Habits as Adaptations: An Experimental Study1
Ludmila Matyskova´2 Brian Rogers345
University of Bonn Washington University in St. Louis
Jakub Steiner§
University of Zurich and CERGE-EI
Keh-Kuan Sun
Washington University in St. Louis
May 18, 2020
Abstract
When observable cues correlate with optimal choices, habit-driven behavior can
alleviate cognition costs. We experimentally study the degree of sophistication in habit
formation and cue selection. To this end, we compare lab treatments that differ in the
information provided to subjects, holding fixed the serial correlation of optimal actions.
We find that a particular cue – own past action – affects behavior only in treatments in
which this habit is useful. The result suggests that caution is warranted when modeling
habits via a fixed non-separable utility. Despite this sophistication, lab behavior also
reveals myopia in information acquisition.
keywords: habit formation, rational inattention.
JEL codes: C91, D8, D9
1 Introduction
Habits play an important role in economic discourse. Economists employ them to explain
diverse phenomena ranging from inertia of consumption to brand loyalty. Often, the
standard modeling approach accounts for habits via the use of a fixed time-nonseparable
utility function, thus leaving the issues of when and why habits form, and their responses to
counterfactual environments, unaddressed. On the other hand, psychologists offer a view
on both the purpose of and the mechanism underlying habit formation. In this literature,
1
2
3
4
5
,habits are typically defined as automated responses triggered by cues, where cues are past
actions or other variables that empirically correlate with optimal continuation choices. In
this view, the purpose of habits is to alleviate cognition costs; see Andrews (1903), Lally et
al. (2010), and Wood & Neal (2007).
We explore the extent to which people’s behavior exhibits sophistication in habit
formation and cue selection in a simple setting. This issue is crucial for an assessment of the
common modeling approach to habits within economics. To the extent such sophistication
is observed, it suggests that habits, at least in some economic applications, should be
understood as endogenous, changing in predictable ways with the decision-making
environment. Our main experimental results largely support this position. Our subjects do
form a habit when available cues are predictive of the current optimal choice and they do
not form a habit otherwise, i.e., when the available cues cease to be informative guides for
current choices. When multiple informative cues are available, subjects select the most
informative one. We view these findings as suggesting that habit formation is a predictable
response to changes in the decision-making environment. An understanding of habit
formation rooted in optimization can inform analysts which cues, out of several available
cues, a decision-maker is likely to leverage. Modeling habits as optimal adaptations also
permits counterfactual predictions of habit strength under various policies.
To study sophistication in habit formation, we compare treatments from a lab
experiment in which subjects face a sequence of tasks generated by a given stochastic
process. The compared treatments differ only in the information feedback. If habits consist
of naive repetitions of past actions, then the variation in feedback should not impact habit
formation. Our data, however, show that subjects form habits with distinct cues across
these treatments; moreover, the cues selected are naturally rationalized as adaptations to
the information provided.
In our experiment, the basic task confronted by subjects is to recognize a binary state
variable presented visually on a computer screen. Correctly identifying the state requires
moderate cognitive effort.6 Each decision problem consists of a two-period sequence of this
state-recognition task, across which the state evolves according to a known stochastic
process with positive serial correlation. In the treatment without feedback, we reveal both
realized states to the subjects only at the conclusion of the two-period sequence. We find
that subjects form a habit in this treatment: the first-period outcome predicts the second-
period choice (controlling for the second-period state) in this treatment. The cue that
subjects leverage is their first-period action; the first-period state does not predict the
6 Identifying the state amounts to conducting a counting process, so that we can plausibly assume that frictions in
the cognitive process are the main source of errors.
1
, second-period action. In other words, the behavioral pattern exhibits action inertia. The
habit alleviates the subject’s cognitive burden since, due to the serial correlation of the
states, the firstperiod action contains useful information about the second-period optimal
choice, and the subjects utilize this information.
In the other treatment, with information feedback, we employ an identical state-
generating process, but we reveal the first-period state before the second period task, so
that subjects now know the first-period state before engaging in the second task. Subjects
again form a habit in this treatment: payoff-irrelevant elements of the history predict the
continuation choice (controlling for the second-period state). However, importantly, the
cue changes relative to the previous treatment. The first-period action is no longer
predictive; all of the predictive power is associated with the first-period state, which
contains superior information about the optimal continuation action relative to the first-
period action. This result suggests that our subjects demonstrate sophistication in the sense
that they select cues according to their informational content. As a further check, we ran
additional treatments (both with and without information feedback) in which the states
were serially independent, so that the first period history contains no cues that correlate
with the optimal second-period choice. As expected, subjects do not form habits in these
treatments; the second-period choice is independent of all first-period variables (for both
information treatments).
To the extent that habits are driven by optimal adaptations, their strength should vary
predictably with the parameters of the environment, in particular, with the incentive stakes
and the serial correlation of states. When stakes are decreased or correlation increased, the
trade-off between reliance on the cues and the acquisition of new information shifts in
favor of the cues. Thus, we predict that habits become stronger – cues become more
predictive of continuation behavior – when stakes are lower and correlation is greater. We
test this hypothesis experimentally. For the correlated treatments, changes in stakes and
correlation have no impact on the cue selection, but they do affect the strength of habits.
We obtain strong statistical evidence in favor of the predicted comparative statics when the
selected cue is the past action. When the cue is the past state, the evidence continues to
support the prediction, although it is less conclusive.
