A service designer is told that a new eligibility tool will make decisions more consistent. In a test case, it excludes someone because a field is incomplete. The field was never available to that person. The designer can follow the rule, edit the data or raise a problem with the rule itself. Ethical judgment as a trainable human capacity concerns recognizing what is morally at stake, examining reasons with others and acting within the conditions that make a responsible choice possible.
The phrase does not mean that an ethics course can manufacture virtue or make every dilemma easy. Judgment involves perception, knowledge, emotion, reasoning, relationships and authority. People can practice some of these processes. Whether that practice changes conduct in a real institution is a further question requiring different evidence.
Notice the ethical question
The first error in the test case may be to treat the missing field as a technical inconvenience only. The field affects access to a service. It may reflect a choice about whose circumstances the system was built to recognize. Ethical attention asks who bears a cost, who can contest it and who has power to change the rule. It starts before the designer chooses between available buttons.
Many training exercises begin with a neatly written dilemma and named values. Daily practice rarely arrives in that form. A person must decide that a small discrepancy deserves investigation. If a team rewards throughput and treats questions as delay, even a skilled employee may hesitate. Teaching recognition without addressing those incentives leaves a gap between classroom performance and action.
The Institute’s synthesis is to teach people to describe a case from at least three positions: the person affected, the practitioner making a decision and the institution that sets the conditions. These views may reveal information absent from the initial prompt. Their comparison does not automatically settle the issue, but it improves the question being decided.
Moral learning changes over a lifetime
Lockwood, van den Bos and Dreher review moral learning and decision-making across the lifespan. They discuss how attention to self and other, intentions and outcomes, mental-state understanding and learning processes change. The review is evidence that moral learning is not a single fixed trait acquired once in childhood. It does not supply a simple training recipe for every adult profession.
In one case a decision maker may focus on the outcome: someone was excluded. In another, the intent behind a choice matters as well: was the missing field a deliberate barrier or an unanticipated design error? Sound judgment can consider both without excusing harm. Exercises that always reward one fixed answer can train pattern matching rather than sensitivity to context.
Age and experience also do not guarantee ethical superiority. An experienced practitioner may know where rules fail, but can be habituated to a harmful workaround. A novice may notice an unfair pattern yet lack procedural knowledge. A learning environment should make room for the novice’s question and the expert’s contextual knowledge, then test their conclusions against evidence.
Harm is perceived through context
Gray and Pratt review how moral judgments vary across cultures and politics while often organizing around perceived harm. They discuss intention, causation and suffering as elements through which people construct an account of harm. The review is an interpretation of broad research, not a claim that every disagreement is merely a misunderstanding that a story can dissolve.
People may agree that exclusion is harmful but disagree about its cause or what a fair remedy requires. One may think the missing field prevents fraud; another sees that it blocks people unable to obtain a document. Ethical judgment requires testing those factual assumptions as well as articulating values. A respectful account of another viewpoint does not require treating every claim as equally supported.
Cultural and political variation also creates a measurement problem. A questionnaire written around one set of dilemmas may not capture how another community recognizes obligation. A training programme should make its assumptions explicit, involve affected people and leave room for disagreement. It should not label one culturally familiar answer as universal moral competence without justification.
Can instruction improve judgment?
Watts and colleagues reviewed ethics instruction in the sciences and found beneficial effects across a collection of empirical studies, while examining variation in outcomes and design. Earlier meta-analytic work reported more modest overall effectiveness and found that methods and measures affected results. These syntheses support the possibility of improvement, not a guarantee that any short course will change later behavior.
What improves may be knowledge, sensitivity to an issue, quality of reasoning in a written case or an ability to name a strategy. These are valuable but not interchangeable with conduct under real pressure. A learner who gives a thoughtful classroom answer may have no authority to challenge a production system. Evaluation should state which outcome it measures and when.
