When you first smell a fragrance, you may not yet be sure which one you like. After using it in everyday life, however, you may discover which you want to choose. A study of fabric softener fragrances investigated whether initial brain activity contains information associated with that later choice.
The 2025 study identified brain regions whose activity patterns distinguished fragrances later selected from those not selected, even when initial ratings of pleasantness and willingness to use showed no clear group-level differences. It also examined whether those classification cues remained after about four months of use.
- From an unfamiliar fragrance to an everyday choice
- Trying three products, then using one choice for about four months
- Initial self-reports did not clearly distinguish later choices
- Looking at combinations of responses rather than the strongest activation
- Could the model simply be identifying fragrance type or intensity?
- The left superior frontal region retained information about four months later
- Why this is not proof that the brain knows your true preferences
- Summary
From an unfamiliar fragrance to an everyday choice
The researchers were from the National Institute of Information and Communications Technology (NICT) and LION. They had two objectives: to investigate whether brain activity in response to an unfamiliar fragrance reflects subsequent preference-based choices, and whether similar information remains after the fragrance becomes familiar.
The study recruited 30 women who used fabric softener at least three times a week and chose and purchased it themselves, mainly on the basis of fragrance. Following withdrawals and exclusions for measurement-data problems and other reasons, the final analysis included 25 participants with a mean age of 42.2 years.
Three commercially available fabric softeners that the participants had never used were selected. Their fragrances were described as sweet floral, fresh floral, and floral. To minimize associations with product packaging, container labels were concealed except during the preliminary selection stage.
Trying three products, then using one choice for about four months
During the first fMRI session, brain activity was recorded while participants smelled the three fragrances. A device delivered each fragrance through a nasal mask for six seconds at a time over repeated measurements. After each measurement run, participants rated odor intensity, pleasantness, and willingness to use on five-point scales.
Within the following two weeks, participants used all three products at home twice each. They then chose a favorite on the basis of that experience. For about four months, they continued using only their chosen softener for everyday laundry.
Finally, they completed the same fMRI measurements and fragrance ratings again. Long-term use involved one self-selected product, not all three. This procedure connected initial responses with subsequent choices and responses after familiarization.
Initial self-reports did not clearly distinguish later choices
The researchers retrospectively compared initial ratings by labeling the fragrance subsequently chosen as selected and the other two as non-selected.
At the first measurement, selected and non-selected fragrances did not differ significantly in pleasantness or willingness to use. At the measurement about four months later, however, the fragrances participants had chosen and used received significantly higher ratings.
This does not mean that nobody initially had a preference. It means that the comparison across the 25 participants found no clear difference corresponding to their later choices on these five-point scales.
Looking at combinations of responses rather than the strongest activation
What about brain activity? An initial analysis comparing how strongly each region responded to the fragrances found no significant difference between selected and non-selected odors.
The team therefore classified responses using the spatial pattern of activity across multiple small regions, rather than simply reducing activity to a single average. Even when overall averages are similar, the combination of which areas respond and how strongly might provide clues.
The classification used a leave-one-participant-out procedure: a model trained on the other participants was tested on the participant left out. It distinguished two categories, later selected versus non-selected. Its accuracy was not a direct three-way prediction of which of the three products someone would buy.
This analysis identified regions where initial brain activity supported classification above chance. Prominent examples included the right nucleus accumbens, the frontal part of the left piriform cortex, and the left superior frontal region. Information associated with later choices was present in olfactory regions as well as reward-related regions.
Could the model simply be identifying fragrance type or intensity?
The three softeners were not chosen equally often: 4, 9, and 12 participants selected them, respectively. Perceived odor intensity could also differ. The researchers therefore needed to consider whether apparent classification of preference merely reflected distinctions between product fragrances or their intensity.
They additionally tested whether activity in the same regions could classify odor identity, intensity, pleasantness, and willingness to use. The right medial orbitofrontal cortex also distinguished odor identity, leaving open the possibility that fragrance-type differences contributed to its results.
In contrast, the same analyses did not significantly classify odor identity or intensity in regions including the right nucleus accumbens, left piriform cortex, and left superior frontal region. Initial pleasantness and willingness to use were not significantly classified either. These findings support the possibility that immediate ratings and choices based on subsequent product experience do not contain exactly the same information.
The left superior frontal region retained information about four months later
At the second measurement, the researchers reexamined the regions that had classified choices initially. Among them, significant classification was found in the left superior frontal region.
Classification also exceeded chance when a model trained on the first session was applied to the second, or vice versa. This suggests that choice-related patterns shared a component across the two time points.
Comparable significant classification was no longer found in the other regions. That does not mean that participants had become bored with the products or that the fragrances had lost their value. Self-reported pleasantness and willingness to use had not declined, and the authors discuss the need to investigate the relationship between subjective ratings and brain patterns further.
Why this is not proof that the brain knows your true preferences
The study showed that, within a particular set of products and participants, initial brain activity contained statistical information associated with later choices. It did not establish a technology that can reliably read preferences unknown even to the individual in every person.
The sample consisted of 25 women who frequently used fabric softener and placed importance on fragrance, and the study examined three products. Because only the self-selected fragrance was used over the longer period, the effects of making a choice, repeated use, and the passage of time cannot be separated independently.
Nevertheless, the study is valuable for combining real-life use with brain-activity patterns to investigate aspects of preference that an immediate question about liking may not fully capture.
The research was funded by LION, and coauthors employed by the company disclosed that they received salaries. No other financial conflicts of interest were declared.
Summary
Even when initial ratings of fabric softener pleasantness and willingness to use showed no clear differences, brain-activity patterns contained information distinguishing fragrances later selected from those not selected. Comparing only the strength of activation had not revealed those differences.
After about four months of using the chosen fragrance, the left superior frontal region still suggested a shared choice-related pattern. By following initial impressions, everyday choices, and preferences after familiarization, the study examined more specifically how fragrance preferences appear in the brain.
References
Okumura T, et al. Latent preference representation in the human brain for scented products: Effects of novelty and familiarity. NeuroImage. 2025;310:121131. doi:10.1016/j.neuroimage.2025.121131
