📈 economics

Information Greenhouse: Optimal Persuasion for Medical Test-Avoiders

This paper characterizes optimal communication strategies for doctors treating information-avoidant patients, demonstrating that when standard warnings deter testing, doctors should construct an "information greenhouse"—a committed post-test environment that reassures patients about the risks of non-treatment to encourage them to undergo diagnosis.

Original authors: Zhuo Chen

Published 2026-05-12

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Zhuo Chen

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are a doctor, and your patient is terrified of finding out if they are sick. They know that getting a test is the only way to get the right treatment, but the idea of hearing bad news is so painful that they would rather stay in the dark. They are avoiding the test not because it hurts physically or costs money, but because the uncertainty is their only comfort right now.

This paper asks: How should a doctor talk to this patient to get them to take the test and, if they are sick, to accept the treatment?

The doctor has a tricky balancing act. She needs to be a Warner (to make the patient realize that not treating the sickness is dangerous) and a Comforter (to make the patient feel safe enough to face the test in the first place).

Here is the breakdown of the paper's findings using simple analogies:

1. The Core Conflict: The "Scary Movie" vs. The "Safety Blanket"

Think of the medical test like a scary movie.

The Warning: If you tell the patient, "If you don't get treated, it could be very bad," they might finally agree to get treated. But if you say this before they take the test, they might be too scared to even walk into the theater.
The Reassurance: If you tell the patient, "Don't worry, even if you are sick, it might not be that bad," they might feel brave enough to take the test. But if they find out they are sick, they might think, "Well, it's not that bad, so I'll just skip the treatment."

The doctor has to decide when to use the scary movie script and when to use the safety blanket.

2. The Two Main Strategies

The paper finds that the best strategy depends entirely on how the patient reacts to fear.

Scenario A: The "Fear-Reacting" Patient

Some people are scared, but that fear makes them want to act. They think, "Oh no, this is bad! I need to fix it now!"

The Strategy: Warning-in-Advance.
The Analogy: The doctor says, "Look, if you have this disease and don't treat it, it's a disaster. So, let's get the test done immediately so we can fix it."
Why it works: Since the patient is already motivated by fear, the doctor doesn't need to hold back. She gives the scary truth before the test. Once the test is done, the patient is already convinced that treatment is necessary, so no more talking is needed.

Scenario B: The "Fear-Avoiding" Patient

This is the harder case. These patients are so scared of bad news that the thought of taking the test paralyzes them. They would rather live in ignorance than face the possibility of a diagnosis.

The Strategy: The Information Greenhouse.
The Analogy: Imagine a greenhouse. It's a controlled, warm, safe environment that protects plants from the harsh outside weather.
- Before the test: The doctor acts like a gardener. She says, "Let's just take the test. If the result is bad, I promise to explain it in a way that gives you hope. We will focus on the possibilities that things aren't as hopeless as they seem." She creates a "safe zone" of hope to get the patient to walk through the door.
- After the test (if the result is bad): This is the tricky part. The doctor must keep her promise. Even though she wants to scream, "You are very sick, you must treat this immediately!" she holds back. Instead, she gives a "reassuring" explanation that softens the blow.
- Why it works: The patient only agrees to the test because they trust the doctor will protect them from the emotional shock after the fact. The doctor sacrifices some of the "treatment motivation" (by being less scary) to gain the "testing motivation" (by being comforting).

3. The "Voluntary Consultation" Twist

What if the patient is so scared they won't even come into the doctor's office to talk?

The Problem: The doctor can't use the "Information Greenhouse" (the promise of future comfort) to get them in the door if they won't listen to the promise in the first place.
The Solution: Precautionary Comfort.
The Analogy: If the patient won't enter the house, the doctor has to stand outside and hand them a warm blanket before they even step inside.
- The doctor must provide the reassurance immediately at the very start. She says, "I know you are scared. Let me tell you right now that even in the worst case, there is hope."
- Once the patient feels safe enough to come in and listen, the doctor stops talking about hope and focuses on the facts.
- The Cost: Because the doctor had to give away the "hope" so early to get the patient to listen, she has less "hope" left to use later to convince the patient to accept treatment. It's a trade-off: you get them in the door, but it's harder to convince them to take the medicine later.

