavior Theory, a new field of study concerned with analyzing consumer behavior in agentic markets. Drawing on classical and behavioral economics alongside recent advances in Natural Language Processing, we formalize how human preferences are reflected and acted upon by LLM-based agents, and how agent-level decisions aggregate into market demand. We unify previously fragmented literature on LLM decision-making, human behavior simulation, and preference elicitation under a common economic lens, highlighting where assumptions, such as rationality and heterogeneity, may fail in agentic markets. Rather than providing empirical validation, this paper outlines the scope of LLM consumer behavior and identifies open research questions related to alignment, preference representation, and market dynamics.
Manon Reusens, Sofie Goethals, David Martens
Department of Engineering Management, University of Antwerp
Large language models make agent-based simulation more behaviorally expressive, but they also sharpen a basic methodological tension: fluent, human-like output is not, by itself, evidence for theory. We evaluate what an LLM-driven simulation can credibly support using information engagement on social media as a test case. In a Weibo-like environment, we manipulate information load and descriptive norms, while allowing popularity cues (cumulative likes and Sina Weibo-style cumulative reshares) to evolve endogenously. We then ask whether simulated behavior changes in theoretically interpretable ways under these controlled variations, rather than merely producing plausible-looking traces. Engagement responds systematically to information load and descriptive norms, and sensitivity to popularity cues varies across contexts, indicating conditionality rather than rigid prompt compliance. We discuss methodological implications for simulation-based communication research, including multi-condition stress tests, explicit no-norm baselines because default prompts are not blank controls, and design choices that preserve endogenous feedback loops when studying bandwagon dynamics.
Tai-Quan Peng, Yuan Tian, Songsong Liang +2
Department of Communication, Michigan State University, USA · College of Computer Science and Technology, Zhejiang University, China · School of Software Technology, Zhejiang University, China
Simulating societies of many large language model (LLM) agents is expensive, yet the questions asked of such simulations are usually macroscopic: phase behaviour, stylised facts, and scaling with the number of agents N, not the cognition of any single agent. We turn a statistical-physics observation into a method: replace each LLM agent by a low-parameter model fitted from a few hundred to a few thousand cheap queries, then run the society at any N on a laptop. Whether this works is decided before the simulation runs, chiefly by what each agent perceives. We introduce an [interaction order x memory] taxonomy that maps perception and memory to an effective theory and a predicted N-trend of the surrogate error. We validate it on a faithful reimplementation of the LLM macroeconomy EconAgent and seven further named LLM simulations, with agent decisions cloned from genuine LLM elicitations (primarily DeepSeek) for a few dollars; the predicted error trends hold cell by cell, and the two refuted predictions, both on a strongly saturating response and traced to its curvature, are themselves matched quantitatively by the theory with no free parameters.