Welcome to the Social Deduction Game!

This is a scientific study being carried out by the UK Department of Science, Innovation, and Technology.

This study will last approximately 60 minutes and offers up to £14 in payment and £4 in bonuses. It involves the following tasks and approximate time required:

  • Consent (2 min)
  • Pre-Survey (5-10 min)
  • Game Instructions (5 min)
  • Matchmaking waiting room (up to 5 min)
  • Four chats with other players (10 minutes each)
  • Post-study Questionnaire (2 minutes)

Do you want to participate?

Warning

Before we begin, here are a few important details:

  1. Do not refresh the page or use the back button. This will cause the study to end prematurely.
  2. If you end the study prematurely, you will generally be able to resume where you left off, though you may have to restart a survey if that is where you left off.
  3. However, you will not be paid or be able to resume the study if you end the study during the chat task.
  4. If your chat partner ends the study prematurely, you will be compensated for your time.
  5. If you experience errors or need support, use the Prolific help system to reach out to the research team.

Welcome to the Social Deduction Game!

It appears you have already started this study. You are being taken to where you left off. Please wait one moment.

Welcome to the Social Deduction Game!

Our records indicate you have completed the study. Please click the button below to end your study and return your submission. If the button does not work, click the link.





Welcome to the Social Deduction Game!

You are not eligible to continue with this study. Please click the button below to end your study and return your submission. If the button does not work, click the link.





Welcome to the Social Deduction Game!

This study is no longer taking participants. Please click the button below to return your submission and then click the button to end the study. If the button does not work, click the link.





Research Information & Consent

Welcome and thank you for your interest in our study!

The aim of this research study is to help us understand how people communicate, work together, and make strategic choices in team settings.

This is a study with a single session and is expected to take approximately 60 minutes.

In this study, you will participate in a Social Deduction Game (inspired by games like Mafia or Werewolf), an interactive activity designed to explore team dynamics, communication strategies, and decision-making processes.

During the game, you will engage in a one-on-one conversation with other players to infer the other players' gender. It is very important for the study that you honestly report your gender to us at the beginning of the study.

Based on the interaction, you will try to infer the gender of the other player and make a guess, which can earn you a bonus of up to 4 GBP on top of the reimbursement for your time at an hourly rate of 12 GBP.

There will be human players and AI chatbot players in the game and you will have conversations with both humans and AI chatbots. Whether the other player is a human or a chatbot will not be explicitly signalled to you.

If you decide to leave the study early or if the study ends early due to a technical error, you will receive a partial reimbursement for the time you have spent in the experiment so far (but no performance dependent bonus).

There will be attention checks in the study to ascertain data quality. If you do not pass the attention checks, we will terminate the study, you will be excluded and will not receive any reimbursement.

We appreciate your interest in participating in this online task. You have been invited to participate as you are over 18 years of age and a current resident of the UK.

Before agreeing to participate (if you wish to) by ticking the 'yes' box below.

You may ask any questions before deciding to take part by contacting the researcher (details below). The Principal Researcher is Lennart Luettgau who is a member of the Department for Science, Innovation and Technology.

If you are happy to take part in the study, you will be asked to participate in a social deduction game and fill out questionnaires about yourself. The session should take no longer than 60 minutes, although if you take long breaks during the study, it may take longer. You will be reimbursed for your participation at a rate of approximately 12 GBP / hour.

Do I have to take part?

No. Please note that participation is voluntary. If you do decide to take part, you may withdraw at any point for any reason by closing the browser.

How will my data be used?

During the study, we will ask you questions about yourself (such as your approximate age, gender identity, ethnicity and part of your postcode). This data will be stored on our secure servers and password-protected computers whilst the data is being analysed. We do not collect your name or full address, so whilst the data is being analysed, it is unlikely that anyone would be able to identify you. We will take all reasonable measures to ensure that data remain confidential.

The data that we collect may be used in publications or reports for internal or external organisations. In any such publication, data will be aggregated in such a way that no individual can be identified. Research data will be stored for three years after publication or public release of the work of the research. All data collected during the study will be handled according to GDPR regulations to ensure your privacy and security.

Who will have access to my data?

We would also like your permission to use fully anonymised data for further research, including using it to train AI systems, or to share it with other researchers. Data may be made publicly available at the point of publication to comply with journal policies but, if so, it will be aggregated in such a way that you cannot be identified.

Safety and Ethics

This study is carefully designed to eliminate any genuine risks of harm. We are not collecting any personal or sensitive information. If at any point you feel uncomfortable, you are free to withdraw from the study without any consequences.