4. The "Commitment" Problem

The paper highlights a crucial requirement: The doctor must be able to keep her word.

In the "Information Greenhouse" scenario, the doctor promises to be gentle after the test. But once the test is done and the patient is sick, the doctor's natural instinct is to be harsh to ensure the patient gets treated.
If the doctor cannot commit to being gentle later (if the patient thinks, "She's just saying that to get me to take the test, but she'll scare me later"), the patient will never take the test. The "Greenhouse" only works if the patient believes the doctor will actually stick to the plan.

Summary

If fear makes people act: Tell them the scary truth before the test.
If fear makes people hide: Promise them a "safe explanation" after the test (The Greenhouse) to get them to take the test.
If fear keeps them away entirely: Give them the comfort before they even enter the room (Precautionary Comfort), even if it makes the later conversation about treatment harder.

The paper concludes that there is no "one size fits all." The best way to communicate depends entirely on whether the patient's fear pushes them forward or holds them back, and whether they are willing to listen to the doctor at all.

Technical Summary: Information Greenhouse: Optimal Persuasion for Medical Test-Avoiders

Problem Statement
This paper addresses the communication problem between a benevolent doctor and an information-avoidant patient. The patient faces a decision to undergo a fully informative, costless medical test to determine if they are sick ( $\theta=0$ ) or healthy ( $\theta=1$ ). If sick, the patient must subsequently decide whether to undergo treatment, which incurs a physical cost $c$ but increases the probability of a healthy final outcome.

The core friction is information avoidance: the patient possesses anticipatory utility, meaning they derive emotional value (or disvalue) from beliefs formed before uncertainty is resolved. Specifically, the patient is modeled with a concave utility function over probabilities, reflecting a preference for hope and a dislike of the psychological pain associated with bad news. Consequently, a patient may refuse a medically beneficial test to avoid the anticipated emotional distress of a potential positive diagnosis.

The doctor's objective is to maximize the patient's physical health probability (the probability of a healthy final outcome). The doctor cannot force testing or treatment but can design the information environment. This involves two stages of disclosure:

Ex ante disclosure: Information provided before the patient decides whether to take the test. This concerns the severity of the "untreated prospect" (the probability of remaining healthy without treatment, conditional on being sick).
Interim disclosure: Information provided after a bad test result but before the treatment decision.

The central tension is between warning (disclosing that non-treatment is severe to induce treatment) and reassurance (softening the outlook on non-treatment to make the prospect of testing psychologically bearable).

Methodology
The paper employs a dynamic Bayesian persuasion framework with a psychological receiver.

Model Structure: A two-stage game. At Date 1, the doctor commits to an information policy (a standard operating procedure). The patient then chooses whether to test ( $a \in \{0, 1\}$ ). If tested and found sick, the doctor provides interim disclosure. At Date 2, the patient chooses treatment ( $e \in \{0, 1\}$ ).
Preferences: The patient's utility follows the psychological expected utility model (Caplin and Leahy, 2001). The payoff at Date 1 is $\phi(v_2)$ , where $v_2$ is the expected physical payoff at Date 2 and $\phi$ is a strictly increasing, concave function. Concavity of $\phi$ captures information avoidance (diminishing marginal utility of hope). The doctor maximizes the expected physical health probability $E[\eta]$ , ignoring the patient's anticipatory emotions and treatment costs.
Information Structure: The doctor controls the disclosure of the parameter $p$ (the health probability without treatment if sick), which takes values in $\{p_H, p_L\}$ . The medical test itself reveals $\theta$ perfectly. The doctor's signals are distributions over posterior beliefs about $p$ .
Solution Concept: Perfect Bayesian Equilibrium. The analysis utilizes the belief-based approach to Bayesian persuasion (Kamenica and Gentzkow, 2011), characterizing optimal signals via the concavification of the sender's value function subject to the receiver's participation constraints.

Key Contributions and Results

1. The "Information Greenhouse" vs. "Warning-in-Advance"
The optimal policy depends on the patient's behavioral response to fear (whether fear triggers action or avoidance) and the nature of the participation constraint.