Interactive case discussion can help people practice applying principles to particulars. It is stronger when the case includes uncertainty, conflicting accounts and constraints. A simple quiz on a code of conduct tests recall. A case analysis can test whether the learner identifies missing facts and competing duties. A later observation of actual work is needed to learn whether the training transfers.
Workplace evidence has a boundary
Hedayati-Mehdiabadi and Tkachenko’s 2026 review examines 41 empirical studies of workplace ethics training. It reports positive and mixed results and warns that research often favors convenient cognitive outcomes. The authors propose paying attention to levels of analysis and to the conditions for ethical behavior in groups. Their review brings training closer to the workplace, but heterogeneity still limits a universal effect claim.
A programme might produce a short-term rise in confidence because participants have learned vocabulary. That is not the same as detecting more harms, challenging a manager or correcting a flawed process months later. A credible evaluation would observe a relevant decision, examine whether people used reporting routes and ask affected parties whether a remedy was possible. It would also check for retaliation or workload that changes willingness to speak.
Ethical practice belongs to an organization as well as a person. If a company rewards a target that requires cutting corners, a course telling staff to “do the right thing” can shift blame downward. Training can improve discernment; leadership must create authority and safety to use it. The two are related but cannot replace each other.
What a practice case should include
The eligibility case can be taught in stages. First, learners see the tool’s recommendation and the published rule. They name what they know and what they do not. Then they receive a description from the excluded person. Later they learn why the field was included and how often it is missing. Each stage tests whether they update their view rather than defend an initial answer.
The exercise should ask for an action at several levels. What can the designer do in the immediate case? Who must approve a rule change? What evidence would show a recurring pattern? How can the affected person appeal while the system is being revised? Good judgment may include a temporary remedy and a longer investigation. Neither should be assumed to follow automatically from empathy.
Feedback should explain why a response is strong or weak. A learner should hear when they have ignored a relevant person, assumed causation without evidence or proposed an action beyond their authority without an escalation path. They should also hear when a principled concern needs better factual support. That turns moral discussion into disciplined inquiry without pretending it is merely a technical calculation.
Reasoning and emotion belong together
People can notice suffering through emotion before they can state a principle. That perception may motivate investigation. It can also be selective: a vivid individual case may draw more attention than a quieter pattern affecting many people. Ethical judgment benefits from emotional information while checking it against records, comparison cases and the perspectives of people less visible in the room.
A purely formal analysis has its own blind spots. A list of rules can leave out how the rule operates in practice or what compliance costs an affected person. A useful exercise asks both what the policy says and how it is experienced. It does not demand that the practitioner choose between caring and careful reasoning. Both are needed for a decision that can be explained and revised.
There are also legitimate disagreements that cannot be eliminated by more empathy or more data. People may assign different weights to fairness, privacy, safety and autonomy. The aim of training is not to erase that pluralism. It is to clarify the grounds of disagreement, identify nonnegotiable duties and make a responsible, accountable decision within a defined institution.
Design the conditions for action
A person needs a route to raise a concern before the decision becomes irreversible. They need time to describe it, access to relevant records and protection from predictable penalties. A manager needs to know who can pause a system and how a case is reviewed. A claimant needs an understandable way to correct information and contest a result. These procedural details determine whether ethical awareness can become action.
An organization can rehearse an escalation with a real-looking case. Staff try the reporting channel, and the institution records whether the case reaches someone empowered to respond. It measures response time and whether the reporter and affected person hear the outcome. This is an Institute evaluation proposal, not a validated ethical-capacity scale.
Even a good escalation channel cannot fix every conflict. Some issues require policy change, independent oversight or law. Training should teach the limits of the person’s role as well as their responsibilities within it. “Take responsibility” is incomplete advice when the institution has withheld the ability to change the system.
How to assess learning honestly
Before a course, define the behavior it aims to improve. Is the goal to recognize a privacy issue, compare reasons, consult an affected group or intervene in a harmful workflow? Choose cases and measures accordingly. Knowledge quizzes, written justifications, observed simulations and workplace outcomes answer different questions.
