When Good Choices Feel Bad: The Brain's Dual Response
Choosing between two or more equally positive outcomes can feel like a contradiction: excitement about the possibilities alongside anxiety about deciding. A 2014 report on research led by Amitai Shenhav describes this combination in choices between desirable items. The findings suggest that positive anticipation and decision conflict can coexist rather than cancel each other out.
This is a useful starting point for considering the AI impact on human psychology. The study itself did not test artificial intelligence; it examined human participants making choices while researchers recorded brain activity. Keeping that distinction clear matters: neuroscience can inform discussion of decision support, but this one experiment does not show that AI can read feelings or resolve choice anxiety.
What the study observed
In one experiment, 42 people rated the desirability of more than 300 products using an auction-like procedure. They then saw pairs of products with different or similar values, made choices, and underwent functional magnetic resonance imaging (fMRI). After scanning, they reported how they felt before and during each choice; participants received one of their selected items at the end.
Choices between two highly valued products—for example, a digital camera and a camcorder—were associated with the most positive feelings and the greatest anxiety. This was compared with choosing between two low-valued items, or between items of different values. The result is not that every attractive choice causes distress. It is an observed pattern in this study, under its particular task conditions.
The scans showed activity in the striatum and prefrontal cortex, regions involved in decision-making. Within those regions, lower portions were more active when participants felt excited about being offered a choice, while activity in upper portions was strongly tied to anxiety. The report describes these as associations between activity and reported feelings, not proof that a scan can diagnose a person's emotion or that a single brain area produces it by itself.
The researchers proposed that parallel processes may help explain why excitement and anxiety can appear together. One process relates to evaluating what a person might receive; another relates to guiding action and working out how difficult the choice will be. A desirable outcome can therefore be appealing while the act of selecting among competing desirable outcomes remains demanding.
More options, conflict, and changing a choice
In a second fMRI experiment, participants were told how similarly they had valued the items before making each choice. The anxiety did not abate simply because they knew how little they might lose by choosing the less-valued item. A third experiment compared two options with six. Six options led to higher anxiety than two, particularly when all six were highly valued; positive feelings about being offered the choice were similar across those set sizes.
These results led the researchers to suggest that conflict in making the decision, rather than only the opportunity cost of the runner-up option, may contribute to anxiety. Opportunity cost refers to the value of the alternative not selected. The report notes that this cost should be the same regardless of how many options are presented, while anxiety rose with six options in the described experiment. Participants in that final study had unlimited time, unlike the 1.5-second limit in the first two, so time pressure was not the main source of anxiety in the reported results.
Participants also had a surprise opportunity at the end of each study to reverse earlier choices. Higher activity in the anterior cingulate cortex around the initial choice predicted whether the decision would later be reversed. Earlier work had associated this region with assessing conflict; the finding suggests some choices may have continued to feel unresolved after selection. People who reported more anxiety in daily life were also more likely to change their minds. This is a group-level observation, not a rule for predicting any individual's behavior.
AI impact on human psychology and decision support
What is AI in psychology? The phrase can refer broadly to using computational systems in psychological research or practice, including tools that analyze information or assist with tasks. That broad definition should not blur the evidence here: the cited study used fMRI and human self-reports, not an AI system. It offers a description of human choice feelings, not evidence about the effectiveness or safety of a particular AI product.
Can AI understand human psychology? An AI tool may process inputs and identify patterns, but this study gives no basis for claiming that a system understands a person's subjective experience. Brain activity and reported emotion were associated in the experiment; they were not presented as a simple code from which another observer can know exactly what someone feels. Any AI-assisted interpretation would need separate validation, and should not be treated as a substitute for a person's own account or professional judgment where that is needed.
The research does suggest questions worth asking when considering decision interfaces. If a person faces many attractive alternatives, presenting a larger set may feel more demanding, even when each option is appealing. A decision aid could be evaluated for whether it clarifies relevant differences or instead adds complexity. But the study did not test interface design, recommendation algorithms, or whether reducing options improves decisions. Those are hypotheses for future evaluation, not conclusions established by these experiments.
A practical human-centered lesson is modest: mixed feelings are not irrational evidence that a person chose badly. The participants could experience positive anticipation and anxiety at once, and the researchers related these responses to different patterns of activity. For an everyday decision, it may help to distinguish liking the available outcomes from finding the act of choosing difficult. That distinction does not prescribe a particular choice; it simply names two different parts of the experience.
What the findings do—and do not—tell us
The study helps explain why a choice among good outcomes can feel more complicated than an economic account focused only on what is gained and what is sacrificed. It links reported affect to activity in different portions of the striatum and prefrontal cortex, and reports that more options and later reversals were associated with aspects of conflict. It does not establish that these patterns apply identically to everyone, every kind of decision, or choices with major real-world consequences.
The article reports that the study, “Neural correlates of dueling affective reactions to win-win choices,” by Amitai Shenhav and Randy L. Buckner, was published in the Proceedings of the National Academy of Sciences on July 14, 2014 (doi:10.1073/pnas.1405725111). The underlying report described experiments with product choices. The authors suggested the work could help illuminate why consequential choices—such as choosing a college or job offer—may feel paralyzing for some people, while also noting that even trivial choices can be difficult.
For readers interested in artificial intelligence and human psychology, the key boundary is equally important: a study of human decision-making can motivate careful questions about technology, but it cannot by itself establish what AI understands or how AI affects mental health. Claims about those effects require evidence from research that directly evaluates the relevant AI systems and people using them.
Source: Neuroscience News: “Brain’s Dynamic Duel Underlies Win-Win Choices”
good choices feel bad