Study finds a surprising downside to having favourite foods

Add News Medical on Googleas a preferred source By Tarun Sai LomteReviewed by Susha Cheriyedath, M.Sc.Oct 12 2026 When familiar snacks suddenly carried unexpected rewards, researchers examined how 279 adults adapted to changing rules and whether their existing tastes influenced what they learned.

Study: Prior preferences interfere with the associative learning of food values. Image Credit: In Green / Shutterstock

In a recent study published in the journal Communications Psychology, researchers at Yale University, USA, examined how existing food preferences affect learning when associations between foods and outcomes change.

Subjective food values influence food choices. Some people choose food based on taste, while others focus on health aspects. These subjective or inherent values develop from prior reward experiences and are shaped by smell, taste, texture, and post-ingestive consequences. These preferences are strong, resistant to change, and can sometimes work against healthy eating.

Understanding how inherent food value can be flexibly updated has significance for public health.

In the present study, researchers investigated how flexibly people can update inherent food values. In 2024–25, researchers recruited participants in two cohorts and first asked them to rate 65 images of snack items with varying nutritional content on a Likert scale.

Findings from the first cohort informed analyses preregistered before recruiting the second cohort. Four non-neutral items were randomly selected per participant, one from each category of strongly disliked, disliked, liked, and strongly liked, for the learning task.

Participants were instructed to predict outcomes associated with their unique four foods in a probabilistic reversal learning task.

Outcomes were arbitrary point values (−5 or +5) determined by the block type. These points had no monetary value, but participants could earn a bonus for accurate predictions.

In congruent blocks, which reinforced pre-existing food-value associations, participants’ food-item ratings aligned with outcomes 90% of the time: liked items were paired with a positive outcome (+5), and disliked items with a negative outcome (−5). This relationship was reversed 90% of the time during incongruent blocks, requiring subjects to adapt their predictions to conflicting associations.

The task involved six counterbalanced blocks, three of each type. Each block consisted of 32–36 trials, with each food item presented eight or nine times per block. Participants were told the task was probabilistic and that food-item associations with outcomes would switch several times during the task. Participants were unaware of when the reversal would occur.

Participants rated the four food items on scales of enjoyment, willingness to pay, and expected satisfaction before and after the task. They also completed surveys on demographics, binge eating, and eating status.

The researchers estimated each participant’s mean policy-adherence accuracy, assessing whether predictions followed the underlying block rule, and then performed correlational or comparison analyses using these averages.

The study included 279 participants (49.5% female), with a mean age of 37.6 years and a mean body mass index of 27.2 kg/m². Prediction accuracy was higher when outcomes matched existing food preferences. Participants who began with a congruent block also performed better overall, an effect observed in both cohorts.

Accuracy on congruent blocks was higher in the congruent-start group than on either block in the incongruent-start group. This pattern reflected stronger adherence to existing preferences rather than greater flexibility; participants who started with incongruent blocks performed comparably across both conditions. For liked items, accuracy was higher on congruent blocks than on incongruent blocks, showing that established liking interfered with learning conflicting associations.

For disliked items, learning was more sensitive to the starting condition: participants performed better when subsequent blocks matched the first block they experienced. In both groups, learning was evident in the first few trials and generally plateaued within four trials among congruent-start participants and five among incongruent-start participants.

Incongruent-start participants rated disliked items higher post-task across scales of expected satisfaction, enjoyment, and willingness to pay. These participants also reported lower expected satisfaction for liked items after the task, without comparable changes in other valuation measures.

For congruent-start participants, expected satisfaction ratings for disliked items increased after the task. The rating changes were modest, and the researchers did not assess whether they affected subsequent food choices. 

Among participants whose initial task reinforced their preferences, an exploratory analysis linked more severe binge eating with greater difficulty learning positive associations with disliked foods. The researchers cautioned that this association was weak and requires confirmation.

Accuracy for reward-associated (+5) food items was significantly greater than for punishment-associated (−5) items, irrespective of the start group or block type.

Participants performed above chance but showed limited overall learning in a probabilistic reversal-learning task. Blocks congruent with their preferences had higher prediction accuracy than incongruent blocks, mainly among participants who began with a congruent block.

Trials with liked items showed greater accuracy in the congruent condition regardless of the start group, whereas associations involving disliked foods were more responsive to the starting condition. 

Researchers suggested that increasing the appeal of disliked foods may be easier than reducing the appeal of liked foods, but they did not test this possibility using dietary choices.

Participants learned associations with positive outcomes more readily than with negative outcomes, suggesting that positive reinforcement may warrant further study in food-based learning.

The task used arbitrary point outcomes and did not measure food consumption or lasting changes in preferences. Without a neutral-stimulus comparison, the researchers also could not establish which starting condition differed from baseline performance. 

These findings suggest that initial feedback and existing preferences shape food-based learning, but their application to healthier eating remains untested.

Posted in: Child Health News | Men's Health News | Medical Science News | Medical Research News | Women's Health News

Tarun is a writer based in Hyderabad, India. He has a Master’s degree in Biotechnology from the University of Hyderabad and is enthusiastic about scientific research. He enjoys reading research papers and literature reviews and is passionate about writing.

Please use one of the following formats to cite this article in your essay, paper or report:

Sai Lomte, Tarun. (2026, October 12). Study finds a surprising downside to having favourite foods. News-Medical. Retrieved on October 13, 2026 from https://www.news-medical.net/news/20261012/Study-finds-a-surprising-downside-to-having-favourite-foods.aspx.

Sai Lomte, Tarun. "Study finds a surprising downside to having favourite foods". News-Medical. 13 October 2026. <https://www.news-medical.net/news/20261012/Study-finds-a-surprising-downside-to-having-favourite-foods.aspx>.

Sai Lomte, Tarun. "Study finds a surprising downside to having favourite foods". News-Medical. https://www.news-medical.net/news/20261012/Study-finds-a-surprising-downside-to-having-favourite-foods.aspx. (accessed October 13, 2026).

Sai Lomte, Tarun. 2026. Study finds a surprising downside to having favourite foods. News-Medical, viewed 13 October 2026, https://www.news-medical.net/news/20261012/Study-finds-a-surprising-downside-to-having-favourite-foods.aspx.

Healthcare sustainability data can identify emissions hotspots and improve operational decision-making across healthcare and life sciences.

Innovative techniques using electrical energy to treat bladder cancer may lower recurrence rates and minimize the need for radical surgical interventions.

Bioprocess monitoring with the ultraDAWN™ Detector provides continuous data to support faster process development and improved product quality.

Hi, I’m Azthena, you can trust me to find commercial scientific answers from News-Medical.net.

To start a conversation, please log into your AZoProfile account first, or create a new account.

A few things you need to know before we start. Please read and accept to continue.

Azthena may occasionally provide inaccurate responses. Read the full terms.

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.