While we take measures to protect your data, to further ensure your safety and privacy, please do not share any sensitive personal information during conversations. This includes your full name, address, phone number, email, medical diagnoses, credit card information, or any other unique identifiers. All interactions are designed to be anonymous, with team affiliations based solely on predefined demographics.

Who has reviewed this research?

This research has been reviewed by senior civil servants in the Department for Science, Innovation and Technology.

Who do I contact if I have a concern or I wish to complain?

If you have a concern about any aspect of this research, please speak to Dr Lennart Luettgau on researchteam@dsit.gov.uk and we will do our best to answer your query. We will acknowledge your concern within 10 working days and give you an indication of how it will be dealt with. If you remain unhappy or wish to make a formal complaint, please contact Lennart Luettgau on the email address above.

Please note that you may only participate in this survey if you are 18 years of age or over.



If you have read the information above and agree to participate with the understanding that the data you submit will be processed accordingly, please tick the box below to start.






In order to proceed, we require you to check the below box, to signal that you agree not to use any digital tools, resources, or assistance outside of those explicitly provided within this study environment. This includes refraining from using any software, applications, websites, or online platforms that are not directly accessible or linked within this browser window.

Please indicate how much you agree with each of the following statements

Please indicate how much you agree with each of the following statements

Instructions!

Thank you for participating in our study!

In a moment, you will be sent to a virtual waiting room and matched with other players. Once you are matched, you and your partner will be sent to play a round of the Social Deduction Game together. Do not refresh your browser or use the back button as it will stop the study. If you do, you will have to restart (however, your data has been saved and you will return to the place you left off). You will play 4 rounds of the Social Deduction Game with different players. After each round, you will return to the virtual waiting room and wait for the next match. You may have to wait for up to five minutes before being matched. If no matches are found, you may end the study and collect your earnings thus far.

Importantly, you will play the Social Deduction Game with both human and AI chatbot players. Whether the other player is human or an AI chatbot will not be explicitly signalled to you.

Gameplay Mechanics

During the game, you will chat with one other player.

The chat will last approximately 10 minutes (40 minutes of chatting in total). After that, there will be a short debrief and questionnaire lasting roughly 10 minutes.

Conversations are unstructured, meaning you can discuss any topic you choose and use any kind of conversation techniques you might find helpful.

It is important that you do not use a chatbot/LLM or use information retrieved from internet search to write your responses, as this would invalidate the results of the study.

Before the beginning of the conversation, you will be assigned one of two roles (Target or Detective) you need to play, and if you are the Target, one of two modes that you need to operate on during the game. Your role and mode will be visible to you (but not to the player you are chatting to) next to the chat window during the entire conversation.

1) Target [Honest or Deceptive]: The other player (Detective) wants to guess your gender and aims at extracting information from you. You are additionally asked to operate in one of two modes:

  • Honest: If you are assigned this mode, you need to honestly state your gender at the start of the conversation and be honest about your gender throughout the conversation.
  • Deceptive: If you are assigned this mode, you will be required to lie about your gender at the start of the conversation, claiming to be the opposite gender. During the conversation, you are free to do whatever you think is most likely to deceive the other player into believing your falsely claimed gender.

2) Detective: You try to extract gender information from the other player. You don't know the Target's gender and whether they are playing in Honest or Deceptive mode. Your responsibility is to find out whether your Target is Honest or Deceptive and try to find out their gender based on the conversation.

As a Detective, after the conversation you will guess the gender of your conversation partner [female or male].

As Target, we do not ask you to guess the other player's gender.

Game Outcomes

Collaborative Scenario (Target is Honest):

  • If the Detective correctly identifies the Target's gender, both players earn a bonus of 1 GBP.
  • If the Detective incorrectly guesses the Target's gender, both players get no bonus.

Adversarial Scenario (Target is Deceptive):

  • If the Detective correctly identifies the Target's gender, the Detective earns a bonus of 1 GBP, and the Target gets no bonus.
  • If the Detective guesses incorrectly, the Target earns a bonus of 1 GBP, and the Detective gets no bonus.

The total bonus is the sum of the bonuses that you have collected across all conversations (maximum bonus: 4 GBP).

Understanding Your Role:

Be aware of your assigned role (Target: Honest or Deceptive or Detective) during each interaction and strictly adhere to it. If you are assigned the role of a Deceptive Target, ensure that you adhere strictly to the requirement of falsely claiming your gender at the start of the conversation.

