Banana split - Molecular Gastronomy

So many sensory experiences have time as a key element. Delight is a combination of Liking (a pleasant hedonic experience) and Surprise (the unexpected). What better way to create surprise in a product than to have an unexpected change in perception taking place? When the molecular gastronomer, Heston Blumenthal makes bacon and egg ice cream for you, it is a series of surprises. When he drops a green tea and vodka ice cream on your tongue after forming it in liquid nitrogen at -80 degrees, it explodes in your mouth and releases the vodka followed by the cooling sweetness of the ice cream and the flavor of green tea. When you consume a refreshing beverage that is initially sweet, but sweetness continues to linger when all other sensation is gone, your liking for the product starts to slide. The first taste of a wine that is astringent may be pleasant, but continued sips intensify the astringency to the point of revulsion. When your chewing gum loses all flavor and becomes a piece of rubber in your mouth, it’s not pleasant.

These are all questions that time-intensity (TI) can help to answer. After all, we measure the sensory properties of products to understand what consumers like and to provide product developers with reliable information to help them create better products. If we don’t have reliable data about the impact of product reformulation, we are just guessing. The role of Sensory Science is to provide valid and robust information to help marketing and product development make well informed decisions.

The details of Time-Intensity procedures can be found in texts like Lawless and Heymann’s Sensory Evaluation of Food, or in the ASTM Standard Guide E1909 on Time Intensity.

I have been interested in following more than just one or two attributes in a dynamic sensory test for quite a long time. Although the Dual Attribute Time-Intensity (DATI) method was published almost 15 years ago, it has proved to be a challenging method to train panelists to use. Very few publications have been written using DATI. Ever since the submission of the first DATI manuscript, reviewers have expressed their concern with the number of tasks a panelist is asked to attend to, almost simultaneously.

In fact, we are expecting panelists to multi-task. A survey of the current literature on human multi-tasking concludes that “multi-tasking is a myth”. Humans cannot focus on two or more tasks simultaneously; we tend to toggle between tasks in a process called “Continuous Partial Attention”. The result is that performance actually gets poorer as the number of tasks increase. There are more errors and a longer time is required than if each task was conducted sequentially. The cognitive load of following sensory attributes and measuring intensity exceeds our ability to process.

In effect, if the attributes that interest us are changing rapidly (measured over a few seconds), for example the bouquet of a wine as it responds to air, we are hard-pressed to register their sequence and to have a computer time-stamp the response. This is the domain of Temporal Dominance of Sensations (TDS) or Temporal Order of Sensation (TOS). However, if the changes are a bit slower, there are several strategies that we can apply to measure their intensity.

Progressive Profiling uses a series of line scales that are on the same screen (practically a maximum of 5). The screen appears and the panelist scores the line-scale attributes in exactly the same way as standard descriptive profiling. Then the screen changes to a neutral background for a fixed time-delay (anywhere from 5 seconds to several minutes, depending upon the product) before the same screen of attributes appears for the next time point. This technique has often been used in conventional profiling to measure aftertaste a fixed time after swallowing. This puts all panelists into the same time line.

Progressive Profiling has the benefit of being a simple descriptive analysis task that our panelists are well trained to accomplish. The attributes are fixed, making their measurement sequential and repetitive, reducing the cognitive load. Data may be analyzed by ANOVA at each time point or it may be presented as actual time-intensity curves for the duration of the test.

A variation on this method that has been conducted by Lisa Methven at the University of Reading is called Sequential Profiling, “The effect of consumption volume on profile and liking of oral nutritional supplements of varied sweetness: Sequential profiling and boredom tests”. 2010. L. Methven, et al. Food Quality and Preference (21:8) 948-955.

We will continue our development of temporal methods in search of reliable and useful tools that answer research questions. Please contact me if you have any ideas or thoughts to share.