Attention and Information Theoretic Data Engineering
Information theoretic data engineering measures and models what decision makers like, what they know, and why they don’t know more. It therefore hews a middle path between structural applied microeconomics, which treats information as perfect, and reduced form applied microeconomics, which focuses on behavioral biases. The key results are surprising in their simplicity and breadth of application. For example, Daniel Martin and I are applying them to more effectively value algorithms.
Rationally Inattentive Behavior: Characterizing and Generalizing Shannon Entropy
We introduce and behaviorally characterize invariant posterior separable and uniformly posterior separable attention cost functions. The Shannon cost function alone is both.
Comparison of Decisions Under Unknown Experiments
An econometrician wants to determine which of two experiments provides higher expected utility but only knows the decisions under each experiment. We provide a simple characterization.
Rational inattention, competitive supply, and psychometrics
We recover attention costs from choice data based on a precise analogy with production theory. Costs of attention determine consumer demand and consumer welfare, just as a competitive firm’s technology determines its supply curve and profits.
Rational inattention, optimal consideration sets, and stochastic choice
We show that rational inattention gives rise to consideration sets. We introduce linear inequalities that identify all priors consistent with each consideration set.
Wishful thinking
We model agents who get utility from their beliefs and therefore interpret information optimistically. They may exhibit excessive optimism, procrastination, confirmation bias, polarization, and the endowment effect.
Defaults and attention: the drop out effect
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Maecenas gravida pretium tristique. Vivamus venenatis ac velit quis bibendum. Vivamus commodo semper elit in cursus.
The Dual‐Process Drift Diffusion Model: Evidence from Response Times
We introduce a model of response time and choice in which rapid (automatic) decisions are qualitatively different from considered decisions. As our model predicts, decision times are bimodal, with automatic decisions of far lower quality than considered decisions.
Naive play and the process of choice in guessing games
We introduce a “choice process” protocol that allows us to separate naive and sophisticted decision makers Naive playrers appear to choose at random, while sophisticated ones show increased understanding as time passes.
Revealed preference, rational inattention, and costly information acquisition
To what extent do mistaken decisions reflect a trade-off between the costs and benefits of learning? We develop a revealed preference test that characterizes all patterns of choice “mistakes” consistent with this cost-benefit trade off.
A testable theory of imperfect perception
We introduce a method of revealing a Blackwell experiment from choice data and the no improving action switch (NIAS) characterization of Bayesian expected utility maximization.
Learning from market share when consumers are rationally inattentive
Combines social learning with rational inattention.
Rational inattention, entropy, and choice: The posterior-based approach
Introduces the posterior-based approach to rational inattention which has since become standard
Shannon Entropy and Rational Inattention: Solutions, Behavioral Regularities, and Extensions
Introduces uniformly posterior separable attention cost functions which have since become standard. Introduces Lagrangean methods of solution, locally invariant posteriors, and the invariant-likelihood characterization of optimal policies in the Shannon model.
Choice sets as percepts
Highlights the importance of developing and integrating data on how choice sets are perceived into the utility-maximizing framework.
Search and satisficing
Experimentally implements data on how choices change with contemplation time..Most subjects behave in line with a reservation-based model of sequential search, altering their reservation utilities in response to the size of the choice set and the complexity of the environment.
Search, choice, and revealed preference
We introduce data on the evolution of provisional choices with contemplation time and characterize sequential search and satisficing behavior in this data set.
Testing the reward prediction error hypothesis with an axiomatic model
We test the reward prediction error (RPE)theory of dopaminergic function, based on a recent axiomatization by Caplin and Dean (Quarterly Journal of Economics, 123 (2008), 663–702). We show that measurements of human neural activity from the striatum, medial prefrontal cortex, amygdala, and posterior cingulate cortex satisfy these conditions.
Measuring beliefs and rewards: a neuroeconomic approach
We test the reward prediction error (RPE) theory of dopaminergic function, based on a recent axiomatization by Caplin and Dean (Quarterly Journal of Economics, 123 (2008), 663–702). These tests are satisfied by neural activity in the nucleus accumbens, an area rich in dopamine receptors. We find evidence for separate positive and negative reward prediction error signals.
Fear as a policy instrument
Models the interaction between attention and fear, and policies that try to grab attention and thereby manipulate behavior (e.g. health messages)