Using minute-by-minute data from over 60,000 smart thermostats in households distributed across the United States, we analyze the persistence of energy consumption behaviors in response to external weather shocks. The analysis examines habitual behavior and provides insight into what affects long term change and what triggers the decision to reconsider one's passive choices. Our preferences for indoor temperatures demonstrate habituation to outdoor temperatures. This habituation is asymmetrical between positive and negative changes and non-linear at the extremes. While our indoor temperature preferences habituate to match small outdoor changes, our preferences revert to long term means in response to extreme temperature change. We also find people are more likely to make active choices when outdoor temperature is salient. Finally, we show there is heterogeneity in how preferences respond as a function of social norms, political preferences, and change costs. Results provide guidance on how conservation policies impact energy use–failure to understand the influence of habit on decision making can lead us to over-estimate the impact of short term policy nudges but underestimate the long run impact of small changes. Our results also inform how changing average temperatures and changing cultural attitudes may affect energy conservation behaviors.

Although first proposed in the database theory as properties of functional dependencies between attributes, Armstrong's axioms capture general principles of information flow by describing properties of dependencies between sets of pieces of information. This article generalizes Armstrong's axioms to a setting in which there is a cost associated with information. The proposed logical system captures general principles of dependencies between pieces of information constrained by a given budget.

The article investigates an influence relation between two sets of agents in a social network. It proposes a logical system that captures propositional properties of this relation valid in all threshold models of social networks with the same structure. The logical system consists of Armstrong axioms for functional dependence and an additional Lighthouse axiom. The main results are soundness, completeness, and decidability theorems for this logical system.