20 Ethics

The way that we do scientific research has impact on the people around us. Some of these impacts are positive: for example, our research might prompt the development of new treatments for clinical disorders, or instigate improved educational practices in schools, or spawn new technologies for helping people discover new music. However, some of these impacts might conversely be negative: certain experiments might cause psychological distress in our participants, might violate their privacy in untoward ways, or might even cause detriment to people’s physical health. The goal of research ethics is to systematically evaluate both these positive and negative potential impacts, and thereby guide our research programmes to maximise the positive and minimise the negative.

Ethics is relevant to many aspects of scientific and non-scientific research. In the context of ‘music and science’, our ethical best practices are mostly inherited from best practices in psychological research. As in psychology, the ethical issues we most commonly have to worry about are issues to do with the participants who take part in our experiments, most commonly humans but also sometimes animals. Ethical issues do also crop up in other areas though, for example when we are constructing musical corpora that deal with music from different cultures.

20.1 Guiding principles

Psychologists across the world are expected to guide their work towards certain ethical principles. One popular codification of these principles has been established by the American Psychological Association (APA) and can be browsed on their website. These principles can be briefly summarised as follows:

Beneficence and Nonmaleficence. We should strive to achieve good and to avoid doing harm.

Fidelity and Responsibility. We should act professionally and responsibly, being aware of the trust that society puts in us.

Integrity. We should be honest and truthful when we communicate to participants and the wider public.

Justice. We should try to conduct research that benefits all people equally.

Respect for People’s Rights and Dignity. We should respect people’s rights and dignity, recognising in particular vulnerable populations who have limited capacity for autonomous decision-making.

20.2 What this means in practice

These principles have certain salient implications for the way that we conduct music psychology research projects.

20.2.1 Good versus harm

Medical research often has to balance significant positive outcomes with significant potential negative outcomes. An experimental drug might have the potential to cure cancer, but simultaneously have the potential to cause dangerous side effects.

Music psychology rarely has to deal with such high stakes. Much of our research has no direct positive outcome on individuals; its primary impact is rather just a general improvement of our foundational knowledge of psychology. Fortunately, our research rarely has much of a negative outcome either: many of our experiments simply involve listening to a series of musical sounds and making some responses, a situation where the most significant side-effect is slight boredom.

When justifying the ethical appropriateness of music psychology studies, we therefore generally do not have to worry about arguing that our research has significant direct positive impact for participants: the stakes are typically low, and so a general contribution to knowledge is sufficient. However, we do have to keep an eye out for situations where harm might come about. Examples include:

• Loud stimuli that might cause hearing damage.
• Long-term repetitive movements that might cause physical problems such as repetitive strain injury.
• Negative reactions to drugs (e.g. from excessive consumption of alcohol).
• Negative reactions to neuroimaging techniques (e.g. transcranial magnetic stimulation).

One strategy is to eliminate the possibility of such negative effects altogether, for example by avoiding running any research using drugs or neuroimaging techniques. However, it is often possible to retain these elements but instead implement measures either for reducing the probability of adverse effects or by mitigating their severity if they do occur. For example, an experiment involving drugs might establish a standardised procedure for obtaining medical assistance should an adverse reaction occur.

20.2.2 Fairness in interventional research

Some music psychology studies involve long-term interventions. For example, one might implement an educational intervention where a random selection of students receive six months of music lessons. This intervention might be considered to be a ‘positive’ intervention in the sense that it involves a ‘gift’ to certain participants. In this context, it is important to reflect on the ‘Justice’ principle outlined above. Does your experiment differentially benefit particular subgroups of the population? Does it reinforce existing disparities in privilege that already exist in that population? If so, it is worth considering whether the experiment might be adjusted to achieve a more equitable outcome.

20.2.3 Privacy

Many legal systems around the world include the notion of a ‘right to privacy’. Privacy means that an individual has control over the use of their personal information. In the UK and Europe, the right to privacy is upheld by the General Data Protection Regulation (GDPR), which establishes a range of principles constraining the use of personal information by third parties. Institutions are under a serious obligation to upheld the requirements of the GDPR, and can be subject to very large fines for violations of it. When we conduct psychology experiments, we must likewise work hard to make sure that we adhere to the GDPR’s requirements.