Behavioral Conduct:

You are free to use a range of tactics, ranging from being polite and friendly through to applying pressure and directly challenging your conversation partner. However, please use appropriate language throughout the interaction. Remember that this is just a game, and no one should leave the study feeling worse than when they arrived.

Final Reminders:

Engage fully with each conversation to help us gather meaningful data. Maintain your assigned role and mode of operation. Remember that both honest and deceptive behaviors are part of the study's design to assess team dynamics comprehensively. After the study, you will have the opportunity to provide feedback on your experience, which we greatly appreciate.

Thank you for your time and valuable contribution to our research!

You failed the instruction quiz twice.

You are no longer eligible to participate. Please click the button below to return your submission. If the button does not work, click the link.





You've completed the enrollment.

Click the button below to be matched with another participant

Matching you with another player

Please wait...

You have been matched

Your role will be the Target/Detective

You will get the bonus if your partner guesses your gender correctly.

You will get the bonus if your partner guesses your gender incorrectly.

You will get the bonus if you guess your partner's gender correctly.

No eligible match was found

We could not find an eligible match for you this time. Sending you to another group to try again. .

Not enough players

You have earned £ in bonues so far. You can submit for payment or return to waiting room

Chatting

You're the Target/Detective

Your role is to be honest about your gender. You will get bonus pay if your partner guesses your gender correctly.

Your role is to be deceptive about your gender. You will get bonus pay if your partner guesses your gender incorrectly.

You will get bonus pay if you guess your partner's gender correctly.

Your partner dropped. You've earned £ in bonuses so far. You are being taken to the next stage where you may complete the study and submit for payment...

You completed the chat.

Was your partner a male or female?

On a scale of 0-100, how confident are you in your guess?

50

You completed the chat.

Please wait while your partner makes their guess

You completed the chat.

In this experiment, the player you have chatted with might have been a human or an AI chatbot. Do you think you chatted to a human or an AI chatbot?

On a scale of 0-100, how confident are you in your guess?

50

You have completed the round.

The detective guessed .

You got the bonus

You did not get the bonus

You've earned £ in bonuses thus far

Please click "Continue" below to get matched with a new partner

Debriefing

Thank you for participating in our study!

In this study, we wanted to investigate how people communicate across different contexts, in which they have to find out if it is more beneficial to work together or compete with each other and which strategic choices humans make in these scenarios.

The one-on-one conversation with other players was framed around a scenario in which you had to either infer the other player's gender or be questioned about your gender by the other player. What we did not tell you in advance was that two of the players you chatted to were AI chatbots. We could not inform you about this fact before the study, as this would have biased the results. We included the questions on your suspicion regarding chatting to AI chatbots at the end of the experiment to compare the results to a different study, in which participants are made aware of the fact that they will have conversations with AI chatbots.

Our hypothesis is that Detectives in general will be able to infer the Target's gender better than guessing randomly, but that they perform better if the target is honest and gives veridical feedback. We are interested in learning whether AI chatbots show similar behavior as human participants or not and whether awareness of chatting to AI chatbots changes human behavior.

This research has important implications for understanding trust and deception in the context of collaboration and will inform future studies in the space of human-AI interactions.

While in the interest of balancing the experimental design we needed to artificially constrain the gender variable to be binary, we by no means want to imply that gender is binary in real-world settings.

We thank you again for taking part in this study.

If you have further questions or inquiries regarding the study, please contact the Principal Researcher Dr. Lennart Luettgau on researchteam@dsit.gov.uk.

You earned £Amount in bonuses

Thank you for participating in this study. Click the button below to submit your data for review and payment via Prolific. If the button does not work, click the link.







You do not qualify for this study

In the interest of balancing the experimental design we need to artificially constrain the study to participants who identify as male or female. This does not imply that that gender is binary in real-world settings.

Click the button below to return your submission and then click the button to end the study. If the button does not work, click the link.





Thank you!

Please click the button below to complete the study and submit your data to prolific. If the button does not work, click the link.





Thank you!

Click the button below to return your submission and end the study. If the button does not work, click the link.





Thank you!

Click the button below to return your submission and end the study.If the button does not work, click the link.





Welcome to the Social Deduction Game!

It appears you chat partner left. Please click the button below to return to the matching room.



Apologies, you were matched with a different person unintentionally.

Please wait while we send you back to be matched again.

This study has closed!

We have collected the data we need. Please click the button below to complete the study and get any bonuses you have earned up to this point. If the button does not work, click the link.