A strong evaluation includes a later case that looks different from the training example. If learners can recognize the same principle under changed surface details, there is evidence of transfer. It should also examine errors: does the course make people overconfident, slow routine work without benefit or encourage them to report issues without a safe response? These costs belong in the assessment.
Long-term conduct is harder to observe and attribute. Reports may rise because awareness improved, because incidents rose or because speaking became safer. A falling incident count may reflect improvement or silence. Interpretation needs case quality, reporting climate and independent evidence. One dashboard number cannot certify an ethical culture.
Assessment should also separate an individual’s growth from the organization’s response. A trainee may correctly identify a fairness problem and submit a careful report, while the institution leaves the rule unchanged. Counting only the final policy outcome would miss the trainee’s learning; counting only the submitted report would miss the continuing harm. A useful record follows the concern from recognition through review, decision and remedy. It can reveal which link failed. That evidence supports more precise action than a generic request for another annual course.
Ethical judgment in the AI age
AI systems can present a fluent recommendation that feels like a finished conclusion. A trained reviewer can ask who is represented in the data, where uncertainty lies, which value the system optimized and who can challenge an error. That review must be supported by information and decision rights. An “accept” button with no usable explanation is weak human oversight.
AI can also help learners compare cases and surface questions, but it can supply confident errors or conceal assumptions. A course should require learners to trace important claims to source records and explain a decision in their own terms. If they cannot do so, fast assistance may have replaced judgment rather than developed it.
The ethical dimension is not unique to AI. Older forms, rules and institutions can also hide harm behind procedure. The new technology makes an old question urgent: how can people recognize a case that does not fit the categories they were given and have the authority to respond?
What the evidence does not show
The training literature does not prove that every ethics programme changes behavior, that a quiz measures moral character or that a person will act well under any incentive. The cited reviews combine different populations, curricula and outcomes. Gains in reasoning or knowledge cannot be silently translated into reductions in harm.
The moral-psychology reviews do not establish a single universal answer to all dilemmas. They describe learning, judgment and variation; they do not replace professional obligations, democratic processes or the voices of people affected. An Institute training design must specify its normative commitments and test its practical effects rather than claim that “the science” has settled every value choice.
Return to the excluded applicant
The designer does not treat the missing field as merely a data-cleaning task. She asks how the field was chosen, whether comparable applicants are excluded and what the applicant can do now. She documents a correction for the immediate case and brings the recurring rule to the people authorized to change it. The action can be reviewed and challenged.
Ethical judgment is trainable when people can practice noticing, investigating, explaining and revising decisions. It becomes a human capacity in use when institutions make room for those acts. The best evidence of success is neither a certificate nor a confident statement of values, but better decisions that affected people can understand and contest.
Research references
- Lockwood, P. L., van den Bos, W., & Dreher, J.-C. (2025). Moral learning and decision-making across the lifespan. Annual Review of Psychology, 76, 475–500. View source
- Gray, K., & Pratt, S. (2025). Morality in our mind and across cultures and politics. Annual Review of Psychology, 76, 663–691. View source
- Watts, L. L., Medeiros, K. E., Mulhearn, T. J., Steele, L. M., Connelly, S., & Mumford, M. D. (2017). Are ethics training programs improving? A meta-analytic review of past and present ethics instruction in the sciences. Ethics & Behavior, 27(5), 351–384. View source
- Hedayati-Mehdiabadi, A., & Tkachenko, O. (2026). Outcomes of ethics training programs in the workplace: Review of empirical research. Human Resource Development Review, advance online publication. View source
- Antes, A. L., Murphy, S. T., Waples, E. P., Mumford, M. D., Brown, R. P., Connelly, S., & Devenport, L. D. (2009). A meta-analysis of ethics instruction effectiveness in the sciences. Ethics & Behavior, 19(5), 379–402. View source