In touch with our emotions, finally: A shift in the science of decision making

  uch-with-our-emotions-finally-a-shift-in-the-science-of-decision-making.html
  Science and Technology on Mon, Aug 25th, 2025 by Phys.org

Emotions, not just evidence, steer scientific judgment – new study reveals

August 2025, Phys.org

A groundbreaking study published this month in the journal Nature Neuroscience shows that the emotional state of a researcher can meaningfully sway the conclusions they draw from data. The research – a collaboration between the University of Oxford, MIT, and the Max Planck Institute for Human Development – suggests that the long‑held assumption that science is a purely rational enterprise may need to be revised.

The experiment

The team recruited 112 professional scientists (physicists, biologists, social‑science researchers) and 88 advanced graduate students. Participants were presented with a set of statistical data sets – some clearly supporting a hypothesis, others that were borderline or ambiguous. Before analysing each data set, subjects were exposed to one of five emotional inductions: joy, sadness, anger, fear, or neutral. The emotional state was triggered by short video clips, music, or autobiographical recall, and participants’ affective responses were verified with self‑report scales and physiological measures (heart rate, galvanic skin response).

While viewing the data, participants performed a series of “decision‑making” tasks: they had to rate the strength of evidence, decide whether to accept or reject a null hypothesis, and suggest next experimental steps. Their responses were recorded verbatim, and the researchers later compared them with pre‑registered “objective” assessments made by a blind panel of experts.

The key finding: Participants in the joy and anger conditions were significantly more likely to favour a hypothesis that matched their pre‑existing beliefs, even when the data were objectively ambiguous. Those in the fear condition were more cautious, often over‑correcting and rejecting clear evidence. Surprisingly, the neutral group performed almost exactly as the blind panel, indicating that the effect was driven by emotion, not by any underlying cognitive deficiency.

The authors argue that these effects mirror well‑documented “affect‑influenced” biases in everyday decision making – the “affect heuristic” – but now provide the first systematic evidence that even seasoned scientists are not immune.

Why emotion matters in science

Historically, the “objective” nature of science has been upheld as a bulwark against personal bias. However, a growing body of social‑cognitive research has shown that emotion can shape attention, memory, and judgment. This study extends those insights to the laboratory setting.

Lead author Dr. Ellen Hsu, a cognitive neuroscientist at Oxford, explains: “We’re not suggesting that emotions undermine the scientific method. Rather, we’re highlighting that affect can alter the interpretation of data – an early, crucial step in the scientific workflow.”

Hsu cites a 2019 review by Psychological Review that identified emotion‑driven “confirmation bias” in medical decision‑making. The present study adds a neural dimension: fMRI scans revealed heightened activity in the amygdala and ventromedial prefrontal cortex when participants in the joy condition were presented with supportive data, correlating with a greater tendency to accept a hypothesis.

Implications for research culture

The findings have resonances beyond individual laboratories. Peer‑review panels, grant committees, and editorial boards are all made up of humans with emotions. If a reviewer is in a positive mood, they may be more generous in their assessment of a proposal’s novelty, whereas a reviewer in a negative or fearful state may be overly critical.

One commentator, Dr. Samuel Ortiz from the University of São Paulo, noted in a commentary linked to the article: “The reproducibility crisis may partly stem from a lack of awareness about affective influences on scientific judgment. Training scientists to recognize and manage their own emotions could be a simple, low‑cost intervention.”

Indeed, the study’s authors are already piloting “affect‑awareness” workshops in several institutions. In one pilot, researchers who attended a 2‑hour training on emotional regulation (including mindfulness and meta‑cognitive strategies) showed a 25 % reduction in bias when re‑analysing the same data sets.

The next steps

Hsu and colleagues plan to investigate whether certain fields are more susceptible to affective bias. Preliminary data suggest that social scientists may be more prone to emotional sway than hard‑science researchers, possibly because their work often deals with complex human behavior where emotion is an intrinsic component.

Another line of inquiry will explore whether the same pattern holds in large‑scale data science projects, where automated algorithms replace some human judgment. “Even algorithmic decision‑making can inherit human bias if the training data are curated by emotionally influenced researchers,” Hsu cautions.

The article ends with a call for a new sub‑discipline: “affective science of science,” which would systematically study how emotions shape every stage of research, from hypothesis generation to publication.

A broader perspective

The study intersects with broader discussions about the “human element” in science. As the Nature editorial board recently pointed out (link), acknowledging emotion does not diminish the rigor of science; it simply adds a layer of realism.

For those interested in the raw data and methodology, the authors provide open‑access code on GitHub (https://github.com/oxford-ehsu/affect-science) and invite replication studies worldwide.

Bottom line: While science strives for objectivity, this new research reminds us that the emotional contours of researchers can influence their judgments. By shining a light on this hidden variable, the study opens pathways to more transparent, reproducible, and ultimately more reliable science.