While, as discussed, some sophistication is observed in how subjects select and use cues
in making their second-period choices, we also find indirect evidence of myopia in the
information-acquisition process. When states are correlated and feedback is not provided
then information is more valuable in the first period relative to the second period, since
firstperiod information is useful in both periods. Consequently, a forward-looking decision-
maker should acquire more information in the first than in the second period, and this
2
Ludmila Matyskova´2 Brian Rogers345
University of Bonn Washington University in St. Louis
Jakub Steiner§
University of Zurich and CERGE-EI
Keh-Kuan Sun
Washington University in St. Louis
May 18, 2020
Abstract
When observable cues correlate with optimal choices, habit-driven behavior can
alleviate cognition costs. We experimentally study the degree of sophistication in habit
formation and cue selection. To this end, we compare lab treatments that differ in the
information provided to subjects, holding fixed the serial correlation of optimal actions.
We find that a particular cue – own past action – affects behavior only in treatments in
which this habit is useful. The result suggests that caution is warranted when modeling
habits via a fixed non-separable utility. Despite this sophistication, lab behavior also
reveals myopia in information acquisition.
keywords: habit formation, rational inattention.
JEL codes: C91, D8, D9
1 Introduction
Habits play an important role in economic discourse. Economists employ them to explain
diverse phenomena ranging from inertia of consumption to brand loyalty. Often, the
standard modeling approach accounts for habits via the use of a fixed time-nonseparable
utility function, thus leaving the issues of when and why habits form, and their responses to
counterfactual environments, unaddressed. On the other hand, psychologists offer a view
on both the purpose of and the mechanism underlying habit formation. In this literature,
1
2
3
4
5
,habits are typically defined as automated responses triggered by cues, where cues are past
actions or other variables that empirically correlate with optimal continuation choices. In
this view, the purpose of habits is to alleviate cognition costs; see Andrews (1903), Lally et
al. (2010), and Wood & Neal (2007).
We explore the extent to which people’s behavior exhibits sophistication in habit
formation and cue selection in a simple setting. This issue is crucial for an assessment of the
common modeling approach to habits within economics. To the extent such sophistication
is observed, it suggests that habits, at least in some economic applications, should be
understood as endogenous, changing in predictable ways with the decision-making
environment. Our main experimental results largely support this position. Our subjects do
form a habit when available cues are predictive of the current optimal choice and they do
not form a habit otherwise, i.e., when the available cues cease to be informative guides for
current choices. When multiple informative cues are available, subjects select the most
informative one. We view these findings as suggesting that habit formation is a predictable
response to changes in the decision-making environment. An understanding of habit
formation rooted in optimization can inform analysts which cues, out of several available
cues, a decision-maker is likely to leverage. Modeling habits as optimal adaptations also
permits counterfactual predictions of habit strength under various policies.
To study sophistication in habit formation, we compare treatments from a lab
experiment in which subjects face a sequence of tasks generated by a given stochastic
process. The compared treatments differ only in the information feedback. If habits consist
of naive repetitions of past actions, then the variation in feedback should not impact habit
formation. Our data, however, show that subjects form habits with distinct cues across
these treatments; moreover, the cues selected are naturally rationalized as adaptations to
the information provided.
In our experiment, the basic task confronted by subjects is to recognize a binary state
variable presented visually on a computer screen. Correctly identifying the state requires
moderate cognitive effort.6 Each decision problem consists of a two-period sequence of this
state-recognition task, across which the state evolves according to a known stochastic
process with positive serial correlation. In the treatment without feedback, we reveal both
realized states to the subjects only at the conclusion of the two-period sequence. We find
that subjects form a habit in this treatment: the first-period outcome predicts the second-
period choice (controlling for the second-period state) in this treatment. The cue that
subjects leverage is their first-period action; the first-period state does not predict the
6 Identifying the state amounts to conducting a counting process, so that we can plausibly assume that frictions in
the cognitive process are the main source of errors.
1
, second-period action. In other words, the behavioral pattern exhibits action inertia. The
habit alleviates the subject’s cognitive burden since, due to the serial correlation of the
states, the firstperiod action contains useful information about the second-period optimal
choice, and the subjects utilize this information.
In the other treatment, with information feedback, we employ an identical state-
generating process, but we reveal the first-period state before the second period task, so
that subjects now know the first-period state before engaging in the second task. Subjects
again form a habit in this treatment: payoff-irrelevant elements of the history predict the
continuation choice (controlling for the second-period state). However, importantly, the
cue changes relative to the previous treatment. The first-period action is no longer
predictive; all of the predictive power is associated with the first-period state, which
contains superior information about the optimal continuation action relative to the first-
period action. This result suggests that our subjects demonstrate sophistication in the sense
that they select cues according to their informational content. As a further check, we ran
additional treatments (both with and without information feedback) in which the states
were serially independent, so that the first period history contains no cues that correlate
with the optimal second-period choice. As expected, subjects do not form habits in these
treatments; the second-period choice is independent of all first-period variables (for both
information treatments).
To the extent that habits are driven by optimal adaptations, their strength should vary
predictably with the parameters of the environment, in particular, with the incentive stakes
and the serial correlation of states. When stakes are decreased or correlation increased, the
trade-off between reliance on the cues and the acquisition of new information shifts in
favor of the cues. Thus, we predict that habits become stronger – cues become more
predictive of continuation behavior – when stakes are lower and correlation is greater. We
test this hypothesis experimentally. For the correlated treatments, changes in stakes and
correlation have no impact on the cue selection, but they do affect the strength of habits.
We obtain strong statistical evidence in favor of the predicted comparative statics when the
selected cue is the past action. When the cue is the past state, the evidence continues to
support the prediction, although it is less conclusive.
While, as discussed, some sophistication is observed in how subjects select and use cues
in making their second-period choices, we also find indirect evidence of myopia in the
information-acquisition process. When states are correlated and feedback is not provided
then information is more valuable in the first period relative to the second period, since
firstperiod information is useful in both periods. Consequently, a forward-looking decision-
maker should acquire more information in the first than in the second period, and this
2
