Tag Archives: Institute of Science Tokyo (Science Tokyo)

Help maintain cognitive and memory functions with virtual reality (VR) game which integrates smell

I always enjoy a research story involving the sense of smell (olfaction); the most recent ones here being an April 22, 2025 posting (also from the Institute of Science Tokyo) “Fragrance design using deep neural networks (DNNs)” and an April 17, 2025 posting “Olfactory ethics.” There’s also this ‘golden oldie’ from May 22, 2017 “Preserving heritage smells (scents).” Now, the sense of smell enters virtual reality.

Caption: Olfactory VR offers the potential for cognitive rehabilitation and dementia preventation [sic] in aging populations Credit: Institute of Science Tokyo

An April 30, 2025 Institute of Science Tokyo press release (also on EurekAlerrt) describes some of the latest investigation into the sense of smell, Note: A link has been removed,

As the global population ages, supporting older adults in maintaining their cognitive and memory functions has become a pressing concern. The United Nations estimates that by the 2070s, there will be over 2.2 billion people aged 65 or older, surpassing the global number of children under 18. This demographic shift is especially pronounced in Japan, the fastest-aging country, where 28.7% of the population is 65 or older.

One promising strategy to counter cognitive decline is through olfactory stimulation—engaging the sense of smell. Smell signals travel directly to brain regions involved in memory and emotion. Building on this knowledge, a joint research team from Institute of Science Tokyo (Science Tokyo), University of the Arts London, Bunkyo Gakuin University, and Hosei University, Japan, has developed the world’s first cognitive training method for older adults by combining olfactory stimulation with virtual reality (VR). The study was published in Volume 15 of the journal Scientific Reports on March 28, 2025.

“VR provides a promising platform to simulate sensory conditions in a controlled yet engaging manner. By combining goal-oriented tasks with real-time feedback, our VR-based olfactory training approach can increase cognitive engagement and maximize its therapeutic impact,” says Professor Takamichi Nakamoto from Science Tokyo.

The method involves an olfactory display that emits specific scents during immersive VR gameplay, activating memory- and emotion-related brain regions. In the activity, participants are asked to memorize and later match scents within a virtual environment. The experience begins in a virtual landscape. Using a VR controller, participants interact with a scent source represented by a stone statue. When touched, the statue releases a specific scent, accompanied by a white vapor cloud as a visual cue to reinforce memory.

Participants then explore the virtual landscape to locate a scent source. As they move through the landscape, the olfactory display emits subtle traces of the scent to guide them to the location. Upon reaching the odor source, shown as a stone lantern, they encounter three colored vapor clouds, each emitting a different scent. Their task is to compare the smells and identify the one that matches the original scent they memorized.

“The smell memory phase strengthens odor recognition and memory encoding by linking the olfactory stimulus with a visual cue. The navigation phase challenges players to integrate spatial navigation with odor recognition while retaining memory of the initial scent. The final odor comparison phase engages olfactory discrimination and working memory retrieval, reinforcing cognitive function,” explains Nakamoto.

The activity led to noticeable cognitive improvements in 30 older adults aged 63 to 90. After just 20 minutes of playing the VR game, participants showed improvements in visuospatial rotation and memory. Visuospatial processing and cognitive function were assessed through different tasks. In the Hiragana Rotation Task, where they had to decide if rotated Japanese characters matched the original, scores improved from 19–82 to 29–85. In a word-based spatial memory recall task, where participants memorized word positions in a grid, scores rose from 0­–15 to 3–15. These improvements were validated through statistical analysis.

With continued research and development toward more affordable olfactory displays or alternate scent delivery methods, olfactory-based VR activities could become an accessible and engaging tool for supporting mental health in older adults.

About Institute of Science Tokyo (Science Tokyo)

Institute of Science Tokyo (Science Tokyo) was established on October 1, 2024, following the merger between Tokyo Medical and Dental University (TMDU) and Tokyo Institute of Technology (Tokyo Tech), with the mission of “Advancing science and human wellbeing to create value for and with society.”

Here’s a link to and a citation for the paper,

Exploring the effects of olfactory VR on visuospatial memory and cognitive processing in older adults by Ryota Sunami, Takamichi Nakamoto, Nathan Cohen, Takefumi Kobayashi & Kohsuke Yamamoto . Scientific Reports volume 15, Article number: 10805 (2025) DOI: https://doi.org/10.1038/s41598-025-94693-9 Published: 28 March 2025

This paper is open access.

Fragrance design using deep neural networks (DNNs)

A December 31, 2024 news item on ScienceDaily announces work from Japanese researchers on fragrance design,

Scientific research explores the potential of DNNs [deep neural networks] in transforming fragrance design. By analyzing the sensing data of 180 essential oils, the DNN was trained using the odor descriptor data from 94 essential oils to generate fragrance profiles, validated through sensory evaluations to align with human olfactory perceptions. The study underscores the technological ability to streamline fragrance creation, reduce costs, and foster innovation, opening up exciting possibilities for personalized and scalable scent development.

Caption: DNNs can transform fragrance design by predicting and creating novel scents from chemical data, ushering in a new era of digital scent creation. Credit: Institute of Science Tokyo

A January 9, 2025 Institute of Science Tokyo press release (also on EurekAlert but published December 31, 2024), which originated the news item, delves further into the topic of digitizing the sense of smell, Note: A link has been removed,

Deep Neural Networks (DNNs) have become an essential driver of innovation across various industries, from healthcare to manufacturing. By analyzing large datasets, identifying patterns, and making precise predictions, DNNs are transforming the way we approach complex tasks. One such area where DNNs are making a remarkable impact is in the digitalization of smell, a field traditionally dominated by human expertise and sensory evaluations. However, a recent study aims to revolutionize this practice by exploring how DNNs can assist in fragrance design.

Moreover, an odor reproduction technique has been developed, enabling a wide variety of scents to be generated by varying the mixing ratio of a small set of odor components. These odor components are prepared by blending essential oils used in the analysis.

A research team led by Professor Takamichi Nakamoto from the Laboratory for Future Interdisciplinary Research of Science and Technology (FIRST), Institute of Integrated Research (IIR), Institute of Science Tokyo, Japan, published their research in Scientific Reports on December 28. This study was driven by the growing need for more efficient and innovative methods of fragrance creation. The study aimed to quickly make the intended scent without trial and error, leveraging DNNs to predict odor profiles based on multidimensional sensing data.

Nakamoto explains “We hypothesized that the DNNs when integrated with chemistry and sensory science could offer new insights into fragrance development. We conducted the study by analyzing mass spectrometry data from 180 essential oils, providing a comprehensive understanding of their odor components. These data were then used to train a DNN designed to predict odor descriptors from the odor-component composition. The DNN employed multiple layers optimized to capture the intricate relationships between its compositions and the resulting scents.” To improve the model’s accuracy and generalization, the team augmented the data with random mixtures of essential oil spectra and introduced noise, ensuring the model could adapt to real-world complexities. Once the DNN model generated the odor-component compositions, human evaluators assessed the DNN- generated scents alongside reference oils.

The DNN achieved the highest accuracy in predicting the odor descriptor “floral” and lower accuracy for the descriptor “woody”. Sensory testing further confirmed the effectiveness of the model, as human panelists found that the DNN-generated oils using odor components were more similar to the reference oils than those with added odor descriptors. These findings highlight the system capability to accurately replicate existing fragrance profiles and, in some cases, generate entirely new combinations.

The study demonstrates numerous benefits, like DNN can significantly reduce the time and costs involved in fragrance development by streamlining both chemical analysis and sensory evaluations. Additionally, DNN makes fragrance creation scalable, allowing it to adapt to diverse market preferences and consumer demands. Most notably, the use of DNN opens up innovative possibilities by enabling the generation of new and unique scent profiles that might not have been discovered through traditional methods.

Looking to the future, the implications of this study are profound. “As DNN models continue to evolve, they could enable the creation of personalized fragrances tailored to individual preferences. Additionally, this approach could be extended to other sensory domains, such as taste, where similar methods could be used to craft personalized flavor profiles,” shares Nakamoto.

By combining DNNs, chemical analysis, and sensory testing, the study emphasizes the potential to replicate and innovate within the fragrance industry. With its ability to enhance efficiency and creativity, a revolution in fragrance design is expected, ushering in a new era of innovation.


About Institute of Science Tokyo (Science Tokyo)

Institute of Science Tokyo (Science Tokyo) was established on October 1, 2024, following the merger between Tokyo Medical and Dental University (TMDU) and Tokyo Institute of Technology (Tokyo Tech), with the mission of “Advancing science and human wellbeing to create value for and with society.”

Here’s a link to and a citation for the paper,

Automatic scent creation by cheminformatics method by Manuel Aleixandre, Dani Prasetyawan & Takamichi Nakamoto. Scientific Reports volume 14, Article number: 31284 (2024) DOI: https://doi.org/10.1038/s41598-024-82654-7 Published: 28 December 2024

This paper is open access.