Fear-Reactive Patients: If the patient is "fear-reactive" (i.e., sufficiently pessimistic beliefs about the untreated prospect induce them to take the test), the optimal policy is Warning-in-Advance. The doctor front-loads the relevant risk information before the testing decision. The warning is strong enough to ensure that if the patient is found sick, they will accept treatment. No further interim disclosure is needed. The testing constraint is satisfied by the pessimism induced ex ante.
Fear-Avoidant Patients: If the patient is "fear-avoidant" (i.e., a warning would make the test psychologically unbearable), the optimal policy constructs an Information Greenhouse.
- Mechanism: The doctor commits to a post-test information environment that reassures the patient about the untreated prospect conditional on a bad result. Specifically, the interim signal is "reassurance-oriented": it provides "perfect bad news" (fully revealing the worst untreated prospect) with some probability, but pools the remaining probability mass into a less conclusive, more optimistic posterior.
- Trade-off: This reassurance lowers the anticipated emotional cost of a bad diagnosis, making the patient willing to take the test. However, it comes at the cost of weakening treatment incentives (since the patient is less convinced that non-treatment is fatal). The doctor accepts a lower probability of treatment conditional on testing to preserve the probability of testing itself.
- Commitment: This policy requires dynamic commitment. Without it, the doctor would re-optimize after a bad result to maximize treatment incentives (switching to a warning), which the patient would anticipate, leading to test refusal.

2. Voluntary Consultation (Ex Ante Participation)
The paper extends the model to allow the patient to refuse the consultation entirely before any disclosure.

Fear-Reactive: The optimal policy remains warning-in-advance, but the ex ante signal must be sufficiently informative to satisfy the participation constraint.
Fear-Avoidant: If the patient is fear-avoidant, no information policy can persuade them to enter the consultation if the fear is too strong. If persuasion is possible, the optimal policy shifts to Precautionary Comfort. The doctor moves the reassurance-oriented disclosure to the ex ante stage. The doctor commits to a reassuring environment upfront to make the consultation acceptable, remaining silent afterwards. This is a "modified information greenhouse" where the emotional reward is front-loaded.

3. General Information Preferences
The analysis is extended to cases where the utility function $\phi$ is not globally concave but inverse S-shaped (concave for low values, convex for high values). This captures preferences for positive skewness and prior dependence.

In this setting, the "motivation-available" region (where testing is induced) can include high beliefs (optimism) because the patient intrinsically values the resolution of uncertainty or the skewness of the signal.
The warning-versus-reassurance logic persists, but the optimal signals may involve richer posterior structures (e.g., trinary signals) to exploit the patient's intrinsic love for information in specific belief regions.

4. Physical Cost Benchmark
The paper contrasts the psychological cost model with a benchmark where testing has a physical cost but no anticipatory utility costs (linear $\phi$ ). In this benchmark, the Information Greenhouse disappears. The doctor never softens post-test disclosure for reassurance; the optimal policy is always warning-in-advance (or no disclosure). This confirms that the Information Greenhouse is a specific response to the psychological cost of information, not merely the cost of testing.

Significance and Claims
The paper claims to characterize the optimal communication strategy for benevolent senders dealing with information-avoidant receivers. Its primary theoretical contribution is identifying the Information Greenhouse as a distinct mechanism where a sender commits to a sub-optimal (from an ex post perspective) information environment to secure participation.

The paper argues that medical communication is not merely about transmitting accurate information but about designing an information environment that balances the willingness to be tested against the willingness to be treated. It highlights that dynamic commitment is crucial: without the ability to commit to a reassuring post-test environment, fear-avoidant patients may never enter the testing process, leading to worse health outcomes.

The results suggest that in contexts where information itself is a source of disutility, "precautionary comfort" or "information greenhouses" can be welfare-improving for the patient's physical health, even if they involve withholding full information or providing reassurance that is technically "less informative" than the truth. The paper concludes that these policies are reference points for evaluating real-world medical communication protocols, acknowledging that ethical and procedural constraints may limit their implementation.