The GDPR is complex, and learning its requirements typically involves taking not insignificant training courses. These requirements involve standardised procedures and expectations for how personal data is accessed, shared, and deleted. For example, the GDPR restricts the transfer of personal data outside the European Union to countries which do not satisfy its own levels of privacy guarantees (e.g. the US). Such principles can seem at the face of it at odds with other scientific goals such as the desire to share one’s datasets alongside academic publications.

Fortunately, there are some basic steps we can take that can simplify all of this for us in the majority of cases (note: I am not a lawyer, this is not legal advice…):

Avoid requesting sensitive personal data. EU law considers the following personal data to be ‘sensitive’, and hence subject to specific processing conditions. If you can, avoid collecting such data, that way you don’t have to worry about these additional conditions.

• Personal data revealing racial or ethnic origin, political opinions, religious or philosophical beliefs;

• Trade-union membership;

• Genetic data, biometric data processed solely to identify a human being;

• Health-related data;

• Data concerning a person’s sex life or sexual orientation.

Avoid requesting personally identifying information unless you need to. Datasets only become subject to GDPR regulations if they contain personally identifying information. It is therefore helpful to avoid collecting such information where possible. Personally identifying information is information that could theoretically be used to identify individual participants, for example:

• Names

• Dates of birth

• Postcodes

• Images or video recordings which include the participant’s face

• Audio recordings of the participant’s voice

• Mechanical Turk Worker IDs, Prolific Participant IDs, etc.

Non-anonymous datasets should only be stored on local, secure machines. Datasets that can be linked to individuals via personally identifying information may be termed ‘non-anonymous datasets’. Sometimes it is necessary to store non-anonymous datasets temporarily for practical purposes, for example administering payments to participants. In such cases, you should:

• Only store your data on a local machine (e.g. a laptop), or a machine known to be in the EU (e.g. a remote web server). Do not allow that data to pass outside the EU while it is still non-anonymous.

• Make sure your machine is secure from intrusion. This means a few best practices: don’t leave your computer unattended, give it a secure password, make sure the hard drive is encrypted. Only named collaborators should have access to the data.

• Institute a workflow which ensures that such information is deleted once it has served its required use. You should not retain such data indefinitely.

Anonymise your datasets as soon as possible. To anonymise data means to remove all of its personally identifying information (names, dates of birth, postcodes etc.). Once a dataset is anonymous, the GDPR rules no longer apply. This means you can then relax your practices about data management, for example uploading it to open-access repositories.

20.2.4 Informed consent

Participants should provide informed consent before participating in your study. For consent to be ‘informed’, the participant must have been told about all ethically relevant aspects of the study, including potential positive impacts, potential negative impacts, what will happen to them in the experiment, etc. This consent should be recorded in a concrete and standardised manner, ideally by signing a consent form, or through some kind of computerised substitute (e.g. ticking a checkbox).

Certain populations are considered to be less able to provide autonomous consent. Such populations include children, prisoners, people with cognitive impairments, and so on. Applications to work with such populations generally require a higher level of ethical scrutiny.

20.2.5 During the experiment

The experiment itself should be designed to minimise discomfort or harm to the participant. It is expected that experimenters will generally follow certain principles:

• Participants should be able to withdraw at any time without penalty (financial, reputational, or otherwise).

• Participants should not be compelled to answer questions that they feel uncomfortable answering.

• Experimenters should avoid using deception. Experiments involving deception are possible but require a higher level of ethical scrutiny.

20.2.6 Debriefing

It is good practice to debrief your participant at the end of the experiment, giving them a concise overview of your research project and how they have contributed to it. A nice way to do this is by giving people a quick verbal summary as well as a paper handout that they can take away and read at their leisure.

20.2.7 Institutional ethics approval

Before starting data collection you must submit an overview of your project to an appropriate ethics body for approval. Most universities have in-house processes for this; you should familiarise yourself with your own university’s systems (for Cambridge’s resources, see this link).

Most institutions have different tiers of review depending on the nature of the project, with fast-track reviews for projects that have minimal ethical implications, and more intensive reviews for projects with more serious ethical implications. Fast-track reviews can take just a few days, whereas intensive reviews can take months. However, timelines are never completely predictable, and it’s usually worth making sure you submit your applications sooner rather than later.

20.2.8 Reporting in publications

When you write up your experiment in a dissertation or a publication you should note that your project received institutional ethics approval, stating both the identity of the approving body and any relevant approval number. Such a statement is mandatory for many scientific venues nowadays.