Trial Analysis of the Relationship between Taste and Biological Information Obtained While Eating Strawberries for Sensory Evaluation

Keisuke Maeda, Ren Togo, Takahiro Ogawa, Shin-ichi Adachi, Fumiaki Yoshizawa, Miki Haseyama

TL;DR

This study explores the complex relationship between strawberry taste perception and biological information, specifically brain activity measured via fNIRS. By employing canonical correlation analysis, the researchers uncover previously hidden links between cultural factors, individual strawberry characteristics (like sugar-to-acid ratio), and taste, suggesting a path towards more objective sensory evaluation.

What Problem Does This Paper Address?

Sensory evaluation of food, like the taste of strawberries, is inherently subjective. While factors like sugar content and acidity are known to influence taste, capturing the full picture requires understanding how these objective measures, subjective preferences, and even cultural backgrounds interact with our brain's perception. Previous research often focused on simple correlations between individual data points and taste. However, this approach can miss more nuanced, indirect relationships. This paper aims to bridge this gap by investigating how brain activity, measured during strawberry consumption, can serve as a mediator to reveal deeper connections between taste and various other data types, including subjective questionnaires and objective strawberry characteristics.

Methods

The core of this research lies in its multivariate analytical approach, specifically Canonical Correlation Analysis (CCA). Unlike simple correlation, CCA identifies latent variables that maximize the correlation between two sets of variables. Here, the researchers used CCA to explore relationships between:

  • Brain activity (fNIRS data) and taste (VAS scores).
  • Brain activity (fNIRS data) and questionnaire responses (covering reward, cultural, and information factors).
  • Brain activity (fNIRS data) and individual strawberry characteristics (sugar content, acidity, sugar-to-acid ratio, and size).

Before CCA, fNIRS data underwent Principal Component Analysis (PCA) for dimensionality reduction, aiming to retain over 90% of the variance. This pre-processing step is standard for handling the high dimensionality of fNIRS signals. The study also performed a preliminary simple correlation analysis to establish baseline relationships before applying the more sophisticated CCA.

Datasets

The study utilized a custom dataset collected from 19 participants (14 men, 5 women) who consumed four types of strawberries: Skyberry (large/small) and Tochiotome (large/small). The collected data included:

  • Visual Analog Scale (VAS) scores: For subjective taste evaluation.
  • Questionnaire responses: Covering reward, cultural, information, and appearance factors, rated on a 5-point Likert scale.
  • fNIRS data: Measuring brain activity during strawberry consumption.
  • Individual strawberry information: Including sugar content, acidity, sugar-to-acid ratio, and size.

This dataset appears to be newly collected for this specific study, combining subjective, objective, and physiological measures.

Key Insights and Ideas

  • Brain activity as a mediator: The study's central hypothesis is that brain activity acts as a crucial link, revealing relationships between taste and other data that simple correlations might miss. This is the paper's main contribution.
  • Cultural influence on taste: CCA revealed a significant correlation between cultural factors (from questionnaires) and brain activity, suggesting that our background and familiarity with certain foods can influence how we perceive their taste, mediated by neural responses.
  • Sugar-to-acid ratio's neural link to taste: The analysis of cross-loading between brain activity and individual strawberry information highlighted the sugar-to-acid ratio as a key factor correlated with brain activity, implying its direct neural impact on taste perception.
  • Limitations of simple correlation: The study demonstrates that while simple correlations might not show a strong link between certain objective data (like sugar content) and taste, CCA can uncover these relationships when brain activity is considered.

Key Results

  • Brain activity and taste correlation: CCA showed a canonical correlation of 0.4 between brain activity and taste, with a p-value of 0.000, confirming a significant potential correlation.
  • Brain activity and questionnaire correlation: High canonical correlations were observed between brain activity and questionnaire responses (up to 0.81), particularly highlighting the cultural factor (Q2) as strongly correlated with brain activity.
  • Brain activity and individual information correlation: Significant canonical correlations were found between brain activity and individual strawberry characteristics (up to 0.53). Specifically, the sugar-to-acid ratio showed a notable correlation with brain activity.
  • Reward factor's direct taste correlation: Simple correlation analysis indicated that reward factors (Q1) from questionnaires had a strong direct correlation with taste.

Limitations and Future Work

The authors acknowledge several limitations:

  • Noise in fNIRS data: While PCA was used for dimensionality reduction, the extraction of principal components might not be entirely sufficient to remove all noise.
  • Limited expressive power of one-dimensional features: For taste (VAS), only one dimension of canonical features was used, limiting its expressive power.
  • Complexity of multivariate data: The current CCA approach is limited to two sets of data.

Future work suggested includes:

  • Applying denoising methods for fNIRS data: To improve the reliability of brain activity measurements.
  • Increasing the number of dimensions for taste features: By incorporating additional evaluations beyond VAS.
  • Exploring multi-view CCA: To handle the multiple types of data (taste, questionnaires, brain activity, individual information) more effectively.
  • Developing machine learning models: To estimate taste using heterogeneous data, leveraging the discovered relationships.

Summary updated

Abstract

This paper presents a trial analysis of the relationship between taste and biological information obtained while eating strawberries (for a sensory evaluation). This study used the visual analog scale (VAS); we collected questionnaires used in previous studies and human brain activity obtained while eating strawberries. In our analysis, we assumed that brain activity is highly correlated with taste. Then, the relationships between brain activity and other data, such as VAS and questionnaires, could be analyzed through a canonical correlation analysis, which is a multivariate analysis. Through an analysis of brain activity, the potential relationship with "taste" (that is not revealed by the initial simple correlation analysis) can be discovered. This is the main contribution of this study. In the experiments, we discovered the potential relationship between cultural factors (in the questionnaires) and taste. We also found a strong relationship between taste and individual information. In particular, the analysis of cross-loading between brain activity and individual information suggests that acidity and the sugar-to-acid ratio are related to taste.
BibTeX
@inproceedings{maeda2022trial,
  author = {Keisuke Maeda and Ren Togo and Takahiro Ogawa and Shin-ichi Adachi and Fumiaki Yoshizawa and Miki Haseyama},
  title = {{Trial Analysis of the Relationship between Taste and Biological Information Obtained While Eating Strawberries for Sensory Evaluation}},
  booktitle = {Sensors},
  year = {2022},
  url = {https://openreview.net/forum?id=zmuuk3uKxA},
